In order to effectively identify the early faults of rotating machinery

a multi-channel input continuous convolutional neural network support vector machine（CNN-SVM） early fault diagnosis model is proposed.The original data collected from multiple sensors in different directions and locations are directly input into the improved CNN model for training.Then

the improved CNN is used to extract features from the new original fault data

and the extracted sparse representative feature vector is input into the support vector machine for fault classification.This method is applied to the diagnosis of multi-channel vibration signal monitoring data of gearbox.The accuracy of this method is 99.94 % compared with improved CNN and PCA-SVM.The results show that the multi-channel input CNN-SVM model has higher diagnostic precision and faster detection speed.