针对多端柔性直流输电线路保护的耐受过渡电阻能力差、识别准确率低的问题，提出了一种改进麻雀搜索算法（improved sparrow search algorithm，ISSA）与双向长短时记忆网络(bidirectional long short-term memory network, BiLSTM)结合的诊断模型。基于小波变换技术提取输电线路故障的特征作为模型输入量对模型进行训练；利用Sine混沌映射、学习粒子群算法策略、引入高斯扰动项对原始麻雀搜索算法进行改进，利用ISSA对BiLSTM目标超参数进行寻优，使故障诊断精度达到最优。最后基于PSCAD/EMTDC仿真平台搭建了四端柔性直流输电系统模型，验证表明，其故障识别准确率高、耐过渡电阻能力强，满足可靠性与速动性的要求。
英文摘要:
Aiming at the problems of poor ability to withstand transition resistance and low recognition accuracy of multi-terminal flexible DC transmission line protection, a diagnostic model combining an improved sparrow search algorithm (ISSA) and bidirectional long short-term memory network (BiLSTM) is proposed. Based on wavelet transform technology, the characteristics of transmission line faults are extracted as model input to train the model; the original sparrow search algorithm is improved by using Sine chaotic mapping, learning particle swarm algorithm strategy, and introducing Gaussian disturbance term. And ISSA is used to optimize the target hyperparameters of BiLSTM, so that the fault diagnosis accuracy is optimal. Finally, based on the PSCAD/EMTDC simulation platform, a four-terminal flexible DC transmission system model is built. The verification shows that it has high fault identification accuracy, strong resistance to transition resistance, and meets the requirements of reliability and rapidity.