Utilizing a neural network modeling process to create an inferred instrument

Utilizing neural network models in process control has increased over the past few years. Neural network models have been used with success for off-line analysis, online control, inferred instrumentation, simulation, cost analysis, etc. In most cases, the success or lack thereof is dependent upon the process used to gather data, create the model, analyze results, and install the model online. This paper describes a neural network application project done by a company and the neural network modeling process (NNMP) that enhanced the success of their project. The company overview and dilemma, solution, neural network overview, NNMP, results, online installation and conclusion, are presented