Reliability
Take the workwith you.
32 topics, two languages. PDF guides and sample data.
ML
Isolation Forest for sensor anomalies: how it works and where it misleads
ML
Multivariate process monitoring with Hotelling T² and SPE
Instrumentation
How to read an FFT spectrum: resolution, windows, peaks
Instrumentation
Envelope analysis for rolling bearing faults: steps and frequencies
ML
Without fault labels: unsupervised machine learning in industry
Reliability
Kalman filtering: a smoothed health indicator from a noisy sensor
Quality
CUSUM and EWMA: catching small, sustained shifts
Instrumentation
Nyquist and aliasing: how fast to sample sensor data
Instrumentation
ISO 20816 vibration severity zones: how to use them and where not to
Quality
Process capability indices: reading Cp and Cpk correctly
Reliability
MTBF, MTTR and Weibull: reading reliability metrics correctly
Control
PLC scan cycle and data sampling error
Networking
Modbus register maps: scaling and byte order
OT
OPC UA security: modes, certificates and trust lists
DataOps
How historian compression distorts machine learning
ML
SPC control charts and anomaly detection: same question, different language
Instrumentation
Vibration analysis basics: velocity, acceleration and spectrum
Control
Alarm management: fix the alarm system before adding AI
Networking
MQTT and Sparkplug: a shared contract for industrial data
ML
Baselines in forecasting projects: why the naive method must be beaten
RAG
A RAG chunking strategy for technical manuals
ML
Prevent leakage in industrial time series
OT
Define the access boundary of an OT AI pilot
DataOps
Prepare OPC UA data for AI
DataOps
Check data quality before calculating OEE
ML
Build a predictive maintenance pilot around a decision
ML
Design a dataset for visual quality inspection
RAG
Industrial RAG: how to measure answer quality
R&D
A one-page brief for an industrial AI pilot
Edge AI
Browser analytics with C++ and WebAssembly
ML
Alert thresholds, precision and recall
Experiment data and data card.
Lab samples are mathematically generated synthetic series with no real site or customer data. They are not suitable for training a production fault model.