| 基于稀疏贝叶斯学习的继电保护装置瞬变脉冲信号识别 |
 点此下载全文 |
| 引用本文：甘永平，李刚.基于稀疏贝叶斯学习的继电保护装置瞬变脉冲信号识别[J].计算技术与自动化,2026,(2):159-163 |
| 摘要点击次数: 38 |
| 全文下载次数: 0 |
|
|
| 中文摘要:为充分掌握瞬变脉冲信号的变化情况，精准衡量继电保护装置运行状态，提出了基于稀疏贝叶斯学习的继电保护装置瞬变脉冲信号识别方法。采集继电保护装置瞬变脉冲信号，提取其分布曲线峰度关键特征。引入期望最大化算法迭代寻优稀疏贝叶斯学习模型超参数，确定合理的稀疏度参数值，采用稀疏贝叶斯学习模型实现瞬变脉冲信号的有效识别。通过实验验证，该方法的稀疏度参数值为0.6时，能够有效识别不同类型单一故障以及多种故障共同产生的脉冲信号，识别准确率均在98.5%以上，且脉冲信号平均识别时间均低于4.46 ms，有效提升电力系统运行安全性与稳定性。 |
| 中文关键词:稀疏贝叶斯学习 继电保护装置 瞬变脉冲信号识别 峰度 期望最大化算法 超参数 |
| |
| Transient Pulse Signal Recognition of Relay Protection Devices Based on Sparse Bayesian Learning |
|
|
| Abstract:To fully grasp the changes in transient pulse signals and accurately measure the operating status of relay protection devices, a method for identifying transient pulse signals in relay protection devices based on sparse Bayesian learning is proposed. Collect transient pulse signals from relay protection devices and extract key features of their distribution curve kurtosis. Introduce the expectation maximization algorithm to iteratively optimize the hyperparameters of the sparse Bayesian learning model, determine reasonable sparsity parameter values, and use the sparse Bayesian learning model to effectively identify transient pulse signals. Through experimental verification, when the sparsity parameter value of this method is 0.6, it can effectively identify pulse signals generated by different types of single faults and multiple faults together. The recognition accuracy is above 98.5%, and the average recognition time of pulse signals is less than 4.46 ms, effectively improving the safety and stability of power system operation. |
| keywords:sparse Bayesian learning relay protection device transient pulse signal recognition kurtosis expectation maximization algorithm hyperparameter |
| 查看全文 查看/发表评论 下载pdf阅读器 |
位访问者,今日访问