Research On Fuzzy Inference Method Based On Variable Universe For Transient Inverse Heat Transfer Problems

Inverse heat transfer problems(IHTP) is a typical inverse problems, which involves the estimation of the unkown parameters such as thermophysical parameters, boundary conditions, geometry shape and the source, based on the internal or surface temperature information of the heat transfer systems. IHTP is widely used in scientific research and the engineering filed such as power engineering, aerospace, construction, environment, biomedical, and nondestructive testing.Decentralized fuzzy inference method(DFI) which based on the fuzzy set theory is a typical uncertain inference method and calculation framework. DFI which describe the knowledge and experience with the language rules, combined with advanced computer technology to mimic human thinking and making decesions was used to solve the steady inverse heat conduction problem, showing good performance of anti-ill-posed problems. Based on the previous research results, this article tries to apply fuzzy inference method to study the transient heat transfer inverse problem. The main works include:(1)Take for example the problem of estimating the one-dimensional transient flat boundary heat flux, the model of inverse heat transfer problems is established by the sequential function specification method(SFSM) that is a typical algorithms. Through numerical simulation experiments, the influence of the number of future time steps, the location of measuring points and temperature measurement errors to the inversion results are discussed, and the limits of the SFSM used for solve the inverse transient heat transfer problems are summarized.(2)For the limits of the SFSM used for solving the inverse heat transfer problems, this article tries to apply fuzzy inference method to study the transient heat transfer inverse problem. Firstly, this method obtain the temperature of measuring point at the future time steps R moment by system temperature at k1t- moment and zero input. Construct a set of decentralized fuzzy inference unit, inputing the bias between temperature calculations and temperature measurements at the measured point at the number of future time steps R to infer get a set of fuzzy inference component, and the fuzzy inference components are weighted and synthesized to get the inverse parameter Estimated value.(3)For the characteristics of the bias between temperature calculation value and the temperature measurements of the measured point may may change, when the thermal boundary condition changed obviously. For this problems, design a scheme of decentralized fuzzy inference method method based on variable universe, which is the input universe of fuzzy inference unit is adjusted online though the maximum absolute value of the bias vector elements, ensure the effectiveness of the fuzzy rules, Matching of input universe of fuzzy inference unit and fuzzy subsets.(4)Research on the comprehensive coordination of the decentralized inference unit result to study transient heat transfer inverse problem. It proposed according to the sensitivity coefficient to set the appropriate weighting matrix. On the basis of the above work, the decentralized inverse inference method based on variable universe(VDFI) was established apply to slove transient heat transfer inverse problem.(5)Using the VDFI to estimate the boundary heat flux inverse problem of onedimensional transient heat transfer model and two-dimensional transient heat transfer model, the inversion results about of the influence of the number of future time steps, temperature measurement errors and the location measuring points are discussed. The results show that VDFI compared to SFSM can significantly reduce the dependence on the future time steps, weaken the effect of the location of the measuring point and has good performance to anti ill-posedness.