A weighted fuzzy inversion method for transient heat load alternating direction

Abstract

The present invention relates to the technical field of heat transfer inversion, and particularly to a transient distributed heat load alternating direction weighted fuzzy inversion method. It includes: Step 1: Establish a heat transfer model of the object to be measured, arrange N m temperature measurement points on the surface to be measured, initialize the transient distributed heat load detection problem, and give the initial guess value matrix #imgabs0# of the boundary heat load. Set the initial value of the iteration number identifier of the inversion process to n = 0; Step 2: Solve the forward heat transfer problem according to the guess value of #imgabs1# to obtain the calculated temperature value matrix #imgabs2# at the measurement points, calculate the temperature error matrix #imgabs3#, and judge whether E n satisfies the iteration stop condition. If it is satisfied, the iteration stops, and the distributed heat load is output. The method of the present invention significantly reduces the requirement for the number of measurement points, is not sensitive to the initial guess value of the boundary heat load, and has good anti-interference ability to measurement errors.

Description

Technical Field

[0001] The invention relates to the technical field of heat transfer inversion, and in particular to a transient distributed heat load alternating direction weighted fuzzy inversion method. Background Art

[0002] Inverse Heat Transfer Problem (IHTP) refers to the inverse solution of some unknown characteristic parameters of the system, such as boundary conditions, thermophysical parameters, geometric shape, initial conditions, and source terms, based on partial temperature information inside or on the surface of the heat transfer system. Inverse heat transfer problems are widely used in engineering fields such as aerospace, power engineering, material processing, bioengineering, and non-destructive testing.

[0003] Inverse problems in heat transfer are usually ill-posed problems in the Hadamard sense, that is, the existence, uniqueness and stability of the solution cannot be satisfied at the same time. Due to the ill-posedness of the inverse problem, for classic inversion algorithms such as the Conjugate Gradient Method (CGM) and the Levenberg-Marquard method (L-MM), when the number of measurement points decreases or the measurement error increases, the inversion results often deteriorate seriously.

[0004] Fuzzy reasoning is based on fuzzy theory, has obvious anti-interference ability for input information, and the reasoning process has good robustness and fault tolerance; it can effectively use imprecise, uncertain and incomplete information for reasoning and decision-making; it can comprehensively use qualitative knowledge (including empirical knowledge) and quantitative knowledge to complete the reasoning process, and has a low computational cost. These characteristics of fuzzy reasoning methods can undoubtedly provide substantial help for solving ill-posed problems including heat transfer inverse problems. Although decentralized fuzzy reasoning methods have been studied in heat transfer inverse problems, this method has not yet been effectively applied to solving unsteady heat transfer inverse problems. The difficulty of using decentralized fuzzy reasoning methods to solve unsteady heat transfer inverse problems lies in constructing an appropriate comprehensive weighting method for unsteady inverse problems. Inappropriate weighting methods will not only seriously affect the convergence speed, but also cause the inversion results to diverge. Summary of the invention

[0005] In order to solve the prior art problems in the background technology, the purpose of the present invention is to provide a transient distributed heat load inversion method which can significantly reduce the requirement on the number of measuring points, is insensitive to the initial guess value of the boundary heat load, and has good anti-interference ability to measurement errors.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A transient distributed heat load alternating direction weighted fuzzy inversion method, characterized by comprising the following steps:

[0008] Step 1: Establish a heat transfer model for the object to be measured and arrange N m Temperature measurement points are used to initialize the transient distributed heat load detection problem, and the initial guess value matrix of the boundary heat load is given. Set the initial value of the inversion process iteration number identifier to n = 0;

[0009] Step 2: According to The guessed value of solves the heat transfer problem and obtains the temperature calculation value matrix at the measuring point Calculate the temperature error matrix Judge E n Whether the iteration stop condition is met, if it is met, the iteration stops and the distributed heat load is output; otherwise, go to step 3;

[0010] Step 3: E n As the input of the fuzzy reasoning module FIM, decentralized fuzzy reasoning is performed to obtain the fuzzy reasoning result matrix ΔU t ;

[0011] Step 4: ΔU t Perform time-direction weighted synthesis to obtain the compensation matrix Δq of the heat load t ′, through the interpolation function, the spatial direction interpolation is performed, and the heat load compensation amount becomes Δq t ;

[0012] Step 5: According to the formula Update the current guess of the heat load at the point to be inverted to obtain an estimate of the heat flow that only takes into account the time weighting

[0013] Step 6: Based on Solve the heat transfer problem to obtain the temperature calculation value matrix at the measuring point And calculate the temperature error matrix

[0014] Step 7: E n+1 / 2 As the input of the fuzzy reasoning module FIM, fuzzy reasoning is performed, and the fuzzy reasoning result matrix ΔU s Perform weighted synthesis in spatial direction to obtain the compensation matrix Δq of heat flow s ;

[0015] Step 8: According to the formula Matrix of current guesses for heat flux Refresh to obtain heat flow estimates that take both time and space weighting into account

[0016] Step 9: Set n=n+1; return to step 2 and continue the loop until the shutdown criterion is met.

[0017] Furthermore, in step 1, the mathematical description of the heat transfer model is as follows:

[0018]

[0019] T(x,y,t)=T0 0≤x≤L x ,0≤y≤L y ,t=0

[0020]

[0021]

[0022]

[0023]

[0024] Among them, x = (x, y) is the spatial coordinate vector; T is the temperature; ρ is the density; c p is the specific heat capacity; λ is the thermal conductivity; h is the convective heat transfer coefficient; T amb is the ambient temperature; T0 is the initial temperature distribution; t f is the simulation end time;

[0025] In the heat transfer model (x m ,y m ) Arrange N m A temperature sensor measures the temperature Y(x m ,y m ,t k )(m=1,2,…,N m ), the unknown boundary heat load distribution q(x,t) is inverted, and the heat load is discretely expressed as:

[0026]

[0027] Furthermore, in step 2,

[0028]

[0029]

[0030] in The boundary heat flow is expressed as When , the calculated temperature matrix is ​​obtained according to the heat transfer direct problem calculation;

[0031] The criteria for stopping the iteration are as follows:

[0032]

[0033] In the formula, ε is a given small positive number. According to the deviation principle, ε=N m N t σ 2 ; When the measurement error is not considered (i.e. when σ=0), take ε=0.001℃ 2 , σ is an estimate of the standard deviation of the measurement error.

[0034] Further, in step 3, the fuzzy inference module FIM uses N m ×N t Decentralized Fuzzy Inference Unit FIU i (i=1,2,...,N m ×N t ), used to realize the temperature error matrix by any element e i To any element Δu of the fuzzy inference result matrix i the reasoning process;

[0035] Take e i and Δu i The fuzzy set domains are [-p e ,p e ] and [-p u ,p u ], e i and Δu i The domain is divided into 7 fuzzy sets {A1, A2, ..., A7} and {B1, B2, ..., B7}. The language values ​​corresponding to these fuzzy sets are: NB (negative large), NM (negative medium), NS (negative small), ZE (zero), PS (positive small), PM (positive medium), PB (positive large). The triangular membership function is used to determine the membership of each fuzzy set.

[0036] FIU i The fuzzy inference rules are as follows:

[0037]

[0038] N m ×N t The inference results form the inference result matrix, that is

[0039]

[0040] Further, in step 4, the time direction weighted synthesis is constructed using a chi-square distribution model;

[0041] The chi-square distribution probability density function is:

[0042]

[0043] where n f is the degree of freedom of the chi-square distribution, which is a positive integer; adjusting the chi-square distribution degree of freedom can change the response mode of temperature to heat flow; t k =kdt(k=1,2...,N t ), is the discrete time; Γ(x) is the gamma function, that is,

[0044] The time weighted coefficient model based on the chi-square distribution is:

[0045]

[0046] After normalization, the weighted coefficient is obtained: When i<j, That is, the time weighting matrix Λ t is a lower triangular matrix;

[0047] Through the weighted comprehensive formula in the time domain [Δq t ′] Nm×Nt =[ΔU t ] Nm×Nt [Λ t ] Nt×Nt , generating the time domain inference compensation matrix. Since the number of inversion points in space is not equal to the number of measurement points, the matrix [Δq t ′] Nm×Nt Transformed into [Δq t ] Ne×Nt .

[0048] Furthermore, step 6 mainly includes:

[0049] based on Solve the heat transfer forward problem model established in step 1 to obtain the temperature calculation value matrix at the measuring point And calculate the temperature error matrix as follows:

[0050]

[0051] Further, in step 7, wherein E n+1 / 2 As the input of the fuzzy reasoning module FIM, fuzzy reasoning is performed, and the reasoning method is the same as step 4;

[0052] Among them, the weighted synthesis of row space direction mainly includes:

[0053] Spatial domain weighted comprehensive matrix Λ s Using a comprehensive weighted model based on normal distribution, the spatial weighted comprehensive matrix Λ s element for

[0054]

[0055] In the formula, θ>0 is the variance coefficient of the normal distribution, x i and x j are the positions of the inversion point and the measurement point in the x direction respectively.

[0056] The present invention has at least the following beneficial effects:

[0057] The present invention can utilize limited local temperature measurement information and, based on the established transient distributed heat load time-space alternating direction inversion method, estimate and detect the widely existing boundary heat load that is difficult or impossible to measure directly.

[0058] In actual use, the present invention can significantly reduce the requirement of the inversion process on the number of measuring points, is insensitive to the initial guess value of the heat load distribution, and has good anti-interference ability to the measurement temperature error. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Schematic diagram of the system of the method of the present invention.

[0060] Figure 2 Schematic diagram of a two-dimensional non-steady-state heat conduction system in an embodiment of the present invention.

[0061] Figure 3 for e i For the fuzzy set A l The degree of membership.

[0062] Figure 4 is Δu i For the fuzzy set B l The degree of membership.

[0063] Figure 5 For x e = 0.065m at different initial guess values ​​of the heat flow inversion results of the present invention.

[0064] Figure 6 For x e CGM heat flow inversion results for different initial guess values ​​at =0.065m.

[0065] Figure 7 For t = 105s, N m =15 when the heat flow inversion results of the method of the present invention are compared with those of CGM.

[0066] Figure 8 For t = 105s, N m =10 when the heat flow inversion results of the method of the present invention are compared with those of CGM.

[0067] Fig. 9 For t = 105s, N m =5 when the heat flow inversion results of the method of the present invention are compared with those of CGM.

[0068] Fig.10 For t = 105s, N m =15, σ=0.05°C. The heat flow inversion results of the method of the present invention are compared with those of CGM.

[0069] Fig.11 For t = 105s, N m =15, σ=0.1℃, the heat flow inversion results of the method of the present invention are compared with those of CGM.

[0070] Fig.12 For t = 105s, N m =5, σ=0.05°C. The heat flow inversion results of the method of the present invention are compared with those of CGM.

[0071] Fig.13 For t = 105s, N m =5, σ=0.1℃, the heat flow inversion results of the method of the present invention are compared with those of CGM. DETAILED DESCRIPTION

[0072] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0073] Taking the boundary heat flow inversion of two-dimensional unsteady heat conduction process of steel billet as an example, refer to Figure 1 and 2 A specific embodiment of a transient distribution heat load alternating direction weighted fuzzy inversion method comprises the following steps:

[0074] 1) Establish a heat transfer model for two-dimensional boundary heat flow inversion ( Figure 2 ), arrange N on the measured surface m Temperature measurement points are used to initialize the transient distributed heat load detection problem, and the initial guess value matrix of the boundary heat load is given. Set the initial value of the inversion process iteration number identifier to n = 0;

[0075] The mathematical description of the heat transfer model consists of control equations, boundary conditions and initial conditions.

[0076]

[0077] T(x,y,t)=T0 0≤x≤L x ,0≤y≤L y ,t=0

[0078]

[0079]

[0080]

[0081]

[0082] Among them, x = (x, y) is the spatial coordinate vector; ρ is the density; c p is the specific heat capacity; λ is the thermal conductivity; h is the convective heat transfer coefficient; T amb is the ambient temperature; T0 is the initial temperature distribution; t f The simulation end time.

[0083] In the heat transfer model (x m ,y m ) Arrange N m A temperature sensor measures the temperature Y(x m ,y m ,t k )(m=1,2,…,N m ), the unknown boundary heat load distribution q(x,t) is inverted, and the heat load is discretely expressed as

[0084]

[0085] Among them, Ne is the number of spatial inversion points.

[0086] 2) According to The guessed value of solves the heat transfer problem and obtains the temperature calculation value matrix at the measuring point Calculate the temperature error matrix Judge E n Whether the iteration stop condition is met, if it is met, the iteration stops and the distributed heat load is output; otherwise, go to step 3)

[0087]

[0088]

[0089] in The boundary heat flow is expressed as The computational temperature matrix obtained from the direct heat transfer problem when .

[0090] The criteria for stopping the iteration are as follows:

[0091]

[0092] In the formula, ε is a given small positive number. According to the deviation principle, ε=N mN t σ 2 ; When the measurement error is not considered (i.e. when σ=0), take ε=0.001℃ 2 , σ is an estimate of the standard deviation of the measurement error.

[0093] 3) E n As the input of the fuzzy reasoning module FIM, decentralized fuzzy reasoning is performed. The fuzzy reasoning module FIM uses N m ×N t Decentralized Fuzzy Inference Unit FIU i (i=1,2,…,N m ×N t ), realize the temperature error matrix by any element e i To any element Δu of the fuzzy inference result matrix i reasoning process.

[0094] Take e i and Δu i The fuzzy set domains are [-p e ,p e ] and [-p u ,p u ]. i and Δu i The domain is divided into 7 fuzzy sets {A1, A2, ..., A7} and {B1, B2, ..., B7}, and the language values ​​corresponding to these fuzzy sets are: NB (negative large), NM (negative medium), NS (negative small), ZE (zero), PS (positive small), PM (positive medium), PB (positive large). The triangular membership function is used to determine the membership of each fuzzy set γ Al (e i ) and γ Bl (Δu i ),like Figure 3 and Figure 4 FIU i The fuzzy reasoning rules are shown in Table 1.

[0095] Table 1 FIU i Fuzzy inference rules

[0096]

[0097] N m ×N t The inference results form the inference result matrix, that is

[0098]

[0099] 4) For ΔU t Perform time-direction weighted synthesis to obtain the compensation matrix Δq′ of the heat loadt , through the interpolation function, the spatial direction interpolation is performed, and the heat load compensation amount becomes Δq t , the time domain weighted comprehensive matrix Λ t The chi-square distribution model is used. The chi-square distribution probability density function is:

[0100]

[0101] where n f is the degree of freedom of the chi-square distribution, which is a positive integer. Adjusting the chi-square distribution degree of freedom can change the response pattern of temperature to heat flow. k =kdt(k=1,2...,N t ), is the discrete time. Γ(x) is the gamma function, that is

[0102] The time weighted coefficient model based on the chi-square distribution is:

[0103]

[0104] After normalization, the weighted coefficient is obtained: When i<j, That is, the time weighting matrix Λ t is a lower triangular matrix. This is because heat flow only affects the moment after the heat flow acts, and has no effect on the moment before the heat flow acts.

[0105] Through the weighted comprehensive formula in the time domain [Δq′ t ] Nm×Nt =[ΔU t ] Nm×Nt [Λ t ] Nt×Nt , generating the time domain inference compensation matrix. Since the number of inversion points in space is not equal to the number of measurement points, the matrix [Δq′ t ] Nm×Nt Transformed into [Δq t ] Ne×Nt .

[0106] 5) According to the formula Update the current guess of the heat load at the point to be inverted to obtain an estimate of the heat flow that only takes into account the time weighting

[0107] 6) Based on Solve the heat transfer forward problem model established in step 1) to obtain the temperature calculation value matrix at the measuring point And calculate the temperature error matrix

[0108]

[0109] 7) Using the same reasoning method as in step 4), n+1 / 2 As the input of the fuzzy reasoning module FIM, fuzzy reasoning is performed to obtain the fuzzy reasoning result matrix ΔU s . N m ×N t The inference results form the inference result matrix, that is

[0110]

[0111] Spatial domain weighted comprehensive matrix Λ s Using a comprehensive weighted model based on normal distribution, the spatial weighted comprehensive matrix Λ s element for

[0112]

[0113] In the formula, θ>0 is the variance coefficient of the normal distribution, x i and x j are the positions of the inversion point and the measurement point in the x direction respectively.

[0114] According to the formula [Δq s ] Ne×Nt =[Λ s ] Ne×Nm [ΔU s ] Nm×Nt Perform weighted synthesis in spatial direction to obtain the compensation matrix Δq of heat flow s ;

[0115] 8) According to the formula Matrix of current guesses for heat flux Refresh to obtain heat flow estimates that take both time and space weighting into account

[0116] 9) Set n=n+1; return to step 2) and continue the loop until the shutdown criterion is met.

[0117] The following is an example of inversion verification of two-dimensional non-steady-state steel billet boundary heat flow using the method of the present invention, and the inversion result of the method of the present invention is compared with the inversion result of the conjugate gradient method (CGM) to verify the effectiveness of the method of the present invention.

[0118] Geometric parameters of the billet L x =0.15m, L y =0.075m. Thermophysical parameter ρ = 7854kg / m 3 , c p =445J / kg·℃,λ=43.5W / m·℃. Ambient temperature Tamb =20℃, convection heat transfer coefficient h=60W / m 2 ℃. Initial temperature T0 = 20℃. Simulation end time t f = 120s. Assume that the heat flux distribution on the upper surface of the billet is as follows:

[0119] q(x,t)=100+50sin(πx / L x )sin(πt / t f )kW / m 2

[0120] In actual engineering problems, due to human or non-human factors, there are inevitable measurement errors in the temperature measurement results of thermocouples. In numerical simulation calculations, the temperature "measurement value" matrix Y of the thermocouple is usually calculated as follows:

[0121] Y=T(q)+σω

[0122] Where T(q) represents the precise temperature value obtained by calculating the direct problem through the accurate heat flow, σ is the standard deviation of the measurement error, and ω is a random number matrix that obeys the standard normal distribution with a confidence interval of [-2.576, 2.576] corresponding to a 99% confidence level.

[0123] The method of the present invention and CGM use the same iteration stop criteria when inverting heat flow.

[0124] The fuzzy domain of the input error of the fuzzy reasoning unit is p e =1℃, the output heat flow fuzzy domain is p u =500W / m 2 , spatially weighted normal distribution with variance θ = 0.1, and time-weighted chi-square distribution with degrees of freedom n f =5.

[0125] Take N m =15,σ=0℃. 100kW / m 2 and 125kW / m 2 (i=1,2,…,N e ; k = 1, 2, ..., N t ), without considering the measurement error, the influence of the initial value of heat flow on the inversion result is investigated. The inversion results corresponding to the method of the present invention and CGM are shown in Figure 5 and Figure 6 .

[0126] Figure 5 It shows that for the given initial value of heat flow, the method of the present invention can obtain good inversion effect, and for CGM, The inversion result is good when and When , the inversion result graph shows a slight offset (see Figure 6 ). It can be seen that the method of the present invention is not sensitive to the selection of initial values.

[0127] Pick σ=0℃. Take the number of temperature sensors N m =15, N m =10 and N m =5, the influence of the number of measuring points on the inversion is discussed without considering the measurement error, and compared with the inversion results of CGM. Figure 7-9 As shown. Figure 7 It can be seen that when N m = 15, both the method of the present invention and CGM can obtain good inversion effect. m =10 and N m =5, the method of the present invention is in N m =10 and N m =5, a relatively accurate inversion result can still be obtained; however, the inversion result of CGM gradually deteriorates, and the fewer the measurement points, the worse the inversion result.

[0128] It can be seen that compared with CGM, the method of the present invention can still obtain stable and effective inversion results when the number of temperature measuring points is significantly reduced, and significantly reduces the dependence on the number of temperature measuring points.

[0129] Take the number of measuring points N m =15, initial value of heat flow The standard deviation of the measurement error is taken as σ = 0.05°C and σ = 0.1°C, respectively, to investigate the impact of the measurement error on the inversion results. Fig.10 and Fig.11 The heat flow inverted by the present invention and CGM at t = 105 s are plotted respectively. It can be observed that the oscillation of the inversion result of CGM is obviously stronger than that of the present invention method.

[0130] Furthermore, the number of measuring points is reduced to 5, and the standard deviation of the measurement error is taken as σ = 0.05℃ and σ = 0.1℃ respectively. The inversion results are as follows: Figure 12-13 As shown in the inversion results, it can be seen that for the method of the present invention, when the number of measurement points decreases, the average relative error of the inversion result remains almost unchanged, but the inversion error of CGM increases significantly. This also proves that the method of the present invention is not sensitive to the number of measurements.

[0131] The above inversion results show that, compared with CGM, the method of the present invention has enhanced anti-interference ability against measurement errors.

[0132] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A weighted fuzzy inversion method for alternating directions of transient distributed heat loads, characterized in that: The following steps are involved: Step 1: Establish a heat transfer model for the object to be measured and arrange N m Temperature measurement points are used to initialize the transient distributed heat load detection problem, and the initial guess value matrix of the boundary heat load is given. Set the initial value of the inversion process iteration number identifier to n = 0; Step 2: According to The guessed value of solves the heat transfer problem and obtains the temperature calculation value matrix at the measuring point Calculate the temperature error matrix Judge E n Whether the iteration stop condition is met, if it is met, the iteration stops and the distributed heat load is output; otherwise, go to step 3; Step 3: E n As the input of the fuzzy reasoning module FIM, decentralized fuzzy reasoning is performed to obtain the fuzzy reasoning result matrix ΔU t ; Step 4: ΔU t Perform time-direction weighted synthesis to obtain the compensation matrix Δq of the heat load t ′, through the interpolation function, the spatial direction interpolation is performed, and the heat load compensation amount becomes Δq t ; Step 5: According to the formula Update the current guess of the heat load at the point to be inverted to obtain an estimate of the heat flow that only takes into account the time weighting Step 6: Based on Solve the heat transfer problem to obtain the temperature calculation value matrix at the measuring point And calculate the temperature error matrix Step 7: E n+1 / 2 As the input of the fuzzy reasoning module FIM, fuzzy reasoning is performed, and the fuzzy reasoning result matrix ΔU s Perform weighted synthesis in spatial direction to obtain the compensation matrix Δq of heat flow s ; Step 8: According to the formula Matrix of current guesses for heat flux Refresh to obtain heat flow estimates that take both time and space weighting into account Step 9: Set n=n+1; return to step 2 and continue the loop until the shutdown criterion is met.

2. A transient distribution heat load alternating direction weighted fuzzy inversion method according to claim 1, characterized in that: In step 1, the mathematical description of the heat transfer model is as follows: Among them, x = (x, y) is the spatial coordinate vector; T is the temperature; ρ is the density; c p is the specific heat capacity; λ is the thermal conductivity; h is the convective heat transfer coefficient; T amb is the ambient temperature; T0 is the initial temperature distribution; t f is the simulation end time; In the heat transfer model (x m ,y m ) Arrange N m A temperature sensor is located at the measuring point m. k The temperature measured at the time is Y(x m ,y m ,t k )(m=1,2,…,N m ), the unknown boundary heat load distribution q(x,t) is inverted, and the heat load is discretely expressed as: 。 3. A transient distribution heat load alternating direction weighted fuzzy inversion method according to claim 2, characterized in that: In step 2, in The boundary heat flow is expressed as When , the calculated temperature matrix is ​​obtained according to the heat transfer direct problem calculation; The criteria for stopping the iteration are as follows: In the formula, ε is a given small positive number. According to the deviation principle, ε=N m N t σ 2 ; When the measurement error is not considered (i.e. when σ=0), take ε=0.001℃ 2 , σ is an estimate of the standard deviation of the measurement error.

4. A transient distribution heat load alternating direction weighted fuzzy inversion method according to claim 1, characterized in that: In step 3, the fuzzy inference module FIM uses N m ×N t Decentralized Fuzzy Inference Unit FIU i (i=1,2,…,N m ×N t ), used to realize the temperature error matrix by any element e i To any element Δu of the fuzzy inference result matrix i the reasoning process; Take e i and Δu i The fuzzy set domains are [-p e ,p e ] and [-p u ,p u ], e i and Δu i The domain is divided into 7 fuzzy sets {A1, A2, ..., A7} and {B1, B2, ..., B7}. The language values ​​corresponding to these fuzzy sets are: NB (negative large), NM (negative medium), NS (negative small), ZE (zero), PS (positive small), PM (positive medium), PB (positive large). The triangular membership function is used to determine the membership of each fuzzy set. FIU i The fuzzy inference rules are as follows: N m ×N t The inference results form the inference result matrix, that is 。 5. The method for weighted fuzzy inversion of transient distributed heat load alternating direction according to claim 1 is characterized in that: In step 4, the time-direction weighted synthesis is constructed using the chi-square distribution model; The chi-square distribution probability density function is: where n f is the degree of freedom of the chi-square distribution, which is a positive integer; adjusting the chi-square distribution degree of freedom can change the response mode of temperature to heat flow; t k =kdt(k=1,2...,N t ), is the discrete time; Γ(x) is the gamma function, that is, The time weighted coefficient model based on the chi-square distribution is: After normalization, the weighted coefficient is obtained: i=1,2,...,N t ; j = 1, 2, ..., N t ; When i<j, That is, the time weighting matrix Λ t is a lower triangular matrix; Through the weighted comprehensive formula in the time domain [Δq t ′] Nm×Nt =[ΔU t ] Nm×Nt [Λ t ] Nt×Nt , generating the time domain inference compensation matrix. Since the number of inversion points in space is not equal to the number of measurement points, the matrix [Δq t ′] Nm×Nt Transformed into [Δq t ] Ne×Nt .

6. A transient distribution heat load alternating direction weighted fuzzy inversion method according to claim 1, characterized in that: Step 6 mainly includes: based on Solve the heat transfer forward problem model established in step 1 to obtain the temperature calculation value matrix at the measuring point And calculate the temperature error matrix as follows: 。 7. The method for weighted fuzzy inversion of transient distributed heat load alternating direction according to claim 1 is characterized in that: In step 7, E n+1 / 2 As the input of the fuzzy reasoning module FIM, fuzzy reasoning is performed, and the reasoning method is the same as step 4; Among them, the weighted synthesis of row space direction mainly includes: Spatial domain weighted comprehensive matrix Λ s Using a comprehensive weighted model based on normal distribution, the spatial weighted comprehensive matrix Λ s element for In the formula, θ>0 is the variance coefficient of the normal distribution, x i and x j are the positions of the inversion point and the measurement point in the x direction respectively.

Images