The Optimization And Experiment Of The Point-Location Algorithm In Explicit Model Predictive Control

In the online computation phase of the explicit model predictive control (EMPC), the core task is to solve the point-location problem. The method to solve this problem, so called point-location algorithm, has three main performance metrics:the storage required by space partitions and characteristic values, the efficiency of solving the problem online, and the time needed to process data into required structure. These metrics are in relation to if the EMPC could be widely applied to increasingly miniaturization, high efficiency, or rapid-designed control situations.On the basis of several classic point-location algorithms, four methods called synonymous redundancy-reduced grid method, two-levels grid method, fast binary tree method and grid binary tree method are put forward step by step. The first two methods both reduce the storage requirements in a certain extent and increase efficiency of the online search drastically. The third one has reduced the time consumed in the pre-processing of the binary tree method without reducing the performance of the other two aspects. And the last one has the highest online search efficiency.The main works and achievements of this paper are as follows:1. Based on consulting a large number of related literatures, the basic principles and research status of the EMPC and its point-location algorithm are introduced.2. Introducing the theoretical basis of point-location algorithm, such as polytope theory, data structures and algorithms.3. Based on the hash table method, this paper puts forward the synonymous redundancy-reduced grid method and the two-levels grid method, and than expounds the principle and realization of these two methods.4. By optimizing the process of establishing a binary search tree, the fast binary tree method is proposed. Than this paper combines the fast binary tree method with the synonymous redundancy-reduced grid method to propose the grid binary tree method, and than expounds the principle and realization of these two methods.5. By three experiments, eight methods in this paper are realized. Than this paper tests and compares their performance in three aspects, analyses their advantages, disadvantages and applicable occasions, and proves the superiority of the newly proposed methods at the same time.6. Based on the MATLAB Multi-Parametric Toolbox, this paper develops software for the design and analysis of the EMPC point-location algorithm.7. Finally, this paper summarizes the full text, and puts forward some prospects for further research.