Optimization of wind and solar energy storage system capacity configuration based on the Parzen window estimation method

Abstract

Compressed air energy storage (CAES) effectively reduces wind and solar power curtailment due to randomness. However, inaccurate daily data and improper storage capacity configuration impact CAES development. This study uses the Parzen window estimation method to extract features from historical data, obtaining distributions of typical weekly wind power, solar power, and load. These distributions are compared to Weibull and Beta distributions. The wind–solar energy storage system's capacity configuration is optimized using a genetic algorithm to maximize profit. Different methods are compared in island/grid-connected modes using evaluation metrics to verify the accuracy of the Parzen window estimation method. The results show that it surpasses parameter estimation for real-time series-based configuration. Under grid-connected mode, rated power configurations are 1107 MW for wind, 346 MW for solar, and 290 MW for CAES. The CAES system has a rated capacity of 2320 MW·h, meeting average hourly power demand of 699.26 MW. It saves $6.55 million per week in electricity costs, with a maximum weekly profit of $0.61 million. Payback period for system investment is 5.6 years, excluding penalty costs.