Knowledgebase

Help and knowledgebase

Online manual for the Trend portal, covering daily use, equipment investigation, predictive analytics, hierarchy governance, data import, administration and the glossary used across the tenant.

29 Portal views

8 Workflows

37 Glossary terms

Start here

The portal is organised around the asset hierarchy. Start from the overview when checking plant condition, move into an asset or equipment view when investigating a problem and use System config only when changing the governed structure, thresholds, sensors or access model. Pages remember the last selected hierarchy node, so once you pick the equipment you are working on, the same node carries across Asset Diagnostics, Equipment Health Index, Trend Analysis and Predictive Analytics. The hierarchy itself is the spine: organisation nodes, equipment nodes and measurement-point nodes are the three node types every page consumes.

For operators

Begin every shift on Fleet Overview to see which area trends amber or red, then drill into Plant Dashboard for the area, Asset Dashboard for the asset and Asset alerts for any open notifications. Maintenance History tells you what was done recently and what is open as a work order. When something needs eyes on the signal, open Asset Diagnostics. Listen to the audio channel where one is configured. Typical first session of the day: Fleet Overview, then any red asset, then Asset Diagnostics, then Maintenance History. If you do not see expected data, check the hierarchy selection first, the time range second and the gateway connection in Online systems third. A sensor that has not uploaded for several hours is usually a wireless mesh or battery issue, not a portal issue.

For reliability engineers

Use Equipment Health Index to spot equipment with a falling score before alarms fire. Trend Analysis lets you overlay scalar values across periods and across assets so a slow drift is visible. Asset Diagnostics carries the full analytical bench: time-domain waveform, spectrum, long waveform, envelope demodulation, cepstrum, waterfall plot, multi-time waveform, multi-spectrum, cross phase, order analysis, order envelope, envelope trend, time-domain index multi-trend and multi-indicator trend. Bearing and gearbox holds component-specific evidence and calculated fault frequencies. Fault cases is the structured record of confirmed defects. Predictive Analytics consolidates the above into a model verdict at equipment level. A productive session moves from a falling EHI, to a Trend Analysis overlay that proves the drift, to an envelope spectrum that names the failure mode, to a Fault case that records the diagnosis and intervention.

For administrators

System config is the governed source for hierarchy, users, departments, online systems, thresholds, data collection trigger, scalar alerts, planned exception, SLA targets, bearing fault frequencies, audit log and site settings. Data integration handles CSV and Excel imports with a mapping review and commit step. Sensor import is the master-data CSV pathway for bulk sensor changes. Hardware admin is the self-diagnosis surface for gateways and sensors and covers collection, storage, network status, transmission, system, rotational speed and module aspects. User regions limits user visibility by geographic or tenant scope. Server config records the runtime ports and connections. License management controls which modules the tenant has access to. Treat every change as a release: backup, change, verify and record what changed and why.

Common workflows

These workflows describe the usual operating rhythm from plant overview to evidence, action and governance. Each follows the same structure: where to start, what to confirm, what good evidence looks like and what to capture before moving on. Where the manual references a specific analysis, the same view is reachable inside Asset Diagnostics.

Check plant condition

Open Fleet Overview and scan the area health bars. Open the area that flags amber or red on Plant Dashboard and check the asset list, KPI cards and notification queue. Drill into Asset Dashboard for the asset with the worst score, confirm scalar values are recent and within configured thresholds and check Asset alerts for active notifications. If nothing is open and the score is still poor, the issue may be a sensor or threshold problem rather than an asset problem. Use this loop at the start of every shift, after any planned outage and any time an operator escalates a concern. If a production line is the natural grouping, the production-line view rotates through every line so a single screen can serve a control room.

Open Fleet Overview

Investigate diagnostic evidence

Open Asset Diagnostics and select the asset or equipment in the left hierarchy panel. Use the chart strip to compare scalar trends across low-frequency acceleration RMS, velocity RMS, high-frequency acceleration RMS and temperature. Open a waveform for the most recent capture and listen to the matching audio channel where one exists. Use spectrum and envelope views to look for repeating frequencies linked to bearing, gear or unbalance signatures. Apply low-pass, high-pass, band-pass or band-stop filtering when the carrier signal hides the modulation you are after. Use the auxiliary cursor to measure a sideband interval or a harmonic spacing. Compare the current capture against a known-good capture from the same equipment using multi-time waveform or multi-spectrum. If you confirm a defect, record it in Fault cases with the supporting captures, the frequency match and any photo evidence. The diagnostic loop ends with a Fault case and a recommended action, not just an observation.

Open Asset Diagnostics

Review predictive analytics

Open Predictive Analytics, pick equipment in the hierarchy (not a single sensor) and check the model readiness panel. Confirm mounted sensors are present, measurement points are configured, recent measurements exist and any retraining feedback has been processed. Review the anomaly check and fault prediction in light of the EHI trend, any open Fault case and the latest maintenance history entries. Add a human label only when you have confirmed evidence (a captured waveform with a clear bearing signature, a confirmed downtime root cause, a thermography photograph). A model prediction without supporting evidence is a hypothesis, not a decision. Use the equipment-led view rather than a single channel: the predictive layer fuses every mounted sensor on the same equipment into one verdict, and a single-channel pick loses that fusion.

Open Predictive Analytics

Assess equipment health

Open Equipment Health Index, select the equipment and read the consolidated score. The score is a governed 0 to 100 health record from mounted sensor evidence, threshold state, recent captures, fault history and reliability rules. Good is 80 to 100, Moderate is 60 to 79, Poor is 40 to 59 and Critical is 1 to 39. Pending means no valid timestamped health record exists. Use the contributing factors panel to see which mounted sensors are pulling the score down. Open the linked Asset Diagnostics view to confirm the signal that the score is reacting to. Cross-check Maintenance History for any recent intervention that should have lifted the score and has not.

Open Equipment Health Index

Close the maintenance loop

Maintenance History is the record of work, not the source of decisions. Use it to confirm what has been done, what is still open, what failed, how long the asset was down and which interventions actually moved the EHI. Before recording a new action, check Fault cases for an open case against the same equipment. Before escalating, check the work order audit trail for a similar fix that resolved the same signature. A closed loop is: detect (Asset Diagnostics or Predictive Analytics), record (Fault case), schedule (Maintenance History work order), execute (field), confirm (re-check EHI, scalar trends and a fresh spectrum or envelope after the intervention), close (Maintenance History action note). For wear-driven failure modes, the post-intervention envelope spectrum is the most convincing evidence the defect has been resolved.

Open Maintenance History

Govern hierarchy and imports

System config holds the canonical hierarchy. When a new line, asset or equipment lands, choose the lowest-risk import path: inline edit for one or two nodes, CSV import for bulk additions, Excel import with mapping review for unstructured spreadsheets, P and ID drawing import only when an OCR-and-review pipeline is in place. The hierarchy schema recognises three node types (organisation, equipment, measurement point); only organisations and equipment can hold children. Every import is a mapping review then a commit. Validate first, look at the inconsistency report, fix at source if you can, then commit. Never edit the node id or node path column when round-tripping through Excel; those fields anchor the validation step. After import, walk the hierarchy on the Asset Diagnostics page to confirm the new nodes appear in the expected position and that mounted sensors carry across.

Open System config

Detect a bearing defect early

Bearing defects show in envelope spectra well before they show in a velocity RMS scalar. Open Asset Diagnostics, select the suspect equipment and open the envelope view. Set the bearing band to 250 to 10000 Hz; for gears, 500 to 20000 Hz is the working range. Look for the bearing inner-race, outer-race, rolling-element or cage frequencies as calculated from the bearing dimensions in the component model. If the envelope shows a peak at one of these frequencies and a harmonic series above it, the early-stage defect is present even if no scalar threshold has fired. Cross-check the audio channel for impact-like character; an impulsive audio character paired with an envelope match is a high-confidence early bearing call.

Open Asset Diagnostics

Separate periodic components

When a spectrum looks busy with sidebands and harmonics, the cepstrum (second-order spectrum) cleanly separates periodic components and surfaces the modulating signal. Use cepstrum when a gear mesh frequency is flanked by sidebands at the shaft rotational frequency, when bearing harmonics overlap a structural resonance, or when sidebands obscure a fundamental. The cepstrum quefrency peak you see corresponds to a modulating period; convert back to frequency to confirm the modulator. Pair cepstrum with the waterfall plot to see whether the periodic component is steady, growing or transient over the sampled period.

Open Asset Diagnostics

Module guide

Every portal page is grouped by purpose. The when-to-use column tells you in one phrase what triggers a visit; the contributing data tells you what feeds the page. Pages marked with a role are restricted; pages without a role badge are visible to client, manager and admin tiers.

Asset Diagnostics

Evidence workbench for waveforms, audio, spectra, envelopes, trend waves and diagnostic traces. Carries the full analytical bench: time-domain waveform, spectrum, long waveform (high-frequency acceleration time series similar to a digital tape recorder), envelope demodulation with adjustable centre frequency and bandwidth, cepstrum, waterfall plot, multi-time waveform, multi-spectrum, cross phase, order analysis, order envelope, envelope trend (bearing 250 to 10000 Hz, gear 500 to 20000 Hz), time-domain index multi-trend covering total, peak, peak-to-peak, kurtosis, pulse, waveform, margin, skewness, root, mean, maximum and minimum, plus multi-indicator trend for temperature, speed, LF total, velocity total, HF total and audio. Use this when you suspect a defect and need to see the raw signal. Drives off mounted sensors, recent captures and the threshold profile. Common pitfall: opening the page without first selecting equipment in the hierarchy.

Open page

Fleet Overview

Fleet-level health and alarm summary across every area. Use this first thing in the morning, after any plant trip and when a manager wants the executive read. Drives off area-level EHI rollups and active alarm counts. Areas appear in the order configured in System config. A red area is not always an emergency: hover the area tile to see whether the score is driven by a single asset or a broad trend.

Open page

Plant Dashboard

Plant-level operating context: KPI cards, area trends, asset rollups. Use this when Fleet Overview points at an area and you need plant context (running shift, energy state, planned vs actual production) before drilling into one asset. Drives off the area scalar feed plus SCADA tags when SCADA integration is configured. The production-line carousel view is available when the plant is laid out as discrete production lines; it rotates through each line so a single control-room screen can serve every line. Useful as the meeting view for plant-side reviews.

Open page

Asset Dashboard

Asset-level snapshot: health score, KPI cards, current scalar values and the health trend over the configured window. Use this as the routine asset check before deciding to open Asset Diagnostics for raw signals. Drives off the asset rollup of EHI, mounted sensors and any open Fault cases. The score is what you defend; the scalars are what you check; the trend is what tells you whether the situation is improving. Clicking a measurement point on the overview image opens the trend chart of the waveform data for that point.

Open page

Equipment gallery

Visual register of mounted equipment in a heat-map style layout, designed for comparing similar machines side by side. Use this when you want to spot the outlier in a row of pumps, drives, mills or motors. Drives off mounted-equipment metadata, the equipment-model image library and the most recent rolled-up health value. Best read at one glance: a row of green tiles with one amber tile is more informative than a long list. Equipment-model images are configured in System config and shared across the gallery, asset overview and reports.

Open page

Maintenance History

Work orders, interventions, downtime, recurring problems and the action audit trail. Use this when an asset has been touched recently and you need to know what was done. Drives off the maintenance history records imported or recorded against each asset. Filters by asset, period and intervention type. The Downtime split panel makes recurring failure modes visible; if the same failure pattern keeps showing, it is a candidate for a Fault case and a thresholds review. The Start/Stop setting in System config feeds the operating-versus-downtime split: shutdown calculation runs against a configured threshold and the shaded shutdown period appears on the trend.

Open page

Trend Analysis

Cross-asset and cross-period overlays for scalar values and waveforms. Use this when Asset Dashboard suggests a drift and you want to see it formally over weeks or months, or when you need to compare two similar assets to decide if a value is normal or elevated. Drives off scalar history (LF acceleration RMS, velocity RMS, HF acceleration RMS, temperature, rotational speed, bearing envelope, gear envelope), configured thresholds and the captured waveform set. Frequency-band energy splits (low, mid, high, ultra-high) can be overlaid as separate traces when a defect signature lives in one band but not the overall scalar. Best practice: always include a known-good baseline asset in your overlay so the comparison is informative.

Open page

Asset alerts

Active alarm and notification queue including severity, status, asset and alarm context. Alarm states are unresolved, in progress and closed; once an alarm is unresolved against a piece of equipment, all subsequent alarms on the same equipment are grouped under the same alarm event until the event is closed. Use this on shift handover and after any threshold change. Drives off configured threshold rules (absolute, relative-value and multi-band) and live scalar values. SMS and email notifications fire when an alarm reaches or exceeds the level configured against the user. An alert without a recent capture is a stale alert; check Asset Diagnostics to confirm there is real signal behaviour behind the alarm rather than a sensor or threshold problem.

Open page

Benchmark metrics

Cross-period and cross-asset benchmarking for identifying persistent performance gaps and measured improvement. Use this for monthly reviews and for proving that an intervention worked. Drives off the rolled-up KPI set per asset, by configured period buckets. Looks for the slope between this month and the comparison period, not just the absolute value. The operating-day and downtime-day split sourced from the Start/Stop setting is the most reliable normalisation when comparing assets with different duty cycles.

Open page

Online systems

Gateway and instrumentation overview including connection state, recent collection activity and channel binding. Use this when an Asset Diagnostics view shows no recent waveforms, or when commissioning new hardware. Drives off the gateway-channels feed which counts bound, spare and unbound channels per gateway. A single gateway typically serves up to about 45 wireless sensors on a short-range mesh with a working line-of-sight of around 300 metres; site obstacles, antenna orientation and mounting location all change the effective range. A channel showing bound but no recent capture is the first thing to investigate; check the wireless mesh signal strength reading against the commissioning threshold (a signal strength of at least 20 is the recommended minimum for healthy live operation).

Open page

Reports

Scheduled and ad-hoc reporting for evidence packs, periodic summaries and operational review. Reports is admin-only by policy: admin reviews and validates the compiler-produced periodic PDF summary on asset performance, then issues. Drives off the periodic report compiler. Diagnostic reports can be authored manually or generated from intelligent diagnosis where the sensor evidence supports an automated reading. Useful as the artefact for monthly client reviews and for evidence of a fault and its resolution sequence.

Open page

Predictive Analytics

Equipment-level machine learning view with fault prediction, anomaly checks, model evidence and feedback labels. Use this once an asset has accumulated enough mounted-sensor history and at least one confirmed fault case for the model to have meaningful inputs. Drives off mounted sensors, measurement points, EHI evidence, diagnostic records, alarms, thresholds and training feedback. The known failure-mode catalogue covers bearing failure, imbalance, resonance, shaft bending, electrical failure, bad lubrication, misalignment, coupling failure, gear failure, mechanical looseness and shaft or bearing wear. A prediction is most reliable when the contributing data set is broad and recent; flag a thin data set rather than acting on a low-confidence prediction.

Open page

Bearing and gearbox

Component-specific diagnostic view for bearing, gear and failure mode evidence. Use this when a spectrum or envelope on Asset Diagnostics shows frequencies you suspect are component-related. Drives off the component model in System config (shafts, bearings, gearboxes) and the bearing fault-frequency library. A new bearing entry computes characteristic frequencies (inner race, outer race, rolling element, cage) from internal dimensions; a new gearbox entry uses parallel-axis tooth counts to compute the speed ratio chain. Marking a shaft as the speed shaft anchors the speed ratio at 1. The view lets you overlay calculated frequencies against a captured spectrum to confirm or refute a component diagnosis.

Open page

SCADA integration

Process-data overlay and tag mapping where SCADA context supports vibration interpretation. Use this when you need to relate a vibration change to a process state (running, idle, start-up, post-trip) or to a load condition. Drives off configured process integration channels including OPC DA, OPC UA, Modbus slave and equivalent industrial protocols. The same configuration controls both import (process tags feeding the portal) and export (portal values published to external systems). Admin-only because mapping the wrong process tag to the wrong asset misleads every downstream interpretation.

Open page

Hardware admin

Sensor, gateway and communication self-diagnosis for the installed data collection estate. The Self-Diagnosis tab covers collection, storage, network status, transmission, system, rotational speed and module aspects per acquisition unit and surfaces firmware versions, on-site photos, abnormal data entries and operation logs. Sensor-level diagnostic shows battery voltage, internal temperature, retransmission count and delay status against the gateway, plus a serial-number lookup. Use this when Online systems flags an issue, when an installation goes live, or as a periodic audit. The Self-Diagnosis tab surfaces bound, spare and offline channels and gives a per-gateway connectivity readout. A spare channel with no sensor is healthy; an offline bound channel needs follow-up.

Open page

Operation statistics

Runtime, utilisation and uptime KPIs used to connect performance to asset behaviour. Use this to interpret whether a falling EHI is a real condition issue or a duty-cycle change. Drives off the Start/Stop setting in System config, shift logs and runtime hours. The operating-day and downtime-day split is the core normaliser. An asset running at 30 per cent more hours than usual will show signal changes that do not reflect a defect; cross-check the runtime trace before raising a Fault case for an apparent drift.

Open page

Spot check

Manual inspection and handheld entry register for field checks and supplementary observations. Use this when a sensor is missing, when a route inspection raises a finding, or when a supplementary observation needs to live in the same record as the online captures. Drives off field operator submissions including photos and signatures. Where a portable handheld is in use, a temporary collection can be associated with the same equipment record so the field reading sits alongside the online history. Closed spot checks contribute to the maintenance and fault case audit trail.

Open page

Orbit plot

Shaft centreline and orbit view for paired proximity-probe evidence on plain-bearing machines. Use this on machines fitted with XY proximity probes when you need to see the shaft motion in the bearing rather than a single-axis vibration signal. Drives off paired X and Y probe captures taken at the same time. Without paired captures, the orbit cannot be rendered; check Online systems for the probe binding. The orbit shape carries diagnostic information: a circular orbit indicates balanced loading, an elliptical orbit suggests preload, a figure-of-eight points to a half-speed instability.

Open page

Equipment Health Index

Equipment-level consolidated health score that rolls up mounted sensors and applicable reliability evidence. Use this as the executive read for equipment health and as the entry point before diving into Asset Diagnostics. The score is produced from mounted-sensor scalars, threshold state, recent captures, health scoring rules, frequency-band energy splits and any open Fault cases. A score that drops without a triggered alarm means the health rules are catching early drift before the threshold gate. Pending means no valid timestamped health record exists.

Open page

Fault cases

Root-cause analysis and historical fault library for repeating defects, evidence records and lessons learned. Use this when you confirm a defect from Asset Diagnostics and need to record it formally, and when you want to check whether the current symptom resembles a previously resolved case. Drives off operator submissions, attached captures and linked maintenance actions. A good fault case has a symptom description, an attached capture (waveform, spectrum, envelope or photo), a confirmed diagnosis with a frequency or pattern match, the recommended intervention, the executed intervention with date and operator and the post-intervention verification (a fresh capture confirming the signature has cleared).

Open page

Energy monitoring

Energy KPIs and consumption trends where energy behaviour informs reliability improvement. Use this when an asset is consuming more energy than its peers or its baseline, which often points to mechanical loss (bearings, alignment, lubrication). Drives off configured energy meters and the same area roll-up used by Plant Dashboard. Cross-reference an unexplained energy rise with the velocity RMS trend and the bearing envelope; mechanical losses usually correlate with at least one vibration scalar.

Open page

Lubrication

Lubrication route, interval and lubricant tracking for reliability control. Use this when bearing fault frequencies climb and lubrication may be a contributing cause, or as part of the planned lubrication route. Drives off configured routes and lubrication history. An overdue lubrication record on the same equipment as a developing bearing signature is the simplest, lowest-risk first action. After lubrication, expect the bearing envelope and high-frequency band energy to fall within one or two collection cycles; a persistent envelope peak after lubrication points to true mechanical damage rather than dry running.

Open page

Spot violations

Compliance and inspection variance register for field observations that need action. Use this when a route inspection raises a finding that is not yet a Fault case but cannot be ignored. Drives off field submissions and configured compliance rules. Resolution rate and aging are the two metrics to watch; an aging violation is a governance risk. Where the same violation type repeats on the same equipment, escalate to Fault cases and review the threshold profile or the inspection route configuration.

Open page

System config

Central governance page for hierarchy, users, departments, online systems, thresholds, data collection trigger, scalar alerts, planned exception, SLA targets, bearing fault frequencies, audit log and site settings. Admin only. Every other portal page consumes the hierarchy from here. Three node types are recognised: organisation, equipment and measurement point; only organisations and equipment can hold children. Acquisition definitions on each measurement point set waveform interval, long waveform interval, audio duration and sampling interval. Trigger collection settings cover acceleration, speed, rotational-speed and PLC rotational-speed triggers with separate impact thresholds for acceleration and kurtosis. Treat changes as releases: backup the live state by recording the current md5, change the smallest viable scope, verify the immediate consumer pages (Asset Diagnostics, Asset Dashboard, EHI), then audit the change in the audit log.

Open page

Login audit

Security audit trail and login history for access review. Use this monthly for tenant-level security review and any time an access concern is raised. Drives off the auth event log. Look for repeated failed attempts on the same account, logins from unusual geographies and any logins outside the user's region scope. Tenant access uses an OAuth-style flow and HTTPS-only transport; sensitive fields including passwords are stored encrypted at rest.

Open page

User regions

Geographic access mapping where users are limited by operating area or tenant scope. Use this when onboarding a user who should only see one area, or when removing access for a user who has moved scope. Drives off the user-to-region join table. A user without a region assignment will see everything by default; explicit region assignment is the safer posture. The SMS and email alarm level configured against each user determines which alarm severities reach them by SMS and email; combine region scope with alarm level so users only get notified about alarms in their area.

Open page

Server config

Collection server ports and connections used by the tenant runtime. Use this when Hardware admin or Online systems flags a connectivity issue, or when adding a new gateway type. Drives off the runtime configuration store. Application server, database server and gateway-facing server roles can be co-located on a single host or split across a distributed deployment depending on tenant size. Changes here require a service restart; coordinate with operations before changing live ports.

Open page

Data integration

External data source and import surface for governed CSV, Excel and P and ID drawing workflows. Use this for any structured import: hierarchy nodes, measurement points, mounted-sensor bindings, threshold packs. Drives off the mapping wizard plus the import job store. Excel threshold imports require the node id and node path columns to be left unmodified; the validation step uses them to anchor the row against the live hierarchy. Always run the parse and the mapping review before the commit; the commit is the only step that mutates the live hierarchy. Sync to device on a measurement point applies the new threshold or acquisition definition to all similar measurement points under the same equipment in one action.

Open page

License management

Registration and module entitlement page for controlling which tenant capabilities are enabled. Use this when a tenant adds or removes a module subscription, or when troubleshooting a hidden card. Drives off the licence record stored against the tenant. A page that is enabled in System config but not licensed will not render; a licence change is the first thing to check when a card disappears unexpectedly.

Open page

Predictive analytics guide

Predictive Analytics is equipment-led, not sensor-led. Select equipment, confirm that mounted sensors and measurement points are present, then review model readiness before running checks or adding feedback labels. A useful prediction relies on a broad evidence set including waveform-derived spectra, envelopes and frequency-band energy splits; a thin set produces low-confidence outputs that should not drive action without supporting diagnostic work.

Select equipment

Use the hierarchy panel to pick equipment, not a loose sensor. The model consolidates every mounted sensor on the selected equipment into a single equipment context including triaxial vibration channels, temperature, audio (where configured) and any associated rotational speed reading. If you pick a single sensor instead, only that channel feeds the analysis and the prediction loses the cross-sensor view that makes it meaningful. Check that the equipment node has a non-empty mounted-sensor list and that at least one recent capture has landed before continuing.

Review model evidence

Open the evidence panel and confirm: mounted sensors are present, measurement points are configured for the relevant analyses (low-frequency acceleration RMS, velocity RMS, high-frequency acceleration RMS, temperature, bearing envelope 250 to 10000 Hz and gear envelope 500 to 20000 Hz where applicable), the EHI score has been computed recently, the diagnostic record has at least one capture from the relevant period, alarms have a configured threshold profile and any training feedback labels have been processed. A missing evidence card is a flag, not a fatal error: the model can still run, but its confidence will be lower and any prediction should be cross-checked against Asset Diagnostics before action.

Run prediction checks

Use the fault prediction and anomaly check to assess the current evidence set. A fault prediction returns a likelihood per fault mode from the known catalogue (bearing failure, imbalance, resonance, shaft bending, electrical failure, bad lubrication, misalignment, coupling failure, gear failure, mechanical looseness, shaft or bearing wear); treat anything above the model's calibrated threshold as a candidate for diagnostic confirmation. An anomaly check flags deviation from learned normal; use it to spot drift even before a fault mode is identifiable. After confirmed evidence is in hand, add a human label so the model learns from the case; an unlabelled confirmed fault is a wasted training opportunity.

Queue retraining only with evidence

Retraining is most useful when there is meaningful new data since the last training: confirmed labels, closed fault cases, fresh diagnostic history and recorded maintenance outcomes against the same equipment context. Without that, retraining bakes in noise. Governance posture: retraining is a release, not a side-effect. Capture which data sets and which feedback labels went in, what the training set size was and what the validation result said. If a retrain regresses against a previously-correct prediction, roll back to the prior model.

Administration and governance

Administration changes affect the reliability of every downstream page. Use staged import, mapping review and role-based access when changing tenant structure or data routes. Every admin change should be traceable: what changed, who made the change, why, when and what the rollback path is.

Hierarchy governance

System config is the single governed source for the hierarchy. Asset Diagnostics, Asset Dashboard, Equipment Health Index, Predictive Analytics and every alert page should all consume the same structure. The hierarchy uses three node types: organisation, equipment and measurement point. Only organisations and equipment may hold children; measurement points are leaf nodes that anchor sensor bindings and threshold rules. When changing nodes, prefer the smallest scope: an inline rename is safer than a CSV-wide replacement. After any change, walk the affected branch on Asset Diagnostics to confirm the new structure renders cleanly and the previously-selected nodes still resolve. The hierarchy audit log records every commit; review it weekly during the first month after any large change.

Import review

CSV and Excel imports must go through the mapping review before commit. The wizard surfaces suggested column-to-canonical-field mappings with confidence bars; manually confirm the low-confidence rows. The validation step flags missing IDs, duplicate names, parent-child cycles and unsupported node types. When round-tripping threshold settings through Excel, never modify the node id or node path columns; the validator uses both to anchor each row against the live hierarchy. Fix at source where possible; if you must override during import, record the override in the import notes so the next operator understands why the imported file differs from the live structure. Commit is the only mutating step.

Thresholds and access

Thresholds and scalar alert rules determine what appears in Asset alerts and how the EHI scores. The portal supports three rule types: normal (manual or Excel-imported absolute thresholds per level), automatic (computed from 100 to 500 historical samples over a chosen reference window) and multi-band (separate thresholds for configured frequency bands with upper and lower limits). A relative-value rule fires when the reading rises by a configured delta against the previous reading, even before the absolute threshold trips. User roles and User regions determine what each user can see. The SMS and email alarm level on the user record gates which alarm severities reach them. A new threshold profile rolls out gradually: apply to one asset, watch for one cycle, then roll across the area. Document every change in the audit log and capture screenshots of the before-and-after state for high-impact thresholds.

Data collection estate

Online systems, Hardware admin and Sensor import describe the same physical reality from three angles. They must agree. A gateway showing online in Hardware admin but no recent captures in Online systems is a binding or channel issue. A mounted sensor in Sensor import without a binding in Hardware admin is half-installed. Reconcile these three pages monthly: pick a gateway, walk its channels, confirm each bound channel has a matching mounted sensor in the hierarchy and a recent capture in Asset Diagnostics. Use the per-sensor self-diagnosis read for battery voltage, internal temperature, retransmission count and delay against the gateway; an MPU log download is available where a sensor needs deeper review. Drift in any one of the three pages breaks downstream reliability. Firmware upgrades roll through an upgrade plan that targets one collector type at a time and reports execution status per acquisition unit; never apply an upgrade file from outside the vendor distribution.

Troubleshooting

Use these checks before escalating a support issue. Each entry follows the same shape: what to look at first, what to capture and what to do if the check does not resolve the symptom.

A page says no configured data

First, confirm the selected hierarchy item. Different pages have different default selections; a freshly opened Asset Diagnostics may resolve to a parent node that has no mounted sensors. Second, check the time range; the default window may be narrower than the most recent capture. Third, confirm the equipment has mounted sensors via the System config hierarchy panel. If all three are correct, check Online systems for a gateway connectivity issue; a bound channel with no recent capture points to a wireless mesh, battery or sensor health problem. If still no data, capture the page URL, selected node id and time range, then raise with admin.

A card or page is hidden

Access is role-based: admin pages are visible only when the signed-in account has the admin role; manager pages need at least manager. Check the user chip top-right for the role currently in effect. Check User regions for any geographic constraint that hides assets outside the region. Check License management for a disabled module. If the user should have access and does not, the issue is most often a role assignment in System config Users tab; an admin can resolve in seconds.

The hierarchy looks different

System config is the canonical reference. If a page shows a different structure, the page is either caching a stale version, looking at a different tenant, or reading from a non-canonical source. Hard-refresh first (Ctrl plus Shift plus R). If the structure persists, capture the URL, the selected node and a screenshot of the visible mismatch and send to admin. Most hierarchy mismatches resolve to a recent import that did not propagate, which an admin can confirm in the audit log.

Prediction cannot run

Predictive Analytics needs: mounted sensors on the selected equipment, configured measurement points covering at least LF acceleration RMS, velocity RMS and HF acceleration RMS, recent captures, a threshold profile and at least one prior fault case or feedback label to anchor learning. Open the model evidence panel and look for missing cards. The most common cause is no recent captures, which points back to Online systems or Hardware admin. If evidence looks complete but prediction still does not return, capture the equipment id and the evidence panel state and raise with admin; the model service may need a restart.

A recent change is not visible

Hard-refresh the page (Ctrl plus Shift plus R) to bypass the browser cache. Sign out and sign back in to refresh the auth-driven content. If the change is an admin commit, allow up to a minute for the change to propagate to the page cache. If still not visible, capture the page URL, the role in effect, the timestamp and a screenshot of the page versus the expected state. Provide this evidence to admin, who can check the live md5 floor against the expected post-change md5.

An import flags inconsistencies

The validation step in Data integration flags six common cases: missing required IDs, duplicate names within a parent, parent-child cycles, unsupported node types, mismatched column counts and ambiguous column mappings. For Excel threshold imports specifically, an unmodified node id and node path column is a precondition; modifying either breaks the anchor and produces orphan rows. Fix at source where possible. If you must override an inconsistency during import, document the override in the import notes so the next operator understands why the imported file differs from the live structure. Never commit an import without resolving every red row in the validation report.

A gateway or sensor looks offline

Open Hardware admin and select the gateway. Check the indicator-light state recorded on the on-site photo: a healthy gateway shows the run light flashing, the network light steady and the status light steady once the connection is established (around forty seconds after power-up). If the network light is off, the WAN uplink is down; check the on-site cable, the WAN configuration and the SIM where the gateway uses a cellular uplink. For sensors, open the per-sensor self-diagnosis row and check battery voltage, internal temperature, retransmission count and delay status against the gateway. A high retransmission count usually means the wireless mesh signal strength is below the recommended commissioning threshold (around 20); reposition the antenna or relocate the sensor closer to the gateway. The MPU log download captures internal sensor state for deeper review.

Captures look noisy or impulsive

Open the equipment in System config and check the data collection trigger settings. A continuous (trigger) collection fires when an acceleration, velocity, rotational-speed or PLC rotational-speed trigger is reached, with a configurable number of waveform groups per trigger. Two impact thresholds gate the capture: an acceleration impact threshold (forces an immediate collection regardless of interval) and a kurtosis impact threshold (discards the current waveform and collects the next one). If captures are unusually noisy, the kurtosis threshold may be too high (suspect waveforms are being kept) or too low (clean waveforms are being discarded). Compare the impact-threshold values against the equipment's known operating profile and tune in small steps. For wireless sensors, very impulsive captures sometimes indicate poor magnetic-base coupling; if the sensor was installed by magnet alone without the bonded option, schedule a re-installation using the bonded-base path during the next planned outage.

Glossary

Common terms used across the Trend portal. Each term lists where it appears, a practical example and why it matters when interpreting evidence.

Asset

A maintained item or system in the hierarchy, such as a mill, pump or drive group. Appears on Asset Dashboard, Asset Diagnostics, Maintenance History and Fault cases. Example: Rod Mill 2 is an asset; its motor, pinion shaft and pump assemblies are equipment under that asset. The asset is the level at which most operational reviews happen.

Equipment

A selectable machine or component level where mounted sensors are consolidated for analysis. Appears on Equipment Health Index, Equipment gallery and Predictive Analytics. Example: a pump-motor assembly is equipment; the bearings on that pump-motor are sensors. Equipment is the level at which the EHI score is computed and at which predictive analytics is most meaningful.

Hierarchy

The governed plant, asset, equipment and sensor structure controlled through System config. Recognises three node types: organisation, equipment and measurement point. Only organisations and equipment can hold children. The hierarchy is the spine of the portal; nearly every page consumes it. A change to the hierarchy propagates everywhere within roughly a minute. Treat it as a release, not a casual edit.

Sensor

A physical device mounted on equipment to provide vibration, temperature, speed or audio evidence. Identified by a serial number and bound to a gateway channel. A typical wireless triaxial vibration and temperature sensor delivers vertical (Z), horizontal (X) and axial (Y) channels with the magnetic-base red mark indicating the X-axis direction at installation. A sensor without an active binding is spare; a sensor that is bound but silent is the first thing to check when data goes missing.

Measurement point

A configured leaf node through which a scalar value, waveform or diagnostic feature is stored. A single sensor can deliver multiple measurement points (a triaxial sensor offers X, Y and Z axes plus a temperature channel and, where configured, an audio channel). Measurement points are what threshold rules apply against and what charts plot.

Scalar value

A single numeric reading such as temperature, velocity total or high-frequency total. Scalars drive the alarms surface and the KPI cards on Asset Dashboard. A scalar trending up is the earliest indicator that something is changing; a scalar above its configured threshold is what raises an alert. The portal's core vibration scalars are low-frequency acceleration RMS, velocity RMS, high-frequency acceleration RMS, temperature, rotational speed, bearing envelope and gear envelope.

Waveform

A time-based signal capture used for vibration evidence and deeper diagnostic review. Waveforms are captured at high sample rates and stored at lengths from 8K to 64K samples; FFT-derived spectra run at 3200 to 25600 lines depending on the configuration. They form the basis of spectra, envelopes and orbit plots. Always check the most recent waveform when an alarm fires; the scalar tells you something changed, the waveform tells you what changed.

Long waveform

A high-frequency acceleration time series captured over an extended period, similar in concept to a digital tape recorder. Used when a transient or intermittent event needs to be captured for after-the-fact review. Long waveforms can be filtered, cursor-measured and spectrum-analysed in the same way as ordinary waveforms but cover a longer window.

Spectrum

A frequency-domain view derived from a waveform, used to identify patterns linked to rotating equipment behaviour. The spectrum shows energy per frequency band; specific frequencies link to specific fault modes (unbalance at running speed, bearing inner-race frequencies, gear meshing frequencies). Overlay calculated bearing frequencies against the captured spectrum to confirm or refute a bearing diagnosis.

Envelope demodulation

A demodulated signal view commonly used for early bearing and impact-related evidence. The envelope strips the carrier signal and surfaces the modulating signal, which makes low-amplitude repeating impacts visible. Configured per measurement point with a centre frequency and a filter bandwidth; the working ranges in the portal are 250 to 10000 Hz for bearings and 500 to 20000 Hz for gears. An envelope spectrum is the most sensitive view for early bearing defect detection, often weeks before the same defect shows in a velocity spectrum.

Cepstrum

A second-order spectrum view that separates periodic components from a busy frequency-domain plot. Useful when sidebands and harmonics obscure a fundamental signature. The cepstrum peak corresponds to a modulating period; convert back to frequency to confirm the modulator. Particularly useful for gear mesh evidence and for bearing harmonics that overlap a structural resonance.

Waterfall plot

A three-dimensional spectrum view that stacks consecutive spectra along the time axis to show how frequency content evolves. Use to spot whether a peak is steady, growing or transient. Reading the waterfall is faster than scrolling through individual spectra when you are trying to date the start of a developing signature.

Cross phase

A phase comparison between two channels, useful for separating shaft rotational behaviour from structural response. Common in alignment, unbalance and resonance investigations where a single-channel signal cannot distinguish between source and structure.

Order analysis

A spectrum representation normalised against rotational frequency, where the horizontal axis shows orders of run speed rather than absolute frequency. Useful on variable-speed equipment because order peaks stay in the same place regardless of the operating speed. Order envelope extends the same idea to envelope-demodulated signals.

Envelope trend

A long-term trend of the envelope peak or peak-to-peak value, plotted against time. Used as a leading indicator for bearing and gear faults: the envelope often climbs steadily for weeks before any velocity scalar moves. Bearing envelope trend is configured at 250 to 10000 Hz, gear envelope trend at 500 to 20000 Hz.

Time-domain index

A family of scalar features computed from a time-domain waveform: total value, peak, peak-to-peak, kurtosis, pulse index, waveform index, margin index, skewness, root index, maximum, minimum and mean. Each carries diagnostic meaning; kurtosis is sensitive to impulsive faults, peak-to-peak is sensitive to gross damage and crest-related indices distinguish steady from impulsive signals.

Frequency band energy

An energy scalar computed across a configured frequency band rather than the full spectrum. The portal supports low, mid, high and ultra-high band splits per measurement point. Useful when a defect signature lives in one band but does not move the overall RMS enough to fire an absolute threshold; combine with a multi-band alarm rule to flag developing damage early.

Alarm

A rule-driven notification raised when configured limits or states require attention. Alarm states are unresolved, in progress and closed; new alarms on the same equipment group under the existing alarm event until the event is closed. Alarms appear in Asset alerts and surface as the notification bell badge. An alarm has a severity, a status and a configured rule; the rule lives in System config under scalar value alerts.

Threshold

A configured limit used to classify readings, raise alerts or support predictive analytics. Three rule types: normal (absolute per level), automatic (computed from 100 to 500 historical samples) and multi-band (separate thresholds per frequency band with upper and lower limits). A relative-value rule fires on a sudden delta even before the absolute threshold trips. Threshold profiles vary by equipment type and operating condition; a baseline that suits a running condition will give false alarms during start-up.

Trigger collection

A capture mode where a waveform is recorded when a configured trigger value is reached rather than on a fixed schedule. Trigger types include acceleration, velocity, rotational speed, acceleration paired with rotational speed, velocity paired with rotational speed and PLC rotational speed (bound to an RTU channel). Two impact thresholds gate behaviour: an acceleration impact threshold forces an immediate collection and a kurtosis impact threshold discards an over-impulsive waveform in favour of the next one.

Equipment Health Index

A consolidated health score for selected equipment, based on mounted sensor evidence, threshold state, recent captures, fault history and reliability rules. Scored 0 to 100. Good is 80 to 100, Moderate is 60 to 79, Poor is 40 to 59 and Critical is 1 to 39. Pending means no valid timestamped health record exists.

Asset Diagnostics

The evidence workbench for viewing diagnostic signals, waveforms, trends and related context. Reached by URL asset-workbench.html; the URL is stable, the label is Asset Diagnostics. This is the page where signal-level investigation happens; every other page describes the result rather than the raw evidence.

OptimalAI

The assistant and model layer used for predictive analytics, anomaly checking and guided evidence review. Sits behind the Predictive Analytics page. OptimalAI is equipment-led and evidence-anchored; it does not act on its own but presents evidence for human confirmation. A prediction is a hypothesis until a human label confirms it.

Anomaly

A pattern that differs from learned normal behaviour. An anomaly is not a fault by itself; it is a signal that something has changed and needs review. Useful as an early-warning indicator before the change reaches an alarm threshold or a fault signature is identifiable.

Fault case

A structured record of defect evidence, analysis, action and outcome. A good fault case has: a symptom description, at least one captured waveform or spectrum, a confirmed diagnosis with frequency or pattern match, the recommended intervention, the executed intervention with date and operator and the post-intervention verification.

RCA

Root cause analysis, the discipline of connecting symptoms to causes and corrective actions. RCA on the portal lives inside Fault cases. A complete RCA distinguishes immediate cause (what failed) from root cause (why it was able to fail) and from systemic cause (why the failure pattern keeps recurring).

RUL

Remaining useful life, an estimate of how long an item can continue before a planned intervention. RUL is an output of predictive analytics and is most reliable when applied to wear-driven failure modes (bearings, gears, seals). It is less reliable for failure modes that are step-change rather than progressive.

Route

A configured inspection or collection pathway for sensors, equipment or manual checks. Routes are common in lubrication and spot check work. A well-designed route reflects the physical walking order on the plant floor; a route that lists assets out of physical sequence wastes inspector time.

Gateway

A collection device that connects field sensors to the Trend tenant. A single gateway typically serves up to about 45 wireless sensors on a short-range mesh with a working line-of-sight of around 300 metres. The uplink to the server is Ethernet, WiFi or a cellular WAN. Gateway run, network and status indicator lights signal power, network connectivity and link state. Gateways appear on Online systems and Hardware admin; the bound-versus-spare-versus-offline count per gateway is the headline metric for data collection health.

Wireless mesh signal strength

The link-quality reading between a wireless sensor and its gateway. A minimum value of around 20 is the recommended commissioning threshold for healthy live operation. A reading well below the threshold predicts retransmissions, battery drain and intermittent captures; reposition the antenna, relocate the sensor or shorten the path. Always confirm signal strength on the bench in the live mounting position before bonding the magnetic base.

Acquisition definition

The capture configuration on a measurement point: waveform interval, long waveform interval, audio duration and sampling interval, plus the waveform length and frequency range. Sync to device applies the same acquisition definition to similar measurement points under the same equipment in one action.

Component model

The shafts, bearings and gearboxes configured against equipment that allow characteristic frequencies to be calculated. A bearing entry derives inner-race, outer-race, rolling-element and cage frequencies from internal dimensions. A gearbox entry uses the tooth counts of driving and driven gears across parallel axes to compute the speed ratio chain. The shaft marked as the speed shaft anchors the chain at a speed ratio of 1.

Self-diagnosis

The periodic check that an acquisition unit (gateway) and its sensors run against themselves and upload to the portal. Aspects covered are collection, storage, network status, transmission, system, rotational speed and modules. Per-sensor self-diagnosis returns battery voltage, internal temperature, retransmission count and delay status against the gateway. The MPU log download captures detailed internal sensor state for vendor review when the surface diagnostic does not explain a symptom.

Import mapping

The review step where incoming file columns are mapped to required hierarchy or configuration fields. The mapping wizard suggests high-confidence mappings automatically; you confirm the low-confidence rows manually. A correct mapping reduces the validation work later; a sloppy mapping floods validation with avoidable inconsistencies.

Validation

The check that confirms an imported or configured record is complete, coherent and ready to commit. Validation runs after mapping and before commit. A red row in validation must be resolved at source or explicitly overridden before commit; never commit an unresolved red row.

Role

The access level assigned to a user, controlling which pages and actions are available. Three roles: client, manager and admin. Client is the bottom tier and is scoped to read access on the assets they own. Manager adds field-engineer write access. Admin adds tenant governance. The role chip top-right shows which role is currently in effect.

Tenant

The dedicated client environment, including its portal, data stores, runtime services and configuration. Tenants are isolated: a change in one tenant does not propagate to another. The tenant is the level at which licensing, billing, hierarchy and user management are scoped.

No help article matched the current search.