能源多元化发展对电力系统发电调度提出新的要求：在以往重点研究的传统调度模型与算法的基础上，赋予发电调度“智慧”，以解决多元化电源接入后系统的复杂性、随机性和动态耦合性造成的传统数学建模与优化方法受限问题。本项目以提高发电调度决策效率和调度目标实现率为着眼点，以调度智慧的挖掘、描述、应用与防误为主线，研究适应能源多元化发展的知识型发电调度决策方法。项目的主要研究内容和关键技术包括：解析多元化电源的随机动态耦合机理，提出知识型调度决策的驱动机制；研究发电调度知识网络的构建及追踪辨识方法，实现调度决策规律的知识化表征及自适应调用；将提炼出的深层调度知识与传统发电调度方法相结合，对含多元化电源的大规模复杂性调度问题进行合理简化与高效决策；基于样本熵提出异常发电调度决策的甄别方法。本课题的研究成果为传统发电调度决策向知识型调度决策发展奠定系统的理论基础，具有重要的研究意义和广阔的应用前景。

Because of the diversified development of power sources, new requirement is raised for power system generation scheduling. In the early years, power system generation scheduling mainly focused on traditional optimized models and algorithms. However, the intelligence in power system generation scheduling must be studied in order to solve the limitations of mathematical modeling and optimization caused by the complexity, randomness, and dynamic coupling of diversified power sources. For the purpose of improving the decision-making efficiency and goal-achievability rate of power system generation scheduling, a new knowledge-based decision-making system under new condition is constructed by means of the supporting of this project. The main contents and key techniques related to the mining, description, application, and error prevention of scheduling intelligence are as follows: the random dynamic coupling characteristics of diversified power sources, and the driving mechanism of knowledge-based power system generation scheduling; the knowledge networks describing the scheduling intelligence, and the tracking and adaptive applications of knowledge networks; the decoupling of the large-scale power system generation scheduling problem based on knowledge networks, and the efficient decision-making of each sub-problem based on traditional power system generation scheduling method; and the prevention of abnormal decision-making based on the sample entropy. The achievement of the theoretical studies of this project will establish systematic foundation for the development of the traditional power system generation scheduling to the knowledge-based one, and it has important research significance and potential application prospects.