Study On Multi-Source Microwave Heating Process Modeling Based On Numerical Simulation And NARX Neural Network

Heating has become one of the indispensable processes in the industrial manufacturing process,and the way of microwave heating stands out by advantages of its selective energy absorption,fast start-up speed,high resource utilization rate,short heating time,clean and environmental protection.Microwave heating not only has a wide range of applications in industrial field,such as curing,sintering and drying,but also plays a pivotal role in promoting physicochemical reactions and improving media processing.For industrial high-power tunnel microwave heating systems,the equipment is relatively closed,and the internal reaction environment is harsh,making it difficult to observe the heating state of the material.The "hot spots" or "thermal runaways" that may occur during the heating process will deteriorate the quality of the product,and even bring hidden safety hazards,seriously hinder the practical industrial application of microwave heating technology.In addition,the microwave heating process is complex and variable,with characteristics of nonlinearity,strong coupling and uncertainty,making it hard to establish temperature model of the microwave heating process.Besides,in the actual production process,frequent material changes,complex production processes,and continuous operation of equipment also pose challenges to the establishment of microwave heating process models.Therefore,it is significant to conduct numerical simulation in order to present the heating state of the material and obtain better working conditions,and build a mathematical model that can accurately describe the characteristics of the system while maintaining a low complexity for the heating process.It will lay a solid foundation for the subsequent selection of reasonable control strategies to ensure safe and efficient operation of the system.The main research contents of this thesis are as follows:(1)COMSOL Multiphysics simulation software is utilized to numerically simulate the microwave heating process.For the problem of low meshing quality when solving,a user-defined method is adopted to divide different size standards according to different model structures to obtain a convergent and reliable model.Based on this simulation software,an application(APP)for mutil-source microwave heating system is created,which is convenient to explore the effects of different factors on the electromagnetic field or temperature distribution of the material during microwave heating process.Finally,visualize the heating state of material and determine the optimal working conditions of the material so as to guide the later experiment.(2)Aiming at the problem of high complexity of mechanism model and numerical simulation solving process,microwave heating process modeling based on neural networks are studied.Using the experimental data obtained under better conditions,BP neural network,Elman neural network,and non-linear autoregressive(NARX)neural network with exogenous input are used to establish temperature identification model of microwave heating process.After analyzing and comparing these three models,the accuracy of the NARX neural network model is verified.(3)For the problem that input delay order,feedback delay order,and the number of hidden layer neurons in the NARX neural network are difficult to determine,a microwave heating temperature model based on the lightning search algorithm(LSA)optimization is built to automatically optimize the structural parameters of the NARX neural network model.The model is compared with a model based on particle swarm optimization,and then verify the effectiveness and accuracy of the NARX-LSA model.