Automatic Operating Regime Selection in Local Modeling

Recently, local modeling has attracted much attention to identify the complex systems. In local modeling, global system model is obtained by combining a number of local models, each of which has simpler structure and has a range of validity less than the full range of operation. Since the local models are identified for corresponding local operating regimes, the performance of the global model is highly affected by the choice of the local operating regimes. This paper addresses automatic selection algorithms of suitable local regimes in local modeling. Based on three criteria, Kullback Discrimination Information (KDI), Akaike Information Criterion (AIC), and Mean Square Error (MSE), we develop regime integration and partition algorithms. Numerical simulation studies illustrate the applicability of the proposed selection algorithms.