Model identification of thermal process in power plant

The identification methods of industrial process we discussed. The applied MATLAB program of least squares method is given and used to identify pulse transfer function. A kind of improved genetic algorithm is introduced to identify the transfer function and parameters of nonlinear dynamic model of thermal process. In the algorithm, floating-point coding, rank-based selection, elitist reservation and grouping method are used, the premature convergence is restrained, and the global and local searching ability is improved. With the genetic algorithm, the transfer function of topical thermal process can be identified accurately. The parameters of nonlinear model can be modified according to operating data of power plant, no matter what kind of input signal is used, such as step signal, random operating signal.