Fault prediction algorithm for fire control system based on improved support vector machine

The structure of the tank fire control system is complex, the fault information acquisition is difficult, and the fault features are more, the maintenance cost is high, and the fault prediction and health management problems need to be solved urgently. The machine learning algorithm of support vector classifier is used to predict the fault of the fire control computer and sensor subsystem. In order to better carry out the fire control system health management, the fault prediction of the fire control system not only stays in the identification of the "normal"and "fault"states, but also distinguishes different types of fault states. The least squares support vector multiclassifier based on decision directed acyclic graph is selected for prediction. The improved separation measure is introduced to improve the decision directed acyclic graph, which reduces the error caused by improper initial sequence. The particle swarm optimization algorithm is used to optimize the parameters of the least squares support vector classifier, which improves the classification accuracy. The experimental test of the tank fire control computer proves that the proposed method has high reliability and effectiveness.