condorcontrol team has developped a robust methodology and a software framework to build real time optimal controllers for industry processes. It makes use of recent progress in machine learning and efficient implementation of state of the art robotics control algorithms.
The software framework is used as a brick to build process specific products. It leverages knowledge about the physics of the system and uses the sensors output of the plant to specialize and improve the controller using statistical methods and machine learning.
Optimality can be a nebulous concept. The most common approach is to define it as achieving the maximum profit for the plant within a specified time frame, taking into account the prices of input and output products and utilities. However, there may also be process constraints, production constraints, and trade-offs for equipment and energy savings. Therefore, the first step in building an optimal controller is to clearly define what optimality means for your plant.
We also have a strong focus on the maintenance of the controller, so that it keeps bringing you value. We keep indicators to estimate the performance of the controller, the model drift, etc. We will perform updates of the controller on a regular basis.
