Discovering patterns in historic operating data
If historic data relating to the operation of your plant or process is available, then XpertFactory can quickly turn it into insights/knowledge that can help you understand the root causes of plant faults and inefficiencies. Our explainable machine learning allows a process/manufacturing engineer to start from a data file and within minutes generate understandable tree patterns relating the key process indicators to other process operating parameters. The transparency of the discovered patterns can help domain experts to infer root causes of problems thereby allowing them to devise corrective actions offline or online. The machine learning patterns can also be used for real-time monitoring of the process.
This can all be done offline so minimising risk as manufacturing operations do not need to be interrupted and can lead to the development of a no risk/low risk digital strategy, that is cost effective.
The machine learning wizard will empower a non-data scientist to derive tree-based patterns in minutes. The resulting machine learning model segments the operating data into a number of understandable operating zones or regions.
Real time Condition Monitoring
Monitoring allows plant operators to predict in real-time, if a plant failure will occur or if a machine might malfunction thereby allowing predictive maintenance. Monitoring can also help improve product quality, process efficiency and energy usage by analysing data in real-time to detect abnormal and sub-optimal operating conditions and to advise on remedial actions. XpertFactory supports comprehensive approaches to process/plant monitoring.
Automated Anomaly detection
Descriptive Analytics is the inspection of the data to determine if anything abnormal is happening Unlike other monitoring systems that simply flag individual process variables falling outside minimum – maximum operating ranges, XpertFactory builds complex machine learning patterns combining multiple interacting process parameters representation normal operating conditions. This allows XpertFactory to detect any anomalous previously unseen combination of process parameters. Furthermore, it gives an understandable description of the anomalous pattern.
In addition to anomaly patterns generated by machine learning, XpertFactory allows the definition of anomaly rules/patterns by domain experts together with the associated corrective actions.
XpertFactory can build predictive machine learning models from historic data which when used for real-time monitoring can predict plant / process failures before they occur. This includes our advanced detection of pre-fault time signal signatures using times series/frequency analysis and time signal similarity algorithms.
Root cause analysis
XpertFactory can capture the best practice expertise of engineers and technicians in diagnosing the cause of faults or detecting symptoms/patterns of faults and sub-optimal operations. The expertise is captured as decision trees, decision tables and pattern rules which can be deployed to provide 24/7 automated monitoring of the process / plant. Diagnostic rules can also be generated using Machine learning if sufficient occurrences of the fault data is available.
Condition monitoring as an engine API
The XpertFactory condition monitoring engine can be used as restful service providing anomaly detection, predictive analytics, and diagnostic alerts for use in conjunction with any Industrial Process Performance Monitoring (dashboarding) system deployed on the edge or cloud. Alternatively, XpertFactory comes with its own comprehensive Dashboarding system which can be deployed on web or mobile web and is integrated with the XpertFactory conversational Advisory Bots.
Conversational Advisory Bots
XpertFactory allows the capture and automation of corrective actions, trouble-shooting and standard operating procedures. These can be deployed as Conversational Advisory Bots that are available on demand by operators or are invoked as prescriptive Bots by the condition monitoring engine. Our “Bots” use both patterns from data and captured human rules to ensure the best decisions and guided actions are delivered consistently 24/7 to plant operators. The Advisory Bots can be deployed through any contact channels: web, mobile web and various messenger channels and again can be deployed on the edge or via the cloud.
Performance optimization and control
It is often desirable to operate a process or plant in an optimal manner by maximising a balance of competing Key Performance Indicators such as throughput, quality, energy use etc. XpertFactory supports our patented technology for performance optimisation and control. It involves building multiple understandable machine learning models for the various process KPIs, fuzzifying them for increased accuracy and then using an evolutionary algorithm that uses these models, production requirements and real time operating conditions to derive optimal process settings that balances the competing KPIs.
The optimal process settings can be used in (open loop) Advisory mode or they can be used in a closed loop as target settings for the plant control systems.
Edge - Cloud
The XpertFactory small footprint high performance Decision, Machine Learning and Conversational engine allows you to deploy the full Condition monitoring, diagnostic, advisory and control solutions where you need them; onsite (at the edge) or remotely in the cloud or on an intranet server. This is made possible by our advanced Containerised node.js Decision & AI engine which can run anywhere from a small compute device at the edge to a cluster of cloud servers. The engine also has interfaces to various PLCs, Data loggers, OPC servers, Data Historians, databases and industrial IOT platforms.
Want to find out more?