Improved fuzzy control through the inference of difficult to measure parameters

Researchers at the U.S. Bureau of Mines have developed an innovative approach to process control that combines the control capabilities of fuzzy logic, the search capabilities of genetic algorithms, and the modelling capabilities of neural networks. One of the key aspects of this approach to process control is the use of a neural network model to infer information from the physical system that is difficult or expensive to measure directly with sensors. Often this unmeasured information is critical to successful control of the system. The unmeasured system information can be inferred by employing the search capabilities of genetic algorithms. In the approach presented, a genetic algorithm is used in conjunction with a neural network model of a physical system and sensory information that is available to obtain needed information that cannot be measured directly. The effectiveness of this approach is demonstrated on a specific system from the mineral processing industry, a hydrocyclone separating device that is used to achieve physical separation of mineral samples