Research On Microwave Heating Temperature Control Based On Event-Triggered Adaptive Dynamic Programming

Compared with traditional heating methods,the advantages of microwave heating are mainly reflected in the fast temperature rise rate and high energy conversion rate.Therefore,it is widely used in food sterilization and drying and physical processing of metals and ceramics.Due to the uneven distribution of the alternating electromagnetic field received by the medium in the microwave heating cavity,the dielectric properties of the medium change.It makes the ability of different parts of the medium to absorb microwaves different,which leads to local overheating phenomena such as hot spots and thermal runaway.This local overheating phenomenon hinders the further development of microwave heating technology to promote and apply.Real-time control of the temperature of the heating medium can effectively solve the local overheating problem in the microwave heating process.However,the microwave heating process involves the evolution and coupling of multi-physical fields such as electric field,magnetic field,and temperature field,and has the characteristics of time-varying,non-linear,and external disturbance.Therefore,it is difficult to establish an accurate,effective and easy-to-control mathematical model to describe the microwave heating process.In recent years,neural networks have been widely used in nonlinear dynamic system modeling due to their powerful nonlinear mapping capabilities and adaptive learning capabilities.Therefore,based on a large amount of offline data during the microwave heating process,neural network training is used to obtain the mathematical model close to the actual microwave heating process.Based on the trained neural network model,an adaptive control strategy based on event-triggered is used to control the temperature of the medium in real time.The main research contents of this thesis are as follows:(1)Aiming at the complex modeling of the microwave heating process,it is difficult to design a suitable controller for real-time temperature control of the medium.Based on a large amount of offline historical data obtained during the microwave heating process,an abstract physical model with multiple inputs and multiple outputs is established.Where the power input of five microwave sources,the conveyor speed and the current temperature at the exit of the three chambers are used as input,and the temperature at the next moment at the exit of the three chambers are used as output.The Elman neural network is used to identify the abstract physical model.(2)Aiming at the optimal control of the medium temperature of the microwave heating process.Based on the trained neural network model,the Heuristic Dynamic Programming(HDP)algorithm is used to control the temperature of the medium.The execution network and the evaluation network are used to approximate the optimal control law and optimal performance index function of the control system respectively.Finally,the temperature at the outlet of the cavity is stabilized by adjusting the five power input of the microwave source and the speed input of the conveyor belt.(3)Aiming at the problem of long calculation time for the traditional HDP algorithm not conducive to the real-time control of the medium temperature and the frequent controller adjustment not conducive to the protection of the magnetron during the microwave heating process.On the basis of the traditional HDP algorithm,an event-triggered control strategy is introduced.The event-triggered threshold condition is designed on the premise of ensuring the stability of the system.When the difference between the current state of the system and the sampling state value exceeds the set threshold,the evaluation network and the execution network weight update are updated.In the end,HDP based on event-triggered greatly reduces the sampling amount compared with traditional HDP algorithm,and greatly reduces the system calculation time while ensuring the stability of the system.At the same time,reducing the update frequency of the controller achieves the purpose of protecting the magnetron.