卷积神经网络在舵机异常检测中的应用

Application of Convolutional Neural Networks in Anomaly Detection of Steering Gear

针对舵机测试数据样本不均衡,小样本难以被准确分类的问题,提出了一种基于卷积神经网络(CNN)的舵机异常检测模型。该模型的优异性适用于对舵机测试数据进行特征提取和多故障分类,从而使得舵机的异常检测从单纯人工走向自动化和智能化。实验结果表明,该方法的准确度可达99.774%,证实了卷积神经网络应用于舵机异常检测的可行性。

In this paper, an anomaly detection model of steering gear based on convolutional neural networks(CNN) is proposed. The model is mainly aimed at the test data samples of the unbalance characteristics of the steering gear to improve the present ability of inaccurate classification of small samples. In the experiment, the model is used to extract the features of the test data of the steering gear and classify multiple faults, so that the abnormal detection of the steering gear can be changed from manual to automatic. The experimental results show that the accuracy of the proposed method can be reached by 99.774%, which confirms the feasibility of the convolutional neural networks applied to the anomaly detection of steering gear.