Parallel Dynamic Programming Algorithm And Its Application In Reservoir Group Optimal Operation

After several years of rapid development of hydro power in China, thirteen large hydropower bases are being built step by step nationwide. The reservoir group optimal operation of basin hydro power station has become the focus of the research in this field. The research of reservoir group optimal operation was mainly focused on the building of new model and the improvement of algorithm. But as the number of reservoirs is increasing and the optimal operation model is becoming more and more exquisite, the scale of calculation is expanded intensely. When solving these problems, low efficiency and easy to involve into local optimum were always the problems of traditional algorithm.Reservoir operation of hydropower station is a multi-stage decision process. Dynamic programming which could achieve globally optimal solution is the most effective way to solve such problems. But for the joint operation of reservoir group, it will cause limits to the application because of the difficulty of "dimension disaster" Therefore the main object of this paper is dynamic programming algorithm. The research is mainly about the study of the parallel processing, and the application to medium and long term optimal operation of reservoir group. The main research content is as follows:Firstly, Based on the structural characteristics of the dynamic programming algorithm, comprehensive parallel feasibility analysis of it was carried out, and the mutual independence characteristics of state points were excavated.The improvement was realized by importing the parallel computing and named pipes. Pipeline parallel dynamic programming was designed and in multi-core computing environment, medium and long term optimal operation of single reservoir was calculated and verified.Secondly, Aiming at the dimension disaster problem of the application of dynamic programming, Multi-dimensional dynamic programming model and its calculation steps were lucubrated. By improving its calculation model and reducing the memory usage of computer, solid foundation was provided to cascade dynamic programming algorithm program.Thirdly, On the basis of parallel feasible analysis of cascade dynamic program-ming algorithm, cascade parallel dynamic programming algorithm of two reservoirs was proposed based on the master-slave mode. And validation was carried out by taking Lixian basin cascade hydropower stations as an example.The calculation results show that the parallel dynamic programming algorithm can make full use of the computer resources, shorten the calculation time effectively while guarantee the quality of the algorithm at the same. The basis of theory and method was provided for the further practical application of parallel computing to joint optimal operation problem of large hydropower station reservoir group. And this research can provide meaningful reference for parallel programming studies of other optimal methods.