基于改进GA-BP和CNN的园区不可调控负荷预测研究

Study on Park Unregulated Load Forecasting Based on Improved GA-BP and CNN

在智能园区的虚拟电厂建设中,常规不可调控负荷的预测准确度直接影响虚拟电厂对园区内可调设备的调度。针对园区负荷中常规不可调控负荷的预测,通过改进遗传算法GA的编码方式和交叉、变异概率,提高了BP网络的全局搜索能力。使用改进的GA-BP与CNN、BP对同一历史负荷数据进行预测对比,以均方根误差、平均绝对误差作为预测精度评判指标。实验结果表明了改进GA-BP在负荷预测中相较于CNN与传统BP具有更高的预测准确度,同时验证了改进的GA-BP在常规不可调负荷预测中引入环境因素对预测准确度具有负面扰动作用。

In the construction of virtual power plant in smart park,the accuracy of predicting the general unregulable load directly affects the dispatch of adjustable equipment in the park.Aiming at the prediction of conventional uncontrollable load in park load,the global search capability of BP network was improved by improving the coding method and crossover and mutation probability of genetic algorithm GA.The improved GA-BP was used to predict and compare the same historical load data with CNN and BP,and the root mean square error and average absolute error were used as the prediction accuracy evaluation indicators.Experimental results show that the improved GA-BP has higher prediction accuracy than CNN and traditional BP in load forecasting.At the same time,it is verified that the improved GA-BP introduces environmental factors into the conventional non-adjustable load forecasting and has a negative perturbation on the prediction accuracy.