论文部分内容阅读
电力负荷预测作为电力企业调度工作的重要组成部分,其预测的好坏将直接决定电力供电服务的质量。但传统的神经网络属于静态网络,而电力负荷属于时间序列,随着时间的变化而不断更新。对此针对上述的问题,提出一种基于动态的Elman神经网络对短期内的电力负荷进行预测。同时,针对神经网络算法容易陷入局部最优解不准确的缺陷,引入遗传算法对最优网络初始值进行求解,进而提高电力负荷预测的准确度。最后,通过以内蒙古赤峰市2016年10月2日至2016年10月21日的用电作为基础数据,以前10天的数据作为训练数据,以后10天作为预测数据,并与实际的数据进行对比,进而验证了本文预测算法的准确性。
Power load forecasting is an important part of the dispatch work of power enterprises. The quality of power supply service will be directly determined by the forecast of power load. However, the traditional neural network belongs to the static network, while the power load belongs to the time series, which is continuously updated with the change of time. In response to the above problems, a dynamic Elman neural network is proposed to predict the short-term power load. At the same time, aiming at the defect that the neural network algorithm is easy to fall into the inaccurate local optimal solution, the genetic algorithm is introduced to solve the initial value of the optimal network, so as to improve the accuracy of power load forecasting. Finally, based on electricity consumption in Chifeng City, Inner Mongolia from October 2, 2016 to October 21, 2016, the data of the previous 10 days are used as training data and the next 10 days are used as the forecast data and compared with the actual data , And then verify the accuracy of the prediction algorithm in this paper.