Research On Fault Tolerant Control Strategy Of Double Converter Based On Model Predictive Control

Open winding motor is commonly used in aerospace,electric vehicles and other business and this is mainly due to its high voltage output level and various output number.Compared to conventional single-inverters,dual-inverter topology can offer lower power density and better fault-tolerant performance.This thesis extrapolates predicted expected to the moment at k+2 to solve the calculation delay.Revise the discrete prediction model to solve the model parameter error of the open-winding motor model,based on Heun’s prediction correction scheme.Aiming at switch deviceS_{x 1}andS_{x 4}open circuit,this thesis takesS_{1 1}andS₁₄ open circuit as an example.The zero common-mode voltage vector is chosen offline to suppress zero-sequence voltage and common-mode voltage,and additional control term of the potential midpoint balance is used in cost function.To further shorten the calculation delay,this thesis proposes two algorithms.（1）Quasi-deadbeat model predictive control.This algorithm introduces the idea of deadbeat in single-step predictive control,and use a simplified SVPWM scheme to partition the remaining vectors.（2）Improved boundary circle control strategy.The switch state combination does not need to be changed at any time,and a boundary circle could be drawn within the allowable range of the current error.Aiming at switch deviceS_{x 2}orS_{x 3}open circuit,this thesis takeS₁₂ open circuit as an example.When the number of remaining zero common-mode voltage vectors is small,the remaining conventional vectors are selected for optimization,in this case there are more switch state combinations to choose from.Based on this,this thesis proposes satisfactory optimized predictive control strategy,which is a hierarchical design of voltage jump,current follow,and midpoint potential balance.To prevent damage to the non-fault switch tube caused by the voltage change,the hard constraint index is the voltage control;in order to achieve better current following the effect,the main control target is the current follow-up;the final control target is the midpoint potential balance.To prevent the existence of a solution for optimization,one solution is to establish a new satisfaction judgment function and the other is to introduce a fuzzy decision idea to weaken the boundary between satisfaction and runaway.