Analysis Of Multi-rate Predictive Control System

In the conventional digital control system design, the samplers and preserves located at different positions were considered working at the same time. But, in many control systems, it is difficult to use a same period in all the positions. The study now about multi-rate control strategy mainly focused on the system which was limited by many external factors, and must be settled by multi-rate control strategy. When the limits did not exist, based on many particular systems, the study about which control strategy was better was little.In this paper, the develop situation of multi-rate control theory was introduced, based on the realization of dynamic matrix predictive control. For some particular systems, raised a problem existed in the control process that the time constants in every passageway had huge difference, and for two control strategies, multi-rate and single-rate, which one was better under the situation that the external limits did not exist.Firstly, the impact of the ratio and basic sampling period in single-rate and multi-rate systems were studied, under the situation that the model-plant ratios were different from each other, the single-rate and multi-rate systems had different control performance. The predictive models for the three types of multi-rate control system: input multi-rate,output multi-rate and generalized multi-rate were all established and the sampling period optimal algorithm was found as genetic algorithm. Then, took the generalized multi-rate system for example, the optimal sampling-period was obtained, and the contrast between the two strategies was made.Secondly, the multi-rate control strategy in single-variable system was applied to multivariable system, and the models of input multi-rate and output multi-rate control for the multivariable system were established. Based on the basic sampling period, genetic algorithm was used to realize the optimization of the sampling period multiple, and the contrast the performance under multi-rate and single rate control strategies, then obtained that we should select different strategies for different systems. Using the simulation to analysis the impact what the basic sampling period T had on system performance.At last, based on the input multi-rate control algorithm introduced in section four, error predictive model was developed and multi-variable input multi-rate predictive control algorithm which based on error predictive model was introduced. The simulation indicated that this algorithm could improve the control performance of multi-variable system when it had model-plant mismatch and random disturbs. And obtained the optimal sampling value, when the system was only a common one that the time const mismatch of model and plant was little, the performance under single rate was better then it under multi-rate control strategy, but when the time const mismatch of model and plant was large, a reversed conclusion was obtained.