A identification method of a nonlinear ARX model with variable order for nonlinear systems

This paper gives a identification method of new input-output model. In a identification of a nonlinear model, a nonlinear ARX model(NARX) is presented by Ohata, Furuta et.al. The NARX model consists of a set of ARX models with same orders at each output level. However, a systems order of a nonlinear system is different for each system state, usually. We propose new NARX model with variable order at a output levels. In addition, the proposed method is compared with the conventional NARX model by estimated accuracy. As a result, the conformance rate of the proposed method were larger than that by one of the NARX model. Furthermore, the mean and the variance of estimated error of the proposed method were smaller than one of the NARX model.