Vacuum high-pressure gas quenching furnace pipe reliability analysis method based on Kriging agent model

Through the method based on the Kriging agent model, the random variables and reliability indicators of vacuum high-pressure gas quenching furnace gas gallbladder were analyzed, and the furnace temperature uniformity problem caused by the uncertainty of furnace gas parameters was solved, the equipment reliability and product quality were improved, and the equipment was achieved. Effective failure probability calculation.

CN119004952APending Publication Date: 2024-11-22BEIJING UNIV OF TECH
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410983403.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-22
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The uncertainty of the furnace gas cylinder parameters of vacuum high-pressure gas quenching furnace affects the uniformity of the furnace temperature, resulting in low product processing quality and equipment reliability. An effective reliability analysis method is needed to improve the reliability of the furnace gas cylinder.

Method used

Using the method based on the Kriging agent model, the Kriging agent model is constructed by determining the random variables and reliability evaluation indicators of the furnace gallbladder, and the sample data is obtained using finite element simulation and Latin hypercube sampling to calculate the furnace temperature uniformity and failure probability, thereby improving the Reliability of furnace gallbladder.

Benefits of technology

It improves the reliability of the furnace gallbladder, ensures the uniformity of the furnace temperature, improves the processing quality of the product, provides reliable failure probability estimates and coefficient of variation, and ensures the effectiveness of the Kriging model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004952A_ABST
    Figure CN119004952A_ABST
Patent Text Reader

Abstract

The invention discloses a vacuum high-pressure gas quenching furnace pipe reliability analysis method based on a Kriging agent model, and belongs to the field of heat treatment equipment reliability analysis. Comprising the following steps: firstly, constructing a performance function by taking the inner diameter of a furnace pipe heating tape, the bandwidth, the tape thickness and the thickness of a heat shield as random variables and taking the furnace temperature uniformity as an evaluation index of the reliability of the furnace pipe; secondly, sample points are extracted through Latin hypercube sampling, and finite element simulation is carried out to obtain sample data for constructing a Kriging agent model; and finally, a Kriging agent model is constructed, and the furnace pipe failure probability is calculated. According to the method, the performance parameter furnace temperature uniformity of the furnace pipe serves as the reliability evaluation index of the furnace pipe, a new method is provided for reliability analysis of the furnace pipe, and the method is small in calculation scale and high in speed and has very high engineering significance in reliability analysis of the furnace pipe of the vacuum high-pressure gas quenching furnace.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a reliability analysis method of a vacuum high-pressure gas quenching furnace core (hereinafter referred to as core) based on a Kriging proxy model, and belongs to the field of reliability analysis of heat treatment equipment. Background Art

[0002] Compared with traditional quenching technology, vacuum high-pressure gas quenching furnace has the advantages of no oxidation on the workpiece surface, less distortion of the workpiece, and no environmental pollution. It has become the mainstream quenching equipment promoted in the world. In the future, with the rapid development of industrial technology, domestic heavy industry, light industry, automobile manufacturing industry, and shipbuilding industry will continue to develop rapidly, and the requirements for heat treatment technology and equipment for various metal parts will gradually increase. These applications and research on vacuum high-pressure gas quenching furnaces will bring huge opportunities and challenges. The furnace is a key component of the vacuum high-pressure gas quenching furnace, which is mainly composed of heat-generating belts and heat insulation screens. However, the uncertainty of the parameters of the heat-generating belts and heat insulation screens will directly affect the furnace temperature uniformity of the effective heating zone of the vacuum high-pressure gas quenching furnace, thereby affecting the processing quality of the product and causing the problem of low reliability of the vacuum high-pressure gas quenching furnace. Therefore, a furnace reliability analysis method is needed. It is very necessary to improve the reliability of the furnace by conducting reliability analysis on the furnace. Summary of the invention

[0003] The present invention provides a vacuum high-pressure gas quenching furnace furnace core reliability analysis method based on a Kriging proxy model, the purpose of which is to improve the furnace core reliability, thereby improving product processing quality, etc.

[0004] In order to achieve the above technical objectives, the present invention is implemented through the following technical solutions:

[0005] A reliability analysis method for a vacuum high pressure gas quenching furnace based on a Kriging proxy model includes:

[0006] Step 1) determine the furnace random variables and reliability evaluation indexes, and obtain Kriging proxy model sample data;

[0007] ①Determine the furnace random variables;

[0008] By analyzing the composition structure and main functions of the furnace, the random variables of the furnace are determined to be the inner diameter of the heating belt, the width of the heating belt, the thickness of the heating belt and the thickness of the heat insulation screen.

[0009] ②Determination of furnace reliability evaluation index and establishment of functional function;

[0010] Furnace temperature uniformity is an important indicator to measure the quality of the furnace. Therefore, furnace temperature uniformity is used as an evaluation index of furnace reliability, and the function is constructed as follows:

[0011] G(X)=Y(X)-|ΔT| (1)

[0012] Where X is a random vector, including random variables such as the inner diameter of the heating belt, bandwidth, belt thickness and heat shield thickness, X = (X1, X2, ..., X M ) T , M is the number of random variables; Y(·) is the furnace temperature uniformity of the furnace, obtained by finite element simulation; ΔT is the design threshold of furnace temperature uniformity; when G(·)>0, the furnace temperature uniformity meets the technical requirements, when G(·)<0, the furnace temperature uniformity does not meet the technical requirements, and G(·)=0 is the critical state, which is called the limit state equation.

[0013] ③Build Kriging proxy model sample data generation;

[0014] According to the random variable data characteristics (mean and standard deviation), Latin hypercube sampling is used to extract m groups of input samples to establish the MCS sample set N MCS , from sample N MCS n groups of samples were randomly selected, and steady-state thermal simulation was performed using ANSYS software to obtain the temperature field distribution of the furnace. Finally, the furnace temperature uniformity ΔT was calculated using the nine-point temperature measurement method. The input parameters (heat-generating belt inner diameter, belt width, belt thickness, and heat insulation screen thickness) and output data (furnace temperature uniformity) for constructing the Kriging proxy model were obtained.

[0015] Step 2) Kriging surrogate model construction and failure probability calculation;

[0016] ① Initial Kriging model construction;

[0017] Combined with the response data obtained by finite element simulation, the initial training sample set T is established, and the initial Kriging model is established according to formula (2):

[0018] g k (x) = f(x) T β+z(x) (2)

[0019] Where f(x)=[f1(x),f2(x),…,f m (x)] T is the m-dimensional regression polynomial basis function vector, β is the m-dimensional polynomial parameter vector; z(x) is a random process.

[0020] ② Update the Kriging model;

[0021] The MCS sample set N MCS The samples in are brought into the initial Kriging model, and the U learning function value U(x) of each sample is calculated. The sample point with the minimum U(x) value is taken as the next required sample point, and the model is updated.

[0022]

[0023] In the formula and are the sample prediction value and prediction variance of the Kriging model respectively.

[0024] x u =argminU(x) (4)

[0025] ③ Determine whether the learning function converges: Determine whether the adaptive learning process of the Kriging model converges. When min{U(x)≥2}, end the learning process and execute step ④. If min{U(x)<2}, calculate x u The response value g k (x u ), and {x u ,g k (x u )} is added to the training sample set T and used as a training sample to update the Kriging model.

[0026] ④According to the Kriging model established in step ③, the probability estimate is obtained That is, as the failure probability P f , and calculate the coefficient of variation.

[0027] The failure probability calculation formula is as follows:

[0028]

[0029] Where N G≤0 It is the number of sample points in the MCS sample set whose predicted response value of the Kriging model is less than or equal to 0.

[0030] The coefficient of variation is calculated by the following formula:

[0031]

[0032] When Cov(P f )<0.05, which means that P f The estimate is acceptable, and the final Kriging model is obtained. f )≥0.05, repeat step 2). BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a technical roadmap for implementing the method of the present invention.

[0034] Figure 2 This is a three-dimensional simulation model of the furnace. (a) is the overall modeling diagram of uniformity simulation; (b) is the detailed diagram of uniformity simulation.

[0035] Figure 3 Mesh division. (a) is the overall mesh division diagram; (b) is the overall mesh division cross-section diagram.

[0036] Figure 4 Schematic diagram of the nine-point temperature measurement method. (a) is the overall diagram of the temperature measuring ball position; (b) is the detailed diagram of the temperature measuring ball position; (c) is the detailed diagram of the temperature measuring ball position.

[0037] Figure 5 Schematic diagram of the effective heating area. DETAILED DESCRIPTION

[0038] The analysis method of the present disclosure is further explained below with reference to specific examples.

[0039] Figure 1 The flowchart of the disclosed method is as follows. In this example, a vacuum high-pressure gas quenching furnace furnace reliability analysis method based on the Kriging proxy model is provided, including:

[0040] 1) Determine the furnace random variables and reliability evaluation indicators, and obtain Kriging proxy model sample data

[0041] ①Determine the furnace random variable

[0042] The inner diameter, bandwidth, thickness of the heating belt and the thickness of the heat insulation screen of the furnace have a great influence on the uniformity of the furnace temperature. Therefore, the above parameters are listed as random variables. The distribution type and parameters of each random variable are shown in Table 1.

[0043] Table 1 Random variable distribution type and parameters of furnace core

[0044] random variable Distribution Type Mean Standard Deviation Heat-generating belt inner diameter R / mm Normal 715 3.5 Heating belt width L / mm Normal 65 2.2 Heat-generating belt thickness d / mm Normal 0.7 0.1 Heat insulation screen thickness D / mm Normal 0.5 0.05

[0045] ②Determination of furnace reliability evaluation index and establishment of functional function

[0046] The furnace temperature uniformity is used as the evaluation index of the furnace reliability, and the furnace function function is established according to the random variables determined in step ①:

[0047] G(X)=Y(X)-|ΔT| (7)

[0048] Where X is a random vector, including the random variables of the inner diameter of the heating belt, the width of the belt, the thickness of the belt and the thickness of the heat shield, X = (R, L, d, D) T ; Y(·) is the furnace temperature uniformity value calculated by finite element simulation data; according to the vacuum furnace grade, the furnace temperature uniformity design threshold ΔT=±3 is taken here;

[0049] ③Build Kriging proxy model sample data generation

[0050] Latin hypercube sampling is used to sample the random input variables of the furnace to obtain 10 6 Group sample points as sample set N MCS , then 12 groups of initial sample points are extracted from the sample set, as shown in Table 2.

[0051] Table 2 Initial sample points

[0052]

[0053] SolidWorks software was used to establish a simplified geometric model of the furnace. The established three-dimensional geometric model of the furnace is as follows: Figure 2 As shown in the figure, the furnace is meshed using ANSYS Mechanical APDL software. The meshing results are shown in the figure. Figure 3 shown.

[0054] ANSYS software is used for steady-state thermal simulation, and the furnace temperature uniformity is calculated by the nine-point temperature measurement method. The principle of the nine-point temperature measurement method is as follows:

[0055] The principle of the nine-point temperature measurement method is to place temperature measuring balls at the eight vertices and the geometric center of the rectangular effective heating area in the furnace. Figure 4 As shown, the effective heating area is Figure 5 As shown, after the furnace runs for a period of time, the temperature of each temperature measuring ball is extracted. Then, the maximum and minimum values ​​of the nine-point temperature are subtracted to obtain the furnace temperature uniformity of the effective heating area in the furnace.

[0056] Based on the above principle, the furnace temperature uniformity response value is calculated and recorded in Table 3.

[0057] Table 3 ANSYS simulation results (unit: °C)

[0058]

[0059] 2) Construct Kriging proxy model and calculate furnace failure probability

[0060] ① According to the sample points determined in step 1) and their corresponding simulation response values, the initial Kriging proxy model is constructed. According to the sample set T, the basis function and co-correlation function of the Kriging proxy model are selected, and the Kriging proxy model is:

[0061] g k (x) = f(x) T β+z(x) (8)

[0062] In this embodiment, the Kriging model basis function is selected as a zero-order polynomial, and the co-correlation function is a Gaussian correlation function.

[0063] ② The sample set N MCS 10 of 10 6 The sample points are brought into the initial Kriging model, and the next sample point is determined by U function (3), and the Kriging model is updated;

[0064] ③Use the U function convergence criterion to determine whether the adaptive learning process of the Kriging model converges. When min{U(x)≥2}, end the learning process and execute step ④. If min{U(x)<2}, calculate x u The response value g k (x u ), and {x u ,g k (x u )} is added to the training sample set T and used as a training sample to update the Kriging model.

[0065] ④ According to the Kriging model established in step ③, the failure probability of the furnace is calculated. The results are shown in Table 4.

[0066] Table 4 Calculated values ​​of failure probability

[0067] <![CDATA[N MCS ]]> <![CDATA[N G≤0 ]]> Failure probability Pf Coefficient of variation <![CDATA[10 6 ]]> 4538 0.0045 0.0149

[0068] The coefficient of variation Cov(P f )=0.0149<0.05, that is, the final Kriging proxy model is obtained, and the failure probability P of the furnace is calculated f =0.0045.

Claims

1. A reliability analysis method for a vacuum high-pressure gas quenching furnace based on a Kriging proxy model, characterized in that: The following steps are involved: Step 1) determine the furnace random variables and reliability evaluation indicators, and obtain sample data for building the Kriging proxy model; The inner diameter R, bandwidth L, thickness d of the heating belt of the furnace and the thickness D of the heat insulation screen are taken as random variables, and the furnace temperature uniformity ΔT is taken as the evaluation index of the furnace reliability, and the function function is constructed; according to the data characteristics of the random variables, Latin hypercube sampling LHS is used to extract sample points, and the steady-state thermal simulation of the furnace is carried out using ANSYS software to obtain the temperature field distribution of the furnace, calculate the furnace temperature uniformity response value, and obtain the sample data for constructing the Kriging proxy model; Step 2) constructing a Kriging proxy model and calculating the furnace failure probability; According to the n groups of sample values ​​and simulation response value data extracted in step 1), the initial Kriging proxy model is constructed, and the Kriging proxy model is optimized by adding sample points using the active learning function U function. Finally, the failure probability and coefficient of variation of the furnace are calculated using the Kriging proxy model.

2. The reliability analysis method of a vacuum high pressure gas quenching furnace based on a Kriging proxy model according to claim 1 is characterized in that: Step 1) is characterized in that furnace temperature uniformity is used as a furnace reliability analysis index, and a furnace function function is established: G(X)=Y(X)-|ΔT| Where X is a random vector, including random variables such as the inner diameter of the heating belt, bandwidth, belt thickness and heat shield thickness, X = (X1, X2, ..., X M ) T , M is the number of random variables; Y(·) is the furnace temperature uniformity of the furnace, obtained by finite element simulation; ΔT is the design threshold of furnace temperature uniformity; when G(·)>0, the furnace temperature uniformity meets the technical requirements, when G(·)<0, the furnace temperature uniformity does not meet the technical requirements, and G(·)=0 is the critical state, which is called the limit state equation.

3. The reliability analysis method of a vacuum high pressure gas quenching furnace based on a Kriging proxy model according to claim 1 is characterized in that: Taking furnace temperature uniformity as the furnace reliability analysis index, the Kriging proxy model sample data generation was constructed; According to the characteristics of random variable data, Latin hypercube sampling is used to extract m groups of input samples to establish the MCS sample set N MCS , from sample N MCS n groups of samples were randomly selected, and steady-state thermal simulation was performed using ANSYS software to obtain the temperature field distribution of the furnace. Finally, the furnace temperature uniformity ΔT was calculated using the nine-point temperature measurement method. The input parameters for constructing the Kriging proxy model, namely the inner diameter of the heat-generating belt, belt width, belt thickness and heat insulation screen thickness, and the output data, namely the furnace temperature uniformity, were obtained.

4. The method for analyzing the furnace core reliability of a vacuum high-pressure gas quenching furnace based on a Kriging proxy model according to claim 3 is characterized in that: Combined with the response data obtained by finite element simulation, the initial training sample set T is established, and the initial Kriging model is established according to formula (2): g k (x)=f(x) T β+z(x) (2) Where f(x)=[f1(x),f2(x),…,f m (x)] T is the m-dimensional regression polynomial basis function vector, β is the m-dimensional polynomial parameter vector; z(x) is a random process.

5. The method for analyzing the furnace core reliability of a vacuum high-pressure gas quenching furnace based on a Kriging proxy model according to claim 4 is characterized in that: The MCS sample set N MCS The samples in are brought into the initial Kriging model, and the U learning function value U(x) of each sample is calculated. The sample point with the minimum U(x) value is taken as the next required sample point, and the model is updated. In the formula and They are the sample prediction value and prediction variance of the Kriging model respectively; x u =argmiU(x) (40) 6. The method for analyzing the furnace core reliability of a vacuum high-pressure gas quenching furnace based on a Kriging proxy model according to claim 5 is characterized in that: Determine whether the learning function converges: Determine whether the adaptive learning process of the Kriging model converges. When min{U(x)≥2}, end the learning process and execute the next step; if min{U(x)<2}, calculate x u The response value g k (x u ), and {x u ,g k (x u )} is added to the training sample set T and used as a training sample to update the Kriging model.

7. The method for analyzing the furnace core reliability of a vacuum high-pressure gas quenching furnace based on a Kriging proxy model according to claim 6 is characterized in that: According to the established Kriging model, the probability estimate obtained is That is, as the failure probability P f , and calculate the coefficient of variation; The failure probability calculation formula is as follows: Where N G≤0 is the number of sample points in the MCS sample set whose predicted response value of the Kriging model is less than or equal to 0; The coefficient of variation is calculated by the following formula: When Cov(P f )<0.05, which means that P f The estimate is acceptable, and the final Kriging model is obtained; If Cov(P f )≥0.05, repeat step 2).

Citation Information

Cited By

  • Engineering structure reliability analysis method based on extended Kriging agent model

    CN120781663A