Design Of Multi-variable Model Testing Signals And Identification

Nowadays, the key problem of MPC is how to get the models of the process rapidly and exactly. System identification based on SISO system has been well developed but the models in industrial process are always MIMO systems combined with interacted variables. The normal system identification method restrict the mpc which applying in industry because of its long time calculating and complexity. Therefore, the primary research object in system identification is the identification of MIMO systems.As the impulse response models are widely used in industry MPC arithmetic and software a representative approach of non-parameter model system identification -correlation analysis method is studied in the thesis to test signal and identify the MIMO system accurately and effectively. The main contribution includes:1. Choosing Pseudo-random Binary sequence (PRBS) as the modelling testing signal of the system and analyzing the relationship between the multi-PRBS signals' irrelevant interval and the detention of signals. Compared with the normal method, a new designing approach was proposed to make full use of the irrelevant interval. Compared with the normal method, the proposed method can effectively shorten the periodic length of PRBS especially to those MIMO systems which have great different settling times of different channels.2. A new designing approach of the relevant identification method is proposed based on the optimized multi-PRBS signals designing. The new approach makes the best use of the PRBS signal's irrelevant interval and the minimal PRBS periodic length is merely the sum of the settling time of each channel. Compared with the normal method, the simulation shows that the proposed method can effectively shorten the periodic length of PRBS.3. By using MATLAB some examples of SISO systems and MIMO systemswere discussed and the normal design method and optimized design method were compared to the same system to validate the conclusion. The emphasis of the research work is the study of MIMO system in the aspect of frequency which can validate the method better.