Research On Optimal Control Systemof NdFeB Hydrogen Decrepitation Based On Data-driven

Sintered NdFeB magnet production process that is a complex systems of industrialrequires more processes. Among them, usually hydrogen decrepitation process is veryimportant, it’s a effective means for breaking NdFeB ingot, because of after hydrogenpulverized NdFeB, crushing degree and hydrogen content of alloy have a direct impact on theperformance of NdFeB magnets. Due to the complexityofthe system’s accurately model that isunable established can not be ideal controled by the traditional control methods. however,model-free adaptive control is a control method that does not depend on the mathematicalmodel of the controlled system, only uses system input and output data designing controller, andhas many capabilities as simple algorithm, less adjustable parameter, good control performance.This paper is based on the depth analysis of NdFeB hydrogen absorption process and principles,that design control system of NdFeB hydrogen pulverization whit model-free adaptive controlmethod, and verified the feasibility for this method by experiment. Main works for this issue areas follows:There are the background, basic principles, control law, characteristic parameteridentification methods and performance of the model-free adaptive control method werestudied. the model-free adaptive control was compared whit conventional control method ofPID by simulation in terms of traceability, delay adaptability and adaptability, which obtainedmany comparison of the data curves that is valued. Meanwhile, in connection withshortcoming that parameters of the model-free adaptive control should be manually adjusted,model free adaptive control is based on a genetic algorithm is proposed, and a simulationexperiment was done with three-tank liquid position control as the background that show thatGA-MFAC controller have many properties, such as converges faster, shorter response timeand smaller overshoot that compared with the model-free adaptive control method.On this basis, a hydrogen forecasting model was established that is based on SOM andRBF hybrid neural network, it can predictive NdFeB hydrogen decrepitation process and thehydrogen content of the alloy, thus provide a model-free adaptive control a base for control. There is a system was builded and designed for NdFeB hydrogen crushing process whitgenetic optimization algorithm model-free adaptive control, we have completed hardware andsoftware of the control system, and the system test runs well. This study have provided a newapproach for the hydrogen pulverization process control.