Temperature Control For Microwave Heating Process Based On Recurrent Fuzzy Quantum Neural Network

Energy is an indispensable part of human survival and social development.Faced with the increasingly serious problem of fossil energy exhaustion and the environmental problems caused by its abuse,the research,development and use of new energy have attracted much attention.As a kind of clean and efficient energy,microwave energy is widely used in industrial heating and home cooking.However,due to its complex heating environment and easy hot spots,thermal runaway and thermal inequality in the heating process,the safe and efficient application of microwave has brought great challenges.This paper is devoted to solving problems related to mathematical modeling and thermal uniformity temperature control of the medium during microwave thermal process.for the uniform time-varying media with known load characteristic parameters,Lambert's law and Maxwell equations were used to approximate and accurately describe the microwave power density distribution in the media under microwave radiation.The experimental results show that when the thickness of the medium is larger than the penetration depth of the microwave,Lambert's law can be used to calculate the microwave power distribution as accurately as Maxwell's equation.However,when the size and thickness of the medium are small,the calculation results based on Lambert's law have a large error,and its simple exponential distribution cannot accurately describe the fluctuation characteristics of microwave power distribution in the medium.To solve the problem that the mechanism model of unknown time-varying microwave thermal process is difficult to establish accurately,this paper proposes a recursive fuzzy quantum neural network(RFQNN)to establish the prediction model of microwave thermal process.The traditional multi-layer feedforward neural network and recursive fuzzy network used in unknown time-varying systems are prone to overfitting,poor generalization ability,and irreconcilable contradictions between training errors and testing errors.In this paper,the proposed RFQNN has strong self-learning ability of network structure and fast parameter updating and convergence ability.In addition,the quantum layer is introduced into the later items to replace the traditional TSK or functional link input,and the precise identification and prediction of the uncertain system of the network are realized by quantizing,rotating and reversing the input eigenvalues.In the application of time series prediction of dynamic system and model prediction of actual microwave heating process,RFQNN shows a better prediction performance than classical self-evolving recursive fuzzy network(RSFNN-TSK),local feedback self-evolving recursive fuzzy network(RSEFNN-LF)and full-feedback functional link self-evolving recursive fuzzy network(IRSEFNN-FL).To solve the thermal uniformity control problem of medium temperature during microwave heating,an interactive self-evolving recursive fuzzy network(IRSFNN)based on functional linkage is proposed.For the unknown time-varying system,IRSFNN as the controller has good dynamic performance and generalization ability.Based on this,a design method of microwave heating temperature control system is proposed.First,IRSFNN was used to study the microwave heating process off-line.Then,the unknown changes in the heating process were studied online to optimize and update the controller parameters.Finally,deionized water and white pudding were used as heating objects for simulation.The results show that the design method of this system has a stronger capability of expected temperature tracking.