Calibration Of Dynamic Deformation Reconfigurable System For Wing Conformal Antenna And Sensor Fault Detection

With the development of health monitoring and intelligent structures,the technology of structural deformation sensing based on strain has gained more and more attention.The high-precision deformation and reconstruction of wing structure plays an important role in ensuring the safe operation of aircraft.In this paper,the calibration of the wing deformation reconstruction system and the sensor fault monitoring are studied and verified by physical experiments.By using the idea of error allocation and the theory of fuzzy identification,the accuracy of wing model deformation reconstruction scheme is improved.However,the performance of structural deformation reconstruction system based on strain depends on the reliability of sensor measurement.To solve this problem,two fault detection algorithms of fiber Bragg grating strain sensor are proposed,and the type and size of the fault are identified.The self-developed wing deformation measurement software realizes the encapsulation of the above theory,and preliminarily verifies the scheme.The main contents are as follows:In view of the errors in the reconstruction results based on the inverse finite element method,a two-step calibration method is proposed.Firstly,the hypothesis test with error allocation is applied to modify the node information of deformed structures.Then,the self-structuring fuzzy network with particle swarm optimization(SSFN-PSO)is applied to automatically approximate the relationship between the strain and the modification of node information.The reliability and robustness of the calibration scheme are verified by simulation analysis and wing deformation experiment.In the measurement process of FBG sensors,the assumption that the same probability distribution will continue to represent the observed variable cannot be met for non-stationary processes.Therefore,a new method is proposed to divide the strain time series into two periods(nighttime and daytime),and to normalize the time series of daytime.Through heuristic segmentation of non-stationary time series,the statistical control of the trend output data of the strain sensor is realized.Local verification experiments of non-stationary sensor data on the wing model show that the method can successfully detect abnormal sensors.The heuristic segmentation algorithm does not reveal the fault types of sensors,and a method of sensor fault detection and classification based on generalized likelihood ratio and correlation coefficient is proposed.The minimum mean square error(MMSE)algorithm was used to evaluate each sensor in the sensor network,and the fault of the sensor was detected by using multiple hypothesis test based on generalized likelihood ratio.Five common types of sensor faults are studied,and the correlation between estimated and measured values is extracted as a classification feature.The types of sensor faults are classified by the unbalanced binary tree method.Fault simulation experiments show that the method can quickly reveal the location and type of fault sensors.Finally,the software platform of deformation monitoring system is designed by analyzing the software requirements.Based on the deformation reconstruction calibration algorithm and sensor fault detection algorithm,the reconstruction deformation displacements of the wing model and the monitoring of sensor faults are displayed on the demonstration interface.The main functions of the software system include three parts: wing deformation reconstruction,sensor fault monitoring,and data management.