An integrated neural network method for market clearing price prediction and confidence interval estimation

Market energy clearing prices (MCPs) play an important role in a deregulated power market, and good MCP prediction and interval estimation will help utilities and independent power producers submit effective bids with low risks. MCP is a non-stationary process, and an adaptive algorithm with fast convergence is important. A common method for MCP prediction is neural networks, and the extended Kalman filter (EKF) can be used as an integrated adaptive learning and interval estimation method, with fast convergence and small confidence interval. However, the EKF learning is computationally expensive because it involves high dimensional matrices. This paper presents a modified U-D factorization method within the framework of decoupled EKF. The computation speed is significantly improved and also is the numerical stability. EKF learning can then be used for high dimensional practical problems. Testing results show that the integrated learning and confidence interval algorithm provides better prediction than the back propagation algorithm and the confidence interval is smaller than that of a Bayesian inference-based interval estimation method.