A method for evaluating faults of unmanned aerial vehicles based on chain neural networks

Abstract

The application discloses a kind of unmanned plane fault evaluation method and system based on chain neural network, method includes: constructing variable database;By pearson correlation coefficient method, the correlation coefficient of variable combination is obtained, and variable combination below correlation coefficient threshold is eliminated;Chain neural network model is constructed, and variable predictor model n is the prediction model for predicting variable n with the variable that exists correlation in correlation matrix;Collect the relevant data of unmanned plane fault-free time and data factor item input chain neural network model in prediction learning training;Real-time acquisition unmanned plane and data factor item relevant data input variable predictor model set, and the number of alarms is counted as unmanned plane fault evaluation result.The application realizes the probability evaluation of fault occurrence in unmanned plane flight, is convenient for timely emergency command unmanned plane to stop flying and landing, and carries out fault inspection in time, prevents unmanned plane fault from further expanding, and improves unmanned plane flight safety.

Images