Smart factories and Industry 4.0 brings together machine, process, and operator data in new and powerful ways. However, in the absence of scalable and insightful analytics, little value can be extracted from this data. Currently, the different types of data live in different databases and systems, which are integrated through a Smart Factory solution. Process Engineers extract data from such systems and use tools such as Matlab to model behavior.
Current data historian implementation follows best practices. Areas of improvement include:
- Process engineers should be able to develop analytical models of critical behavior themselves.
- Sensitive process data is not used in analytics because it cannot be exposed to vendors.
- Detecting conditions of quality and efficiency across raw materials, operator training, and machine data will result in substantial benefits in yield, quality, uptime, and productivity.
Falkonry enables MES users and process engineers to apply complex pattern recognition to their smart factory data. The results of pattern recognition are available as additional attributes on the equipment or batch, as appropriate. These attributes can be used to alert workers, trigger workflows, and be rendered in dashboard displays.
A Smart Factory solution integrated with Falkonry can use Falkonry’s pattern recognition as the back-end and surface the findings through familiar means to operators and technicians. Falkonry-based Smart Factory enables companies to scale data-driven processes across their plant and produce more relevant and accurate results than Anomaly Detection, Statistical Process Control, or calculated thresholds.
A custom solution can be put together by combining machine data, process database, and Falkonry in the following steps.
Step 1: Integrate data from machines and processes and supply it to one or more Falkonry Data Streams via Falkonry client libraries. If your data lives in a custom data lake, create a Falkonry connector to directly use the data from the data lake.
Step 2: Select signals from Event Buffer to combine in pipelines that can automatically generate cognitive models for the patterns in the smart factory data. At first, use controlled historical data for Event Buffers to ensure high quality models are generated.
Step 3: Process Engineers use Falkonry UI to review patterns as well as behaviors found from historical data and add facts, if not already made available from historical records. Evaluate the performance of Falkonry results against expectations on historical data to determine operational readiness.
Step 4: Turn on live monitoring in Falkonry UI and observe cognitive condition assessments flowing continuously into Smart Factory System.
Step 5: Enhance existing Smart Factory dashboards to include Falkonry condition assessments. Alternately, use generate notifications to issue alerts for undesirable quality or efficiency conditions based on Falkonry assessments.