Fault reconstruction method based on PD (Proportional Differential) type learning observer

Abstract

The invention discloses a fault reconstruction method based on a PD (Proportional Differential) type learning observer. The method comprises the following steps that: taking an unmanned aerial vehicleas a rigid body, ignoring the inertia product of the unmanned aerial vehicle, and selecting a corresponding state variable to construct the state-space equation of the longitudinal flight control system of the unmanned aerial vehicle; on the basis of the constructed state-space equation, constructing a PD-type iterative learning observer; defining a state estimation error, an output estimation error and a fault reconstruction error, and according to the constructed state-space equation and the constructed PD-type iterative learning observer, obtaining a corresponding error equation; setting the infinite norm of an executor fault matrix to be bounded, utilizing the PD-type iterative learning observer to reconstruct executor additive faults, and according to the error equation, enabling thestate estimation error and the fault reconstruction error to be stabilized in a preset interval, wherein the output estimation error meets a preset standard. By use of the method, a purpose of constructing the PD-type iterative learning observer to realize accurate fault reconstruction is realized, and the accuracy of accurate reconstruction is improved.

Description

technical field [0001] The invention relates to the technical field of fault reconstruction, in particular to a fault reconstruction method based on a PD type learning observer. Background technique [0002] As an effective method to improve system security and reliability, Fault Diagnosis and Fault Tolerant Control (FTC) technology has attracted widespread attention from experts and scholars at home and abroad. Fault diagnosis mainly includes: fault detection, fault separation, and fault reconstruction. Its main tasks are: judging whether a fault occurs, locating the location and type of the fault, and determining the size and change trend of the fault. The model-based fault detection and isolation (FDI) method mainly detects and isolates system faults by designing a residual generator and comparing the residual with a threshold. However, for early micro-faults and slowly-varying faults, the above method is prone to fault under-reporting. Contents of the invention [00...

AI Technical Summary

Problems solved by technology

However, for early micro-faults and slow-changing faults, the above method is prone to fault under-reporting.

Images