Study On Methods Of Stochastic Optimal Operation Of Multi-Reservoir Based On IGA-SA

Stochastic Dynamic Programming (SDP) is a basic method to achieve optimal operation of reservoir, while which is applied to multi-reservoir containing multi-annual regulating reservoir with characteristic of unregulated cycle, the problem of the "curse of dimensionality" will be aroused and the characteristic of multi-annual regulating reservoir can't be showed. Thus a method of multi-level optimization decompounded the optimal problems of multi-reservoir into two levels where the first level and the second level was optimized with SDP and Improved Genetic Algorithms & Simulated Annealing Algorithms (IGA-SA), respectively, has been applied and the two problems mentioned above were solved. The new method and new theory brought forward by this dissertation were proved by the example of multi-reservoir in series of Wujiang River, GuiZhou Province, China, and plentiful and substantial research outcomes as following were gained:(1) The model of optimal operation of multi-reservoir in series with a objective of maximal annual capacity of generate electricity was built and solved with the method of multi-level optimization by which the optimal operation of multi-reservoir was decompounded into two levels where the first level and the second level was optimized with SDP and IGA-SA, individually. Thus the problem of the "curse of dimensionality" was solved and an algorithm of man-machine interaction operated with even outflow when SDP was appliedto multi-annual regulating reservoir and a method which describes space-time relationship among inflows of each reservoir of multi-reservoir were brought forward.(2) A better algorithms, that is IGA-SA, was created combined IGA which has better global searching capability with SA which has better local searching capability in order to strengthen global and local searching capability, enhance searching efficiency. Some parameters of IGA-SA algorithms such as crossover operator, mutation operator and temperature showed a increasing or degressiving trend in the course of evolution. Increasing trend of fitness value of the most optimal individual agree with which of even fitness value of population.(3) With the application of the new method and new theory proposed by this dissertation to optimal operation of multi-reservoir of Wujiang River, Gui Zhou Province, China, the multi-annual even capacity, of generate electricity of multi-reservoir in series increased 3.21 × 10~8 KW.h and the demand of firm capacity was satisfied. This shows the new method and new theory applied to optimal operation of multi-reservoir are effectual and the application field of SDP method has been widen.