PRACTICAL APPLICATIONS OF ARTIFICIAL NEURAL NETWORKS IN CHEMICAL PROCESS-DEVELOPMENT

Artificial neural networks (ANN) have the ability to map non-linear re lationships without a-priori information about process or system model s. This significant feature allows the network to ''learn'' the behavi or of a system by example when it may be difficult or impractical to c omplete a rigorous mathematical solution. Recently ANN technology has been leaving the academic arena and placed in user-friendly software p ackages. This paper will offer an introduction to artificial neural ne tworks and present a case history of two problems in chemical process development that were approached with ANN. Both optimal PID control tu ning parameters and product particle size predictions were constructed from process information using neural networks. The ANN provides a ra pid solution to many applications with little physical insight into th e underlying system function. The amount of data preparation and perfo rmance limitations using a neural network will be discussed. However, the properly applied ANN will generally provide insight to which varia bles are most influential to the model and evolve dynamically to the m inimum performance surface squared error. Neural networks have been us ed successfully with non-linear dynamic systems and can by applied to chemical process development for system identification and multivariat e optimization problems.