The Application of LMS Adaptive Method in Time Delay Estimation for Order Reduction Identification

The actual industrial systems are usually high order system with long time delay. These characteristics will bring a lot of troubles in building the models. Sometimes we need to know the pure delay of the system when designing the controller and at the same time we hope the model has low order. The traditional methods like unit step response and pade approximation to estimate time delay have some limitations. In this paper, we will use LMS(Least mean square) adaptive method to estimate time delay first and then use ARMAX model to reduced the order. The simulation has been conducted using the actual industrial data. The industrial system´s order is very high even can reach 30 while we want to use this new method to reduce its order to about 10. At last, by comparing the performance index of 3 models, we prove this method can achieve the desired purpose.