As a novel recurrent neural network (RNN), an echo state network (ESN) that utilizes a reservoir with many randomly connected internal units and only trains the readout, avoids increased complexity of training procedures faced by traditional RNN. The ESN can cope with complex nonlinear systems because of its dynamical properties and has been applied in hydrological forecasting and load forecasting. Due to the linear regression algorithm usually adopted by generic ESN to train the output weights, an ill-conditioned solution might occur, degrading the generalization ability of the ESN. In this study, the ESN with Bayesian regularization (BESN) is proposed for short-term power production forecasting of small hydropower (SHP) plants. According to the Bayesian theory, the weights distribution in space is considered and the optimal output weights are obtained by maximizing the posterior probabilistic distribution. The evidence procedure is employed to gain optimal hyperparameters for the BESN model. The recorded data obtained from the SHP plants in two different counties, located in Yunnan Province, China, are utilized to validate the proposed model. For comparison, the feed-forward neural networks with Levenberg-Marquardt algorithm (LM-FNN) and the generic ESN are also employed. The results indicate that BESN outperforms both LM-FNN and ESN.
China is implementing a new power system reform, with one goal of renewable energy absorption such as hydropower. However, the forthcoming spot market challenges cascade hydropower generation in terms of the short-term hydro scheduling (STHS) problem. Specifically, STHS involves fulfilling bilateral market obligations and bidding for the day-ahead market with uncertainty. Coordination of these two tasks while managing market risks becomes a problem that must be urgently solved. Herein, we propose a method based on the information-gap decision theory (IGDT) to solve the cascade hydropower STHS problem, wherein the aforementioned tasks are coordinated simultaneously. The IGDT method was used to deal with the uncertainty of the day-ahead market price, and the robustness function was derived. A mixed-integer nonlinear programming model was used to describe the proposed problem, and a commercial solver was used to solve it. A four-reservoir cascade hydropower company was used as the research object. Through the robust dispatching results, the preset profit objectives of the power generation company were satisfied within the price information gap, and the day-ahead market bidding strategy and daily contract decomposition curve were obtained. The proposed model is found to be superior to the scenario-based probability method. Moreover, a comparative analysis of bilateral contract fulfillment showed that more profits can be obtained by coordinating contract fulfillment in the day-ahead market.
To further promote market competition, enrich trading varieties, alleviate information asymmetry, and improve trading efficiency during electricity market reform in China, the continuous bidirectional transaction (CBT) was designed and applied in the Yunnan electricity market (YNEM), which is dominated by medium- and long-term power energy trading. The clearing model for the CBT with the goal of maximum social welfare is proposed in two bidding stages, including call auction (CA) and continuous double auction (CDA). Correspondingly, the integrated two-stage market clearing algorithm is also introduced to ensure the data consistency and business continuity. Finally, the analysis of the practical application shows that the proposed model, algorithm, and various key implementation strategies of the trading platform support the bidding and clearing of the CBT well. In addition, the research and application of CBT may also provide valuable insights for other electricity market construction.
China is currently engaged in a nationwide electricity market reform, aiming to reduce electricity price and improve clean energy consumption. In many areas, thousands of members are participating these special markets; they are all facing with bidding decision-making problems in these unanticipated markets. Due to the particularity of market rules and the shortness of data history and real-time info, plants’ bidding is lacking in theoretical support, their fate in the market is unknown. In this paper, an evolutionary game based electricity market-bidding model is proposed, in which all market members are bidding under limited market info. Each bidding strategy is an exploration of the market. The power plants take ‘market’ itself as their game opponent, instead of ‘other plants’ in normal game theory models. They adjust decisions according to market feedback and attempt to get some payoff from next market. Loss-making power plants will be acquired by other plants and inherit their bidding strategies. By simulating Chinese typical electricity market, the most competitive strategies are found and power generation groups are formed. Furthermore, this paper takes into account the possible impact of different type of demand and the supply-demand ratio. Comparing with constant demand, the demand curve can prevent clearing price from rising effectively. With a certain demand curve and the acquisition mechanism, the market systems achieve stability, in which the clearing point fluctuate little and periodically, the final clearing price is much lower than that before the reform, and the reform goal was achieved.
The newly-reformed Yunnan electricity market has been operating successfully for three years with more and more abundant trade varieties and perfect trading mechanism. The vitality of the market is increasing year by year. Market participants are gradually exploring reasonable bidding strategies, and the accurate electricity price forecasting is a joint demand for all the market members. With comprehensive considerations of the social economic background, the structure of the power system, the market development state, and other factors in Yunnan, this paper analyzes various factors and rules that have effects on the electricity price in day-ahead market. Furthermore, a successful application of an auto regressive moving average (ARMA) model to electricity price forecasting (EPF) is presented. The case study based on real data of day-ahead market in Yunnan shows that the proposed model could reflect the characteristics of electricity price and the results prove this model’s effectivity.