【摘要】 为了提高断路器故障预测结果的可靠性，提出一种基于电流-振动信号特征自适应融合的BWO-CNN-LSTM故障预测方法。首先采用变分模态分解处理振动信号，提取能量熵、样本熵和散布熵等特征；同时提取线圈电流信号的关键特征，通过特征自适应融合技术构建断路器完整动作过程的电流-振动联合信号特征集；随后构建白鲸算法优化的CNN-LSTM故障预测模型，对断路器的不同机械状态进行预测。与其他算法相比，BWO-CNN-LSTM模型在故障预测准确率上显著提高，有效解决了断路器故障预测中对人工经验过于依赖的问题。更多还原

【Abstract】 In order to improve the reliability of circuit breaker fault prediction results, a BWO-CNN-LSTM fault prediction method based on adaptive fusion of current-vibration signal features is proposed. First, the vibration signal is processed by variational modal decomposition to extract features such as energy entropy, sample entropy and scatter entropy; and the key features of the coil current signal are also extracted, and the feature set of the current-vibration joint signal for the complete action process of the circuit breaker is constructed by feature adaptive fusion technology; subsequently, the CNN-LSTM fault prediction model optimized by the beluga whale optimization is constructed to predict the different mechanical states of the circuit breaker. Results show that compared with other algorithms, the BWO-CNN-LSTM model significantly improves the fault prediction accuracy, effectively solving the problem of manual experience dependence in circuit breaker fault prediction.更多还原