Modeling Method For Volumetric Specific Heat And Heat Transfer Of Floor Heating Materials Based On Fuzzy Regression Analysis

With the development of society and people's increasing needs for a beautiful life,the public's requirements for indoor heating conditions are also increasing.The combination of floor heating and wooden floor decoration has gradually become the first choice for residential,office and other places because of its comfort and beauty.In recent years,the scale of floor heating industry is expanding,and the theme of "energy saving,environmental protection and comfort" is gradually clear.However,there is no perfect testing technology system for ground heating performance in our country.The research on the specific heat and heat transfer performance of wood floor is the basis of analyzing and evaluating the comfort and energy saving indexes.Wood floor heating is a kind of heterogeneous biomass material with anisotropy.In order to study the volume specific heat and heat transfer performance of wood flooring,the thermal physical properties of heterogeneous materials should be studied instead of the thermal physical properties of homogeneous materials.Therefore,according to the properties of wood volume specific heat and heat transfer,this thesis established the regression model of wood volume specific heat based on fuzzy regression analysis,the anisotropic heat conduction model and the wood classification model based on the value of wood heat transfer parameters,to provide the basic theoretical support for the subsequent research.The volume specific heat capacity,radial thermal diffusivity,radial thermal conductivity,chord wise thermal diffusivity,chord wise thermal conductivity and specific emissivity spectra of 130 kinds of wood samples were measured by Hot Disk thermal constant analyzer and BRUKER Fourier infrared spectrometer.Screening and feature extraction processing of experimentally collected data provide a data basis for modeling work.Furthermore,according to the experimental data of wood volume specific heat capacity,the wood volume specific heat model is established by the modeling method based on the semi-tensor product multivariable fuzzy system.This method used the semi-tensor product principle of the matrix to establish a fuzzy relationship matrix between multiple variables,instead of the fuzzy inference process in the traditional fuzzy logic theory.The fuzzy logic system was constructed and the model was optimized by the recursive least square method which was suitable for experimental data acquisition.The validity of the established model was verified by a variety of evaluation indexes.After many experiments,the fitting degree of the verification set was all greater than 0.99.The wood volume specific heat model established in this study was compared with the model established by ANFIS,which proved the superiority of the wood volume specific heat model established in this study.Aiming at the experimental data of anisotropic heat conduction of wood,a fuzzy least absolute nonlinear regression method was proposed.In this method,the least absolute regression criterion is combined with fuzzy logic theory.Firstly,Gaussian membership function is constructed to fuzzy the data.Furthermore,a single value fuzzier was constructed to generate a fuzzy rule base.Use the product inference engine to perform fuzzy inference on the input space elements and get the inference result.Finally,the least absolute regression criterion was used to optimize the obtained results.In this way,a kind of wood radial and chord thermal diffusivity and thermal conductivity models with nonlinear properties were established,which can analyze and predict the thermal diffusion and thermal conduction process of anisotropic wood in real time.Finally,a classification method for wood species identification was proposed by applying the model of wood thermal physical properties to the study of wood classification.Based on the thermo physical properties of wood,the differences of thermal conductivity,thermal diffusivity and specific emissivity of different tree species under the same experimental conditions were considered.The characteristic spectral data and thermal property parameter data selected by principal component analysis were fused and input into the neural network model to identify the wood species.The simulation results show the feasibility of wood classification based on the value of wood thermal physical properties.