A data-driven method for aircraft static strength test anomaly recognition

The strain data from the static strength test of the aircraft is the correct response to the structure bearing deformation. Real-time monitoring and fast analysis of strain data is an important factor. In this paper, we propose a data-driven approach for aircraft static intensity test anomaly identification in order to address the problems of large aircraft strain gauge numbers, weak data interference resistance, and low manual detection efficiency. First, the regression equations are constructed based on the five points of the test data, and the predicted values of the remaining test data are obtained by the regression equations. The residual analysis is then performed with the actual values measured by the strain gauge and the residuals are calculated using the standard method. Finally, machine learning models are used to discriminate between binary anomalies. By comparing the classification effect of SVM and random forest algorithm, it is found that both algorithms achieve high accuracy, reaching 92% anomaly recognition accuracy. This approach has great engineering applications in real-time monitoring and early warning of static intensity tests of aircraft.