The nonlinear identification of a heat exchanger

The practical application of a recently introduced orthogonal parameter estimation algorithm is investigated by identifying a nonlinear model of a heat exchanger based on input-output records. A new criterion called ERR. (Error Reduction Ratio) is employed to select significant terms in the NARMAX model expansion to yield a parsimonious model. The fitted model is then compared with a model previously obtained using a prediction error algorithm coupled with stepwise regression, and validated by computing various correlation tests and plotting predicted outputs.