Improving Efficiency and Product Quality by Optimizing Steady State Neural Network Process Models, 2001 Process Control, Electrical & Info. Conference Proceedings

Jeffrey DeCicco--Explicitly or implicitly engineers and operators utilize models to operate and optimize processes. Neural network modeling and optimization formalizes and simplifies this procedure by incorporating historical data with a basic understanding of the process. Improving process operation involves defining the overall goal or objective function i.e. increased capacity, higher quality, lower energy costs, etc. with the constraints of the process i.e. available equipment or raw materials, etc. The objectives and constraints are incorporated with a neural network model representing the steady state operation of the process to produce a set of operating conditions that meet desired objective function. Limited working knowledge of neural network modeling and optimization is required rather a good understanding of desired objective and constraints is essential. A relevant example is provided that illustrates the strategy highlighting the importance of good engineering practices rather than detailed theory of neural network modeling and optimization.