基于神经网络的电容器介质损耗角检测方法

Neural Network-based Method for Detecting the Dielectric Loss Angle of Capacitors

在电力系统中,准确计算电容器的介质损耗角对于评估其性能和可靠性至关重要。传统的正向求解方法在实际应用中存在不确定性,本文提出了一种基于神经网络的介质损耗角识别新方法。通过采集电压、电流和介质损耗角数据,训练神经网络模型进行预测。文章详细阐述了介质损耗角的求解过程,包括幅值求解方法,并在深度神经网络模型中纳入电容器扰动因素。仿真测试基于实测数据,结果表明,即使在环境干扰下,所提方法的平均辨识错误率低至2.74%,显示出优越的抗噪性能和精度稳定性。与传统加汉宁窗口谐波分析法相比,本方法在介质损耗角识别方面具有明显优势。为电容器介质损耗角的准确识别提供了一种新的有效途径。

Accurately calculating the dielectric loss angle of capacitors is crucial for evaluating their performance and reliability in the power system.The traditional forward solving method has uncertainty in practical applications.This paper proposes a new method for identifying dielectric loss angles based on neural networks.Train a neural network model for prediction by collecting voltage,current,and dielectric loss angle data.The article elaborates on the process of solving the dielectric loss angle,including the amplitude calculation method,and incorporates capacitor disturbance factors into the deep neural network model.The simulation test is based on measured data,and the results show that even under environmental interference,the average identification error rate of the proposed method is as low as 2.74%,demonstrating superior noise resistance and accuracy stability.Compared with the traditional Hanning window harmonic analysis method,this method has significant advantages in identifying the angle of dielectric loss.This provides a new and effective way to accurately identify the dielectric loss angle of capacitors.