一种基于改进负荷矩的台区电压快速估算方法

针对不同截面导线混接给负荷矩计算带来的困难,提出了一种负荷矩归一化方法,在保证线路压降相同的情况下,能够通过归一化系数统一线路型号,将复杂的配电网络简单化、标准化。在此基础上,建立了节点负荷矩与节点电压之间的映射关系,通过基于Levenberg-Marquardt算法的BP神经网络,实现了对台区节点电压的全覆盖估算,克服了传统梯度下降算法在收敛速度、计算稳定性上的缺陷,所需训练样本少,计算效率高。选取某实际台区网络进行了仿真分析,算例结果验证了方法的有效性。

A fast voltage estimation method based on improved load moment

In order to solve the difficulty of load moment calculation of different cross section wires, a method of load moment normalization is proposed. In the case of ensuring the same voltage drop of the conductor, the model of wire can be unified through the normalization coefficient, and the complex distribution network is simplified and standardized. On this basis, the mapping relationship between the node load moment and the node voltage is established. Through the BP neural network based on the Levenberg-Marquardt algorithm, the full coverage estimation of the node voltage is realized, and the shortcomings of the traditional gradient descent algorithm in the convergence speed and the calculation stability are overcome. The required training samples are few, and the calculation efficiency is high. A practical network is selected for simulation analysis. The effectiveness of the proposed method is verified.