| Time delay characteristics generally exist in the industrial processes. For this reason, current control action will not be reflected in the output signal until after some time intervals. Therefore, the difficulty for the designing of control system will be increased greatly. For most of these processes, their primary dynamic behaviors can be approximately described by rational transfer function plus pure time delay. The model parameters estimated from testing or actual running data can be helpful for the subsequent design of the controller and the analysis of performance of the control system.Firstly, some fundamental knowledge in the process parameters estimating is analyzed, such as selecting driving signal, collecting and pretreating data, choosing and checking model structure, detecting error source and determining evaluation criterion for estimating error. Meanwhile, several commonly used least-square estimating methods with their coherence and unbiasedness are listed. Secondly, after finding methods to identify the delay time and obtaining estimating equation by multiple integration operation, the idea of processing unknown initial conditions is discussed. Thereupon a new mean suitable for all orders is presented. The aim that deleting these conditions is achieved by introducing time variables in the multiple integrating process without adding new parameters in the measurement vector and parameter vector under partial integration principle. With this mean, open loop direct and iterative parameter estimating methods are realized. Finally, the application of the iterative estimating method in the closed loop condition is discussed.To detect the effectiveness of the estimating method, some simulation experiments compared with other techniques are implemented both in open loop condition and closed loop condition by MATLAB software. The result shows that the methods can hold an excellent accuracy level with some deviations in the noisy environment. |
PDF Full Text Request