基于BP神经网络的高压电能计量装置误差校正研究

Error Correction of High Voltage Power Metering Device Based on BP Neural Network

由于传统方法忽略了对高压电能计量装置比例系数、积分系数及微分系数关系的调整,导致误差校正耗时长、校正效果差,为解决该问题,提出了基于BP神经网络的高压电能计量装置误差校正方法。采集高压电能计量装置运行数据,根据采集数据对高压电能计量装置比例系数、积分系数及微分系数进行调整,使三者可以相互匹配且能相互约束,在相互约束条件下,从非线性组合中寻找高压电能计量装置误差校正的最佳权值。采用递阶遗传优化方式进行BP神经网络结构优化,设计权值学习过程,利用优化后的BP神经网络对权值不断调整,实现高压电能计量装置误差校正。仿真实验表明,研究方法的误差校正效果好、耗时短,具有可靠性。

Because the traditional method ignores the adjustment of the relationship among the proportional coefficient,integral coefficient and differential coefficient of the high-voltage electric energy metering device, the error correction is time-consuming and the correction effect is poor.In order to solve this problem,this paper presents an error correction method of high-voltage electric energy metering device based on BP neural network. The operation data of the high-voltage electric energy metering device is collected,and the proportional coefficient,integral coefficient and differential coefficient of the device are adjusted according to the collected data,so that the three can match and constrain each other.Under the condition of mutual constraint,the optimal weight of error correction of the high-voltage electric energy metering device is sought from the nonlinear combination. The hierarchical genetic optimization method is adopted to optimize the structure of the BP neural network,and the weight learning process is designed. The optimized BP neural network is used to continuously adjust the weight to realize error correction of the high-voltage electric energy metering device. Simulation results show that the proposed method has good error correction effect,takes short time for correction time and is reliable.