ThingWorx Analytics Training: Module 9 Part 2
- December 23, 2022
- 0 replies
- 1724 views
This video continues Module 9: Anomaly Detection of the ThingWorx Analytics Training videos. It begins with a ThingWatcher exercise, and concludes by describing Statistical Process Control (SPC). The "SPC Accelerator" will be covered in Module 9 Part 3.
This video contains a hands-on exercise which you may complete as you wish. However, it's not necessary, as the following modules do not rely on the product of this exercise. It shows you how to configure a property on a thing for anomaly detection, and view the results in a mashup. Download and import the attached "Project_AnomalyDetection_ThingWatcher.xml". NOTE: It is recommended to import into a sandbox or non-production environment running ThingWorx/ThingWorx Analytics 9.2 or later.
Chapters are available in this video — use the chapter menu in the player to jump to any section.
Chapter Summaries
0:00 – Intro: Thing Watcher Anomaly Detection Demo with a Simulator
Kicks off a hands-on walkthrough of configuring ThingWorx Thing Watcher anomaly detection on a property, using a data simulator to stream values because the environment isn't connected to a live physical asset.
0:21 – Blower Timer Entity: Numeric Properties – Air Flow & In/Out Temperature
Opens the blower timer entity from the Timers folder and inspects its four numeric properties — air flow rate, in temperature, out temperature, and spare air flow rate — refreshing to confirm the values update frequently.
1:04 – Creating a Logged Property (out_temperature_C) on the Anomaly Services Thing
Adds a new Number-type property named out_temperature_C on the AnomalyServices_92 Thing and marks it persistent and logged so values save to the value stream, giving Thing Watcher the data points it needs for its calculations.
2:37 – Binding the Property to the Blower Data Simulator via Manage Bindings
Uses the Manage Bindings dialog to search "*blower" and drag the four-sided arrow to bind the new property to the simulator's outlet temperature tag, confirming the green link icon and refreshing to see live values flow in.
4:03 – Configuring a Thing Watcher Anomaly Alert & Alert Types
Creates an "out temperature anomaly" alert on the bound property and reviews the available alert types — above/below threshold, within range, equal to a value — before selecting the Anomaly alert type.
5:17 – Anomaly Parameters: Outbound Anomaly Rate & Minimum Data Collection Time
Walks through the anomaly parameters, keeping the 5-second outbound anomaly rate and 1-minute minimum data collection time, and explains how Thing Watcher needs roughly 600 data points to learn a property's natural fluctuations before training.
6:36 – Certainty Percentage: Tuning Sensitivity for Critical vs. Non-Critical Assets
Lowers the default 95.4% certainty percentage to 85% and explains the trade-off: a lower value reports more frequent but less confident anomalies (ideal for critical assets), while a higher value returns fewer, more certain alerts for non-critical assets.
7:34 – Saving the Alert & Thing Watcher States (Initialized → Calibrating → Monitoring)
Leaves priority at the default of one, saves the alert, and shows its "initialized" status while Thing Watcher collects values, previewing the progression to calibrating and the subsequent monitoring states.
8:12 – Anomaly Dashboard Mashup: Real-Time Plots & Training Chart
Opens the anomaly_dashboard_92 mashup, uses the entity picker's magic-wand to select AnomalyServices_92, and reviews the training-points chart alongside the live out_temperature_C blue plot overlaid with the red 0/1 anomaly-status plot.
9:48 – Restart & Retrain: Rebuilding the Anomaly Model
Demonstrates the Restart and Retrain buttons, showing how Thing Watcher drops from monitoring back through initialized, buffering, and calibrating as it collects new data points and trains a fresh anomaly model.
11:07 – Statistical Process Control (SPC) in ThingWorx: Concepts & Benefits
Introduces Statistical Process Control and its benefits inside ThingWorx — instant process feedback, fewer production-line failures, less scrap and rework, reduced process variability, and anomaly labeling for use as features or goal variables in predictive models.
12:19 – PTC SPC Accelerator: Western Electric Rules & Cp/Cpk/Pp/Ppk
Presents PTC's free SPC accelerator as a turnkey launching point that combines streaming telemetry ingestion, SPC logic built on the Western Electric rules with Cp, Cpk, Pp, and Ppk values, and design-spec checks with the ability to annotate or ignore points.
13:23 – SPC Charts & No-Code Config GUI: Xbar/R/S/np, Production Lines & CSV Import
Covers the accelerator's chart types — Xbar, R, S, np, and capability charts plus historical analysis — and its no-code configuration GUI for grouping production lines and assets, setting long-term control limits, and importing/simulating CSV historian data.