基于神经网络的电力负荷预测方法研究

电力负荷预测易受到高频、低频和超低频振荡干扰,导致预测准确性不高,提出基于神经网络的电力负荷预测方法。在无线ZigBee组网协议下进行电力负荷传感器信息组网,构建电网负荷数据采集模型并进行模型修正。根据电力负荷数据采集结果,去除高频、低频和超低频振荡干扰因子。进行神经网络样本数据训练,去除冗余数据,输出电网负荷数据集合。对获得的数据集采用神经网络分类器进行分类融合处理,根据电力负荷数据的融合结果实现电力负荷预测。仿真结果表明,采用该方法进行电力负荷预测的准确性较高,预测过程的抗干扰性较好,在电力负荷的实时监测和信息调度中具有很好的应用价值。

Research on power load forecasting method based on neural network

Power load prediction is easily disturbed by high frequency,low frequency and ultra-low frequency oscillation,which leads to low accuracy of prediction.Sensor information networking of power load is carried out under the wireless ZigBee networking protocol,and the data acquisition model of power grid load is constructed and modified.According to the results of power load data collection,high frequency,low frequency and ultra-low frequency oscillation interference factors are removed.Neural network sample data training is conducted to remove redundant data,and network load data set is output.The obtained data sets are classified and fused by neural network classifier and the power load prediction is realized according to the fusion results of power load data.The simulation results show that the method has high accuracy and good anti-interference ability in power load forecasting,and has good application value in real-time monitoring and information dispatching of power load.