Applications of Neural Nets and Evolutionary Programming to Process Monitoring

Process modeling benefits are well understood: better quality, and minimal scrap, environmental impact, and unplanned maintenance. The cost of implementing and maintaining process models, however, has limited application of model-based process monitoring and control.

New technologies have changed this, with adaptive process modeling reducing the cost of developing and maintaining process models, and thus broadening applications. Applications of neural networks and evolutionary programming have demonstrated quantifiable benefits in process performance, maintenance costs, emissions, and scrap rates. Discrete part, and batch and continuous processing applications are presented to illustrate application qualification criteria and typical costs and benefits.