Once the large-scale equipment breaks down, it will be stopped for maintenance, which will cause economic losses and serious safety accidents. Therefore, it is necessary to monitor the running status and predict the faults of machine tools. To solve this problem, this paper summarizes a multi-sensor information fusion analysis method of machine tool operation based on empirical mode decomposition algorithm (EMD) and improved BP neural network of dingo algorithm (DOA). Firstly, the processing data of eight typical operating states are collected and analyzed, and the original signal is decomposed and reconstructed by empirical mode decomposition algorithm to filter out clutter, and features are extracted to form feature vectors. Finally, the feature sample set is input into BP neural network improved by Australian dingo algorithm for training and the network model is saved. Experiments show that compared with the traditional BP neural network, the classification accuracy of DOA optimized BP neural network is improved, reaching 87.64%; The classification accuracy of BP neural network based on EMD and DOA optimization can reach 94.72%.