Model Predictive Inverse Method For Heat Transfer Process And Application

Inverse Heat Transfer Problems(IHTP)are to use the partially observable information to estimate the internal characteristics or thermal boundary conditions of the heat transfer system.The inverse problem of heat transfer is a typically ill-posed problem.The results of inversion usually do not have the continuous dependence of the measurement information.The measurement error is likely to be significantly amplified in the inversion process,resulting in the instability of the inversion process.The key problem of the inverse heat transfer is to seek a new method to solve the inverse heat transfer problem,especially for good anti-ill-posed inversion method,which also meets the practical needs of engineering field.This paper proposes the method to solve the inverse heat transfer problem based on the model prediction thought,and use the method to study several kinds of typically inverse heat transfer problem.The main research works include:(1)A model predictive inverse method(MPIM)of heat transfer process is proposed.Based on the heat transfer model,the step response function corresponding to the control equation is built up.As a result,we obtain the system's dynamic matrix and set up the predictive temperature model of spatial points.Then,according to obtained information,we can get the optimal time series of the pending parameter in the future limited time domain by the rolling optimization.The research shows that the rolling optimization inversion scheme based on model prediction can effectively reduce the sensitivity of the inversion results to temperature measurement error and dependence of the future observation information,which has obvious anti ill-posed ability and also improves the reliability of the inversion results.(2)To solve multiple thermal boundary conditions inversion problem and the temporal spatial distributed thermal boundary conditions inversion problem,a set of predictive model units(PMUs)corresponding to the temperature observation point is constructed based on system input and output space decomposition.In each PMU,we can get the whole space predictive temperature matrix by using the temporal temperature vector which obtained from the inversed information.Further,we can obtain the optimal estimation of thermal boundary condition based on the whole space predictive temperature matrix by rolling optimization.A new method is proposed from the above work for multiple thermal boundary conditions inversion problem and the temporal spatial distribution inversion problem.(3)To deal with the inverse problem of the nonlinear heat transfer system,a fuzzy adaptive predictive inverse method(FAPIM)is proposed.In order to solve the high dimension fuzzy inference problem of fuzzy adaptive prediction method,we design a decentralized fuzzy inference(DFI)structure.Through a set of fuzzy inference unit,based on the time domain prediction error vector,we go on decentralized fuzzy inference to obtain different time prediction model of fuzzy inference components corresponding to the error.And through the weighted fuzzy reasoning component,we can adjust the prediction matrix of the temperature in measuring points adaptively and update temperature prediction model online.Furthermore,the adaptive inversion of the thermal boundary conditions of the nonlinear heat transfer system is realized by using the fuzzy adaptive prediction model of the heat transfer system and the aforementioned MPIM.(4)Based on the actual temperature measurement information,the simultaneous estimation of the transient heat fluxes on the surface of the billet in the steel rolling furnace and the transient heat flux distribution on the surface of the disc brake are studied by using the aforementioned MPIM.There is a significant difference between the upper,lower and side surface heat flux intensity.Aiming at the problems of the traditional unified regularization method in the study of inverse heat transfer problems,a new method of regularization is proposed.A same temperature discrepancy curve(STDC)is designed to estimate the optimal values of the regularization parameters of the upper as well as lower surface heat flux and the side surface heat flux respectively.The research results show that the above scheme effectively improves the stability of the inverse results of surface heat flux with steel slab.In addition,for the above two types of actual heat transfer system,the influences of the factors such as the temperature measurement errors and the number of the temperature measurement points on the inversion result are discussed by the numerical experiments.The validity of MPIM for solving the inverse problem of real linear heat transfer system is proved.(5)Based on the aforementioned FAPIM,the inverse problems of two typical nonlinear heat transfer systems are studied.According to the measured temperature data obtained from the bone grinding experiment in minimally invasive neurosurgery,the instantaneous grinding heat of the process is inversed,and the transient temperature field of the bone is reconstructed.The effects of different inverse conditions on the inverse results of bone grinding heat sources are discussed and the reliability of the inverse results is verified.The transient boundary heat flux inverse problem of molten material reactor phase change heat transfer process is studied by the FAPIM.The influences of temperature measurement errors on the inverse results are investigated by the numerical experiments.The inverse results of boundary heat flux are used to identify the movement trajectory of phase change interface about molten material reactor,which provides an effective way to reconstruct the movement trajectory of phase change interface about phase change heat transfer process.