煤矿工人心律失常检测的LSTM-CNN T波识别方法

LSTM-CNN T wave recognition method for arrhythmia in mine workers

为更好地保护煤矿工人的身体健康,提出一种长短期记忆神经网络和卷积神经网络组合的T波识别算法。采用卡尔曼滤波器预处理数据,提取特征,利用鲸鱼优化算法求得神经网络参数的最优解,以464名某煤矿公司井下一线作业工人的心电数据作为样本,训练集样本为18000个,测试样本为7700个,验证模型的有效性与准确率。结果表明,与现有方法相比,所提方法能够有效识别出T波异常变化形态,识别准确率达到99.35%。该算法可用于煤矿井下工人的心律失常检测。

This paper proposes a T wave recognition algorithm building on the combination of long and short-term memory neural network and convolutional neural network,which better protect the health of mine workers.This is obtained by using Kalman filter for data preprocessing and feature extraction;using the whale optimization algorithm to find the optimal solution of neural network parameters;and collecting the ECG data of 464 workers in a coal mine company as samples,including 18000 training samples and 7700 test samples to verify the effectiveness and accuracy of the model.The results indicate that the proposed method proves better than other existing methods thanks to an effective identification of the abnormal change of T wave,with the recognition accuracy rate up to 99.35%.The algorithm could work better for detecting the arrhythmia of workers in coal mines.