ALTERNATIVES TO NEURAL NETWORKS FOR INFERENTIAL MEASUREMENT

Neural networks have attracted much attention as a means of modelling nonlinear phenomena, for example for inferential measurement in proces s control. A neural network is a nonlinear multivariable function whos e main potential advantage: the ability to represent highly nonlinear input-output relationships, is in fact not essential in many potential process engineering applications. This advantage is in any case frequ ently outweighed by its major disadvantage: the intractability of the parameter estimation procedure resulting from the highly nonlinear for m of its parameters. In this work we show how moderately nonlinear fun ctions, with easily estimated parameters may be used in certain infere ntial measurement applications for which neural networks have been pro posed. These functions are as effective in representing input-output r elationships and their parameters can be fitted far more rapidly than can the ''weights'' of a neural network. Furthermore, we show that the performance of both these functions and neural networks, being arbitr ary representations having no physical basis, may almost invariably be improved upon by the use of even very simple approximate models based on proper physical understanding.