The invention provides a storage battery feed risk identification method based on a machine learning algorithm, and the method comprises the steps: carrying out the data exploration and mining and feature variable construction through the related network signal data of a storage battery collected and uploaded by a cloud end under the condition of not adding hardware, carrying out the training through the machine learning algorithm, obtaining a more complete feed risk prediction model, using the trained model, and when the cloud monitors that real-time related data of the storage battery is uploaded, outputting the feed risk prediction result of the storage battery, and timely carrying out feed risk early warning reminding on a vehicle owner. As the storage battery fault sample data is continuously accumulated subsequently, the characteristic variables and the algorithm parameters can be continuously optimized, and the accuracy of the prediction model is gradually improved.