DESIGN OF AUTOMATED SUPERVISORY CONTROL & ADVANCED PROCESS CONTROL BASED ON MACHINE VISION OR SENSOR DATA FOR REAL TIME PRODUCTION PARAMETERS TUNING
Adjusting the settings of your machine or production process by means of human intervention is perfectly suitable if such adjustments are needed only infrequently, and if the need to make adjustments can be easily and quickly detected. However, that is not always the case.
For example, standard machine vision applications are used to check quality or presence and generate an alarm or send a signal to an actuator when a problem is detected. Changes to the process parameters to avoid bad quality or defects are usually done offline, manually. This leads to long periods of sub-optimal production quality. The detection of problems occurs only after they cross the threshold set into the computer system.
For example, standard machine vision applications are used to check quality or presence and generate an alarm or send a signal to an actuator when a problem is detected. Changes to the process parameters to avoid bad quality or defects are usually done offline, manually. This leads to long periods of sub-optimal production quality. The detection of problems occurs only after they cross the threshold set into the computer system.
Thanks to the combination of our competences in computer vision, machine learning and advanced (process) control, we can go one step further: derive process parameters from measurements made on industrial camera images and/or make use other sensor and process data. Predictive models and artificial intelligence algorithms first detect anomalies or process drift in-line in pictures and/or sensor data. The smart software then adjusts the right process parameters to prevent the problems from occurring, or to at least reduce the damage caused by the problems.
The result is automated tuning of process parameters - supervisory control - in real time to optimize product quality and avoid problems and defects as much as possible. It also improves operational efficiency and productivity and reduces downtime.
Some reference projects:
The result is automated tuning of process parameters - supervisory control - in real time to optimize product quality and avoid problems and defects as much as possible. It also improves operational efficiency and productivity and reduces downtime.
Some reference projects:
- Real-time adjustment of the pressure and temperature set points of refiners through the extraction of features characterizing production quality and efficiency from color and thermal images, combined with a supervisory control algorithm that maximizes production volume & quality. Customer under NDA.
- Real-time adjustment of the steam flow set points of gas turbines and backup boilers based on the measurement of steam flow demand, pressure controls, and a predictive model of the electricity price on the intra-day market. This was realized with customer Umicore (non-ferro metals) in Belgium as well as with a non-disclosed customer in the Netherlands.
- Real-time adjustment of the setpoints of a tunnel oven to bake cookies that have precisely the desired brown color for customer La Confiance.