Federal learning-based intelligent intrusion detection method and system

The invention provides an intelligent intrusion detection method and system based on federal learning, and the method comprises the steps: step 1, initializing a long short-term memory (LSTM) network model, and carrying out the deployment on all user servers; step 2, enabling each user to use a local command sequence to train a single model of the user, and uploading model parameters to a central server; step 3, enabling the central server to execute model parameter aggregation to form a new global model and distributing the new global model to the user server, executing the step 2 and the step 3 circularly in sequence, and stopping training of the detection model until a set training round n is reached; and step 4, storing the detection model of the nth round, inputting the command sequence into the model to obtain a classification result, and realizing intrusion detection. According to the method, the plurality of sub-servers are coordinated through the central server, the user data sets are unified to esta