| Frequency control system is widely used in aviation, aerospace, industrialproduction and other fields, and as people drive the improvement of performancerequirements, improving the accuracy and reliability of the inverter is increasinglybecoming a hot issue. The motor’s stator and rotor parameters are required in thevector control system. In order to obtain these parameters, parameter identificationis needed before running. In some special occasions, the control system suddenlystops due to a fault, which can cause huge economic losses, and even casualties. Sothe inverter is required with fault diagnosis function and tolerant operation function.For off-line parameter identification problem, this paper proposes an approachbased on linear neural network, and the off-line identifications of induction motor’sstator and rotor parameters are realized based on single-phase AC experiment.Firstly, according to the mathematical model of the induction motor, therelationships between linear neural network parameters and the motor parametersare derived. Then induction motor off-line parameter identification model is builtthrough Matlab simulation platform, and a7.5kW induction motor is identificatedwith this method. Finally, the simulation results are compared with the true value,which verify the effectiveness of the proposed method. This proposed method hasnot only high identification accuracy but also fast identification speed.In order to enhance the inverter’s reliability, the current sensor fault diagnosisand fault-tolerant control methods are researched in detail. In this paper, single ordouble current sensor failure diagnostic methods and double current sensors faulttolerant control method-single current sensor vector control method are proposed.,When single or double current sensors are at fault, with this method, the controlsystem can continue to run smoothly.For the above researches, the hardware interface circuit boards which is used tosimulate inverter fault, conduct the signal and so on, and1.1kW induction motorexperiment platform are designed. Experimental results show that the proposedmethod can achieve the current sensor fault diagnosis and fault tolerant control. |