Open AccessApproximate Dynamic Programming of Long Term Power Generation Schedule
Author(s) -
Zhencheng Liang,
Yiming Li,
Yixin Zhuo,
Yinsheng Su
Publication year - 2021
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/645/1/012071
Subject(s) - mathematical optimization , randomness , stochastic programming , computer science , curse of dimensionality , dynamic programming , recursion (computer science) , scheduling (production processes) , theory of computation , stochastic optimization , schedule , mathematics , algorithm , statistics , machine learning , operating system
With respect to the features in optimal scheduling of long-term power generation such as long cycle, large scale and strong randomness, the paper proposed a multi-stage optimized decision model based on approximate dynamic programming. In this model, the influence of random variables including load, fuel price, and water was tested by means of stochastic simulations. Staged solution was used to reduce the size and difficulty of solving the problem. The proposed value function approximation strategy of pondage and coal inventory solved the "curse of dimensionality" in staged recursion and maintained the overall optimization characteristics after staged decomposition. Through the iterative update between the decision and the approximation function, the approximate optimal decision sequence was solved to obtain the optimization scheme of power generation scheduling. The calculation and analysis results in a real system in some province showed that the modeling by approximate dynamic programming is an effective way to solve such large-scale optimization and has a promising future by virtue of its conciseness, effectiveness, convenience in dealing with random factors, high calculation accuracy and rapidity of solution.