Research On Fault Detection Of Double Tank Liquid Level System Based On Filter

With the continuous development and progress of modern science and technology,the industrial process is continuously intelligent,and the control system is becoming increasingly complex and sensor drift fault occurs in the operation process.In this thesis,the double tank liquid level system is taken as the research object,and the fault detection method is studied for the liquid level sensor drift fault of the upper tank of the system.The core technology of the model based fault detection strategy is to generate residual and calculate the threshold and residual evaluation function.The residual evaluation function is compared with the threshold to determine the occurrence of system fault.The main research contents of this thesis are as follows:Firstly,for the double tank liquid level system,considering the nonlinearity and uncertainty of the system and the drift fault of the upper water level sensor,the linear parameter variation system with double-layer polytope structure is introduced to model it.The first layer of the polytope model is used to approximate the nonlinearity of the system,and the second layer is used to describe the uncertainty of the flow valve.Secondly,a fault detection method based on parameter dependent full order filter is proposed.The introduction ofH_∞performance index is introduced to improve the robustness of the residual to disturbance and indirectly increase the sensitivity of residual to fault.The linear matrix inequality design condition of filter is given by using slack matrix and projection theorem.The design algorithm converts the solving problem of the filter gain matrix into the optimization problem.The simulation of the double tank liquid level system verifies the effectiveness of the design method.Finally,in order to improve the fault detection performance of the system,a parameter dependent non-minimal order fault detection filter is designed based on the full order filter.The auxiliary system is constructed to make the order of the state variables of the auxiliary system consistent with that of the non-minimal order filter,and the auxiliary matrix is constructed to solve the coefficient matrix of the auxiliary system.Compared with full order filter,non-minimal order filter uses more past output information of double tank liquid level system to improve fault detection performance.The effectiveness and superiority of the design method are verified by the simulation of double tank liquid level system.