基于智能个体信念学习的电力市场模拟的理论框架

为了降低电力改革的风险,需要利用电力市场模拟的方法来研究电力市场的演化规律,对现有电力市场方案进行验证.基于多智能体结构和复杂适应系统的思路,作者引入信念学习的思想,提出了电力市场模拟总体设计架构.在该模拟思路中,每个市场成员基于报价中标概率信念拥有统计学习能力和决策能力,同时拥有探索能力和记忆能力,可以自主地从市场中学习知识和调整各自报价决策,从而实现模拟电力市场自主、动态演化的目标.新的市场成员个体学习和决策模型不但提高了新的电力市场模拟方法的计算效率,还能够在仅仅了解电力系统固有特性(如:电网结构、电源结构等)和市场规则等有限信息的情况下模拟和研究电力市场演化过程,检验市场设计方案中的不足和缺陷,对现阶段的我国电力市场改革具有重要的现实意义.

THEORETICAL FUNDAMENTAL AND CONCEPTS OF ELECTRICITY MARKET SIMULATION BASED ON AGENTS' BELIEF LEARNGING

To reduce risks in electricity market reform, simulation methods should be applied to the study of principles of electricity market evolution and to the evaluation of market arrangements. The authors design and propose theoretical principles and framework of an electricity market simulation method based on the ideas of Multi-Agent systems (MAS) and Complex Adaptive Systems (CAS). In this simulation method, based on its own belief of Bid Acceptance Probability (BAP), each market participant agent possesses the abilities of probabilistic learning and bid decision-making, as well as autonomous exploration and memory. Thus, each agent can simulate the dynamic behaviors of market participants in electricity market by actively learning knowledge from electricity market transactions and applying the knowledge to the modification of its own bidding strategies, so the objective to simulate the independent and dynamic evolution of electricity market can be realized. The new designed learning model and bid decision-making model of agents can not only improve the computational efficiency of the proposed simulation methods, simulate and study the evolution process of electricity market even though only limited information such as inherent characteristics of power system, e.g., power grid topology and power source structure, etc., and transaction arrangement of the market is understood, but also verify the insufficiency and defects in market design proposal, therefore it is especially valuable for the reform of electricity market at present stage in China.