论文部分内容阅读
由于电力拍卖市场具有明显的寡头市场特征,且由于发电领域依然存在规模经济性,电力拍卖市场的定价规则出现了多种选择,如边际成本定价、按报价支付定价和当量电价定价法。文中构造了一个基于智能代理的仿真模型讨论3种定价方法的市场运行特征。该模型用具有分散自主决策和智能学习的代理表示发电厂商,在重复进行的报价博弈中,发电厂商通过利润高低按照生物的条件反射原理学习并改进报价策略,追求其利润最大化,随着迭代次数的增加,市场将逐步收敛于均衡位置。然后,通过不同定价方式的拍卖市场均衡状态的比较给出了3种定价方式的市场运行特点。给出了智能代理仿真的模型及其发电厂商的学习算法。最后用一个算例阐述了学习算法的特点。
Because of the obvious oligopolistic market characteristics in the electricity auction market and due to the economies of scale still existing in the electricity generation field, the pricing rules of the electricity auction market have many choices, such as marginal cost pricing, price quotation pricing and the equivalent price pricing method. In this paper, a simulation model based on intelligent agents is constructed to discuss the market operation characteristics of the three pricing methods. In this model, power generation companies are represented by agents with decentralized autonomous decision and intelligent learning. In the repeated bidding game, power companies learn and improve their bidding strategies according to the principle of conditional learning by profit, and pursue profit maximization. With the iteration The increase in the number of markets will gradually converge to a balanced position. Then, by comparing the equilibrium state of the auction market with different pricing methods, the market operating characteristics of the three pricing methods are given. The model of intelligent agent simulation and its learning algorithm of power plant are given. Finally, an example is given to illustrate the characteristics of the learning algorithm.