Estimation method, analysis method, program, and estimation device for estimating estimation object included in analysis

Abstract

The invention provides an estimation method capable of accurately estimating an estimation object included in an analysis condition in a short time. This estimation method is characterized by comprising: a measurement step for measuring measurement data at at least one measurement point; a first analysis step in which an analysis condition including the estimation object is used to perform analysis to obtain an analysis result; a second analysis step in which a value obtained by applying a disturbance to the value to be estimated used in the first analysis step is set as the value to be estimated, and analysis is performed to obtain an analysis result; and a calculation step for calculating a sensitivity matrix on the basis of the analysis result of the first analysis step and the analysis result of the second analysis step, calculating the measurement data of the measurement point or the error between the data calculated from the measurement data and the analysis result of the first analysis step, and calculating the amount to change the value to be estimated from the sensitivity matrix and the error.

Description

Technical Field

[0001] The present invention relates to an estimating method, an analyzing method, a program, and an estimating device for estimating an estimated object included in an analyzing condition. Background Art

[0002] The temperature field, displacement field, stress field, etc. can be analyzed by using the finite element method (FEM) (for example, refer to Non-Patent Document 1).

[0003] In addition, it is known that in a system having an input and an output, there is a sensitivity analysis for examining the ratio of a change in output relative to a change in input (for example, refer to Non-Patent Document 2).

[0004] Prior art documents

[0005] Non-patent documents

[0006] Non-patent document 1: Journal of the Welding Society Vol. 80 (2011) 519-522

[0007] Non-patent document 2: Journal of the Meteorological Society of Japan, "Weather", Vol. 59 (2012), No. 11, P49-51 Summary of the invention

[0008] Problems to be solved by the present invention

[0009] When performing FEM analysis, differential analysis, boundary element analysis, particle analysis, meshless analysis, analysis based on theoretical formulas, etc., various input information is required as analysis conditions. Among these input information, there are cases where it is necessary to determine by experiments. In order to determine the input information, trial and error is required, which takes time. For example, when performing FEM heat conduction analysis, it is necessary to determine the value of the standard deviation σ, an indicator representing the spread of the heat input distribution, through experiments.

[0010] The present invention has been made in view of such circumstances, and provides an estimating method capable of accurately estimating an estimation target included in an analysis condition in a short time.

[0011] Means of solving the problem

[0012] The present invention provides an estimation method, which is an estimation method for estimating at least one estimation object included in an analysis condition based on an analysis result and measurement data corresponding to a part of the analysis result or data calculated from the measurement data, and is characterized in that the method comprises: a measuring step, which measures the measurement data of at least one measurement point; a first analysis step, which uses the analysis condition including the estimation object to perform FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on a theoretical formula to obtain an analysis result; a second analysis step, which sets a value obtained by applying a disturbance to the value of the estimation object used in the first analysis step as the value of the estimation object, and performs FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on a theoretical formula The method comprises the steps of: analyzing the formula to obtain an analysis result; and a calculation step, which calculates a sensitivity matrix based on the analysis result of the first analysis step and the analysis result of the second analysis step, calculates the error between the measurement data of the measuring point or the data calculated based on the measurement data and the analysis result of the first analysis step, and calculates the amount of change in the value of the estimated object based on the sensitivity matrix and the error, wherein the first analysis step, the second analysis step and the calculation step are repeated until the error becomes smaller, and in the first analysis step for the first time, an initial value is set for the estimated object for analysis, and in the first analysis step after the second time, a value obtained by changing the value of the estimated object used in the previous first analysis step by only the amount of change calculated in the previous calculation step is set as the estimated object for analysis.

[0013] Effects of the Invention

[0014] According to the present invention, by repeating the first analysis step, the second analysis step, and the calculation step, it is possible to accurately estimate the estimation target included in the analysis conditions in a short time. This was made clear by experiments and analyses conducted by the inventors of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flowchart of an estimation method according to one embodiment of the present invention.

[0016] Figure 2 (a) is a diagram for explaining the standard deviation, and (b) is a diagram for explaining the time shift.

[0017] Figure 3 This is an explanatory diagram of a method for calculating the error between the measurement data and the analysis result of the first analysis step.

[0018] Figure 4 This is a diagram and mathematical formula for explaining the calculation method of the sensitivity matrix.

[0019] Figure 5is a mathematical formula including the sensitivity matrix.

[0020] Figure 6 This is a diagram for explaining a method of calculating the amount by which the value of the estimation target is changed based on the sensitivity matrix and the error.

[0021] Figure 7 It is a figure which shows the measurement position in a measurement step.

[0022] Figure 8 This is a graph showing the temperature history measured in the measurement step.

[0023] Fig. 9 It is a graph showing the change in the value of thermal efficiency with respect to the number of trials.

[0024] Fig.10 It is a graph showing the change in the value of the standard deviation with respect to the number of trials.

[0025] Fig.11 It is a graph showing the change of the time deviation amount with respect to the number of trials.

[0026] Fig.12 It is a graph showing the change of specific heat value with respect to the number of trials.

[0027] Fig.13 (a) to (e) are graphs comparing the analysis results of the heat conduction analysis of the thermal efficiency, standard deviation, time shift, and specific heat calculated using the estimation method of the present invention with the temperature history measured in the measurement step.

[0028] Fig.14 This is the distribution of out-of-plane displacement of the metal plate subjected to linear heating treatment.

[0029] Fig.15 (a) is a graph showing the initial value of the yield stress used in the analysis conditions, and (b) is a graph showing the estimated value of the yield stress calculated using the estimation method of the present invention.

[0030] Fig.16 (a) is an analytical model of the analytical results of FEM thermoelastic-plastic analysis under analytical conditions including estimated values ​​calculated using the estimation method of the present invention, and (b)-(d) are graphs comparing the analytical results (out-of-plane displacement distribution) at the dashed lines A-A', B-B', and C-C' with the measured out-of-plane displacement distribution at the dashed lines A-A', B-B', and C-C'.

[0031] Fig.17 It is an analytical model of complex structures.

[0032] Fig.18Graph showing the out-of-plane displacement caused by assembly welding at the dashed line AA' of the bottom plate of the complex structure.

[0033] Fig.19 This is a graph showing the transition of torch speeds v0 and v1 to be estimated.

[0034] Fig. 20 This is a graph showing the hypothetical out-of-plane displacement calculated using the estimated torch speed and the measured value of the out-of-plane displacement after assembly welding of an actual machine. DETAILED DESCRIPTION

[0035] The estimation method of the present invention is an estimation method for estimating at least one estimation target included in an analysis condition based on an analysis result and measurement data corresponding to a part of the analysis result or data calculated based on the measurement data.

[0036] The estimation method of the present invention includes: a measuring step of measuring the measurement data of at least one measurement point; a first analysis step of performing FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on a theoretical formula using an analysis condition including the estimation object to obtain an analysis result; a second analysis step of setting a value obtained by applying a disturbance to the value of the estimation object used in the first analysis step as the value of the estimation object, performing FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on a theoretical formula to obtain an analysis result; and a calculation step of calculating a sensitivity matrix based on the analysis result of the first analysis step and the analysis result of the second analysis step, calculating an error between the measurement data of the measurement point or data calculated based on the measurement data and the analysis result of the first analysis step, and calculating an amount of change in the value of the estimation object based on the sensitivity matrix and the error.

[0037] In the estimation method of the present invention, the first analysis step, the second analysis step and the calculation step are repeated until the error becomes smaller. In the first analysis step for the first time, an initial value is set for the estimation object for analysis. In the first analysis step after the second time, the value of the estimation object used in the previous first analysis step is changed only by the amount of change calculated in the previous calculation step, and is set as the estimation object for analysis.

[0038] Preferably, the analysis conditions include a plurality of the estimated objects, the second analysis step is a step of setting a value to which a disturbance is applied to one of the plurality of the estimated objects as the value of the estimated object for all types of the estimated objects, performing FEM analysis, differential method analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on theoretical formulas to obtain analysis results, and the calculation step is a step of calculating the amount by which the values ​​of the various types of estimated objects are changed.

[0039] Preferably, the measurement data is time series data.

[0040] Preferably, the measuring step is a step of measuring the measurement data of multiple measurement points, and the calculation step is a step of calculating a sensitivity matrix based on the analysis results of the first analysis step and the analysis results of the second analysis step, calculating the measurement data of each measurement point or the error between the data calculated based on the measurement data and the analysis result of the first analysis step, and calculating the amount of change in the value of the estimated object based on the sensitivity matrix and the error.

[0041] Preferably, the error is a k-th power error.

[0042] Preferably, in the calculation step, a parameter of the analysis result and a parameter of the estimation target are weighted to calculate an amount to change a value of the estimation target.

[0043] Preferably, the first analysis step and the second analysis step are steps of performing heat conduction analysis, and the measuring step is a step of measuring a temperature change at the measuring point when the target object is heated.

[0044] Preferably, the estimated object includes the thermal efficiency of the heat source, the heat transfer rate, the standard deviation of the heat input distribution, the Goldak heat source parameter, the offset of the heat input distribution in the time direction, the specific heat or heat capacity of the object, the density of the object, the thermal conductivity of the object and at least one of the actual coordinates (x, y, z) of the measuring point.

[0045] Preferably, the first analysis step and the second analysis step are steps for performing thermo-elastoplastic analysis, and the estimated objects include at least one of the parameters related to the construction conditions, such as Young's modulus, Poisson's ratio, yield stress, linear expansion coefficient, annealing temperature, processing solidification parameters, creep constant, time offset variable, space offset variable, mechanical melting temperature and heat input rate of linear heating.

[0046] Preferably, the measuring step is a step of measuring a displacement of the measuring point caused by heating or cooling the target object.

[0047] Preferably, the first analysis step and the second analysis step are steps of performing electric field analysis, and the estimation object includes at least one of contact resistance, conductivity, time offset variable and space offset variable.

[0048] The present invention also provides an analytical method, which uses analytical conditions including the value of the estimated object estimated by the estimation method of the present invention to perform FEM analysis, differential method analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on theoretical formulas.

[0049] Preferably, the analysis method of the present invention includes the step of performing heat conduction analysis using analysis conditions including the value of the estimated object estimated according to the estimation method of the present invention, and the output of the heat conduction analysis includes at least one of the temperature history, temperature distribution, penetration shape parameter and molten pool shape parameter.

[0050] Preferably, the analysis method of the present invention includes the step of performing thermoelastic analysis using analysis conditions including the value of the estimated object estimated according to the estimation method of the present invention, and the output of the thermoelastic analysis includes at least one of displacement history, displacement distribution, stress history, stress distribution, strain history, strain distribution and residual stress distribution.

[0051] Preferably, the analysis method of the present invention includes the step of performing electric field analysis using analysis conditions including the value of the estimated object estimated according to the estimation method of the present invention, and the output of the electric field analysis includes at least one of the history of heat generation, the distribution of heat generation, the history of current density, the current density distribution, the history of potential and the potential distribution.

[0052] The present invention also provides a program for causing a computer to execute an estimating method for estimating at least one estimation object included in an analysis condition based on an analysis result and measurement data corresponding to a portion of the analysis result or data calculated based on the measurement data.

[0053] The present invention also provides an estimation device having a control unit, wherein the control unit is configured to execute an estimation method for estimating at least one estimation object included in an analysis condition based on an analysis result and measurement data corresponding to a part of the analysis result or data calculated based on the measurement data.

[0054] Preferably, the control unit is configured to send at least a part of the analysis condition to the outside of the estimation device and receive at least a part of the analysis result from the outside of the estimation device in the first analysis step or the second analysis step. The outside of the estimation device is, for example, a server on the Internet.

[0055] Hereinafter, one embodiment of the present invention will be described using the drawings. The configurations shown in the drawings and the following description are merely exemplary, and the scope of the present invention is not limited to the configurations shown in the drawings and the following description.

[0056] Figure 1 This is a flowchart of the estimation method according to the present embodiment.

[0057] The estimation method of the present embodiment can estimate at least one estimation object included in the analysis condition based on the analysis result and the measurement data corresponding to a part of the analysis result or the data calculated based on the measurement data. The estimation method of the present embodiment can be executed by a computer. In addition, the estimation method of the present embodiment can be executed by an estimation device having a control unit. The control unit can, for example, have a processing unit (e.g., a CPU), a RAM, a storage device (e.g., an HDD), a network controller, a video controller, etc. In addition, the computer or the estimation device can be connected to the Internet. The computer or the estimation device can perform analysis such as FEM analysis, or can receive online the analysis results of FEM analysis etc. performed on a server on the Internet.

[0058] The analysis conditions are the analysis conditions of FEM analysis (finite element method analysis), difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on theoretical formulas, and the analysis results are the analysis results of FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on theoretical formulas.

[0059] For example, heat conduction analysis, elasticity analysis, elastoplastic analysis, thermoelastoplastic analysis, thermal stress analysis, vibration analysis, electric field analysis, spot welding analysis, electromagnetic field analysis, etc. can be performed through FEM analysis, differential method analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on theoretical formulas.

[0060] The target object of FEM analysis, differential analysis, boundary element analysis, particle analysis, meshless analysis or analysis based on theoretical formulas is, for example, a metal plate, a T-joint, a lap joint, a multi-layer welded joint, etc. The target object may be a solid.

[0061] The heat source model of FEM analysis, differential analysis, boundary element analysis, particle analysis, meshless analysis or analysis based on theoretical formulas, for example, is a heat source model for gas heating, gas welding, gas cutting, arc welding, laser welding, resistance welding, spraying, etc. In addition, the heat source model may be a heat source model for heating any part of a solid object (object) as an object.

[0062] For example, the time series data of the temperature distribution of the target object (for example, a metal plate), the temperature history, the penetration shape parameters (penetration width B a and penetration depth B b When at least one of the melt pool shape parameters (thickness, width, length of the melt pool) and the melt pool shape parameters (analysis result Y) is obtained, the analysis model divided into multiple elements (meshes) corresponding to the shape and size of the target object, the heat source model (thermal efficiency, heat transfer rate, etc.), material constants (specific heat, heat capacity, density, thermal conductivity, etc.), heat source parameters (standard deviation of normal distribution (heat input distribution), Goldak heat source parameters), time offset variables, space offset variables (coordinate values ​​x, y, z), linear heating heat input speed and other parameters related to the construction conditions are analyzed as analysis conditions X.

[0063] For example, when calculating at least one of the displacement distribution, displacement history, stress history, stress distribution, strain history, strain distribution and residual stress distribution of an object (e.g., a metal plate) through thermoelastic-plastic analysis (analysis result Y), an analysis model divided into a plurality of elements (meshes) corresponding to the shape and size of the object, time series data of temperature distribution, material constants (Young's modulus, Poisson's ratio, yield stress, linear expansion coefficient, creep constant, mechanical melting temperature, annealing temperature, processing solidification parameters, etc.), time offset variables, spatial offset variables (coordinate values ​​x, y, z), parameters related to construction conditions such as heat input rate of linear heating, etc. are analyzed as analysis conditions X.

[0064] For example, when at least one of the heating history, heat generation distribution, current density history, current density distribution, potential history and potential distribution (analysis result Y) is calculated through electric field analysis (such as spot welding analysis), contact resistance, conductivity, time offset variables, spatial offset variables (coordinate values ​​x, y, z), etc. are used as analysis conditions X for analysis.

[0065] The calculations based on these analyses can be expressed as Y=F(X).

[0066] The estimation object is information to be estimated among various input information included in the analysis condition X. The estimation object may include a variety of information. In the following description, the estimation object is also referred to as a parameter. The estimation object may also be an estimation object for each temperature (for example, each temperature at intervals of 10° C.).

[0067] Examples of the types of estimation targets include heat transfer parameters such as specific heat and thermal conductivity (estimation of temperature dependence), and mechanical parameters such as Young's modulus, yield stress, and linear expansion coefficient (estimation of temperature dependence).

[0068] For example, in the case of calculating time series data (analysis result Y) of the temperature distribution of an object (e.g., a metal plate) through heat conduction analysis, the estimated object (parameters) may include, for example, the thermal efficiency η of the heat source, the standard deviation σ of the heat input distribution, the offset t0 of the heat input distribution in the time direction, the specific heat c of the object, the actual coordinates (x, y, z) of the measuring point, the heat transfer coefficient, the Goldak heat source parameter, the heat capacity of the object, the density of the object, and at least one of the thermal conductivity of the object.

[0069] Figure 2 (a) is an illustration of the standard deviation σ of the heat input distribution, Figure 2 (b) is an explanatory diagram of the time-direction shift amount t0 of the heat input distribution.

[0070] The heat input distribution of the heat source model included in the analytical conditions can be assumed to be a normal distribution (or its superposition), with a standard deviation such as Figure 2 (a) shows the expansion of the normal distribution.

[0071] like Figure 2 As shown in (b), the time-direction offset t0 of the heat input distribution represents the time-direction deviation between the temperature change (temperature time-series data) measured in the measurement step S1 and the temperature change (temperature time-series data) calculated by the FEM heat conduction analysis.

[0072] For example, in the case of thermo-elastoplastic analysis, the estimated objects (parameters) may include at least one of the parameters related to the construction conditions, such as Young's modulus, Poisson's ratio, yield stress, linear expansion coefficient, annealing temperature, processing solidification parameters, creep constant, time offset variable, space offset variable, mechanical melting temperature and heat input rate of linear heating.

[0073] For example, in the case of electric field analysis, the estimated object (parameter) may include at least one of contact resistance, conductivity, time offset variable, and space offset variable.

[0074] The measurement data is data corresponding to a part of the analysis result or data used to calculate the data, and is measured in the measurement step S1. The measurement data is measurement data of the measurement point of the target object corresponding to the measurement point set in the analysis model. In addition, there may be an error between the measurement point of the analysis model and the measurement point in the measurement step S1. In addition, the analysis condition may also include the coordinates of the actual measurement point as the estimation object (parameter). The measurement data or the data calculated based on the measurement data may also be time series data.

[0075] The measuring step S1 may also include a step of calculating data corresponding to a part of the analysis result based on the measurement data. For example, out-of-plane displacement data or target data may be calculated based on the measurement data, and the out-of-plane displacement data or target data may be used instead of the measurement data of steps S3 and S9. The out-of-plane displacement data or target data is data corresponding to a part of the analysis result.

[0076] The estimation method of the present embodiment includes: a measurement step S1, which measures measurement data of at least one measurement point; a first analysis step S2, S8, which uses an analysis condition including an estimation object to perform FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on a theoretical formula to obtain an analysis result; a second analysis step S5, which sets a value obtained by applying a disturbance to the value of the estimation object used in the first analysis step as the value of the estimation object, performs FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on a theoretical formula to obtain an analysis result; and a calculation step, which calculates a sensitivity matrix based on the analysis result of the first analysis step and the analysis result of the second analysis step (step S6), calculates an error between the measurement data of the measurement point or data calculated based on the measurement data and the analysis result of the first analysis step (steps S3, S9), and calculates an amount of change in the value of the estimation object based on the sensitivity matrix and the error (step S7).

[0077] In the estimation method of this embodiment, the first analysis step, the second analysis step, and the calculation step are repeated until the error between the measurement data of the measurement point or the data calculated from the measurement data and the analysis result of the first analysis step becomes small.

[0078] In the first analysis step (step S2), the estimation object (parameter) is analyzed by setting an initial value.

[0079] In the first analysis step (step S8) after the second time, the value (parameter) of the estimated object used in the previous first analysis step (step S2 or S8) and the value obtained by changing only the amount of change calculated in the previous calculation step (step S7) are set as the estimated object (parameter) for analysis.

[0080] Next, follow Figure 1 In addition, Figure 1 Steps S1-S11 shown in the flowchart may be switched in a different order except when logical matching cannot be achieved.

[0081] First, in the measurement step S1, measurement data (e.g., time series data of temperature distribution, displacement data, etc.) of at least one measurement point of the target object (e.g., a metal plate) is measured. The measurement step S1 may include a step of calculating data corresponding to a part of the analysis result (e.g., out-of-plane displacement data, target displacement data, etc.) based on the measurement data.

[0082] For example, when the time series data (analysis result Y) of the temperature distribution of the target object (analysis model corresponding to the target object) is calculated by heat conduction analysis, heat input treatment is actually performed on the target object, and the time series data of the temperature of the measurement point of the target object when the heat input treatment is performed is measured. When a plurality of measurement points are set, the time series data of the temperature of each measurement point of the target object when the heat input treatment is performed is measured. The measurement point can be set at a location where the temperature change is large.

[0083] For example, when the displacement of the target object (analysis model corresponding to the target object) is calculated by thermoelastic analysis, heat input is actually performed on the target object to measure the displacement of the measurement point of the target object. When multiple measurement points are set, the displacement of each measurement point is measured.

[0084] The heat input process is, for example, gas heating, gas welding, gas cutting, arc welding, laser welding, resistance welding, spraying, etc.

[0085] For example, when calculating the heat generation distribution of the object by electric field analysis (spot welding analysis), voltage is applied to the object to measure the time series data of the temperature of the measurement point of the object. When multiple measurement points are set, the time series data of the temperature of each measurement point is measured.

[0086] In step S2, the estimated object (parameter) included in the analysis condition X is set to an initial value, and FEM analysis, differential analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on a theoretical formula is performed to calculate the analysis result. These analyses can be performed by a server on the Internet. In this case, the analysis condition X is sent to the server, and the analysis result Y is received from the server.

[0087] In addition, through these analyses, it is possible to obtain analysis results corresponding to the measurement data measured in the measurement step S1 or the data calculated based on the measurement data. For example, when the time series data of the temperature of the measurement point is measured in step S1, the analysis results also include the time series data of the temperature of the measurement point (there may be an error between the measurement point of the measurement data and the measurement point of the analysis result).

[0088] In order to perform FEM analysis, first, an analysis model corresponding to the shape and size of the target object (for example, a metal plate) is generated. The analysis model has a plurality of elements (meshes) divided by lines connecting node coordinates.

[0089] The generated analytical model and analytical conditions including the estimated object are used to perform FEM analysis and calculate the analytical results. In step S2, appropriate initial values ​​are set for the estimated object (parameters) and FEM analysis is performed.

[0090] For example, when the time series data (analysis result Y) of the temperature distribution of the target object (analysis model corresponding to the target object) is calculated by FEM heat conduction analysis, the heat source model corresponding to the heat input processing of the measurement step S1 is set as the analysis condition. Unknown heat source parameters in the heat source model (for example, standard deviation σ and thermal efficiency η of the heat source) are set as estimation objects (parameters), and in step S2, appropriate initial values ​​can be set for each estimation object. In addition, the offset amount in the time direction of the heat input distribution, the specific heat of the target object, the actual coordinates of the measurement point, etc. are set as estimation objects (parameters), and appropriate initial values ​​can be set for each estimation object.

[0091] When the displacement of the target object (analysis model corresponding to the target object) is calculated by FEM thermoelastic analysis, for example, Young's modulus may be set as an estimation target, and an appropriate initial value may be set for the Young's modulus in step S2.

[0092] In step S3, an error is calculated between the measurement data of the measurement point measured in the measurement step (step S1) and the analysis result Y corresponding to the measurement point among the analysis results calculated in step S2. This error is a residual error or a k-th power error.

[0093] Figure 3 This is an explanatory diagram of the calculation method of the k-th square error. i , k) represents the deviation between the measurement data of the measurement point measured in the measurement step of step S1 and the analysis result calculated in step S2. Figure 3 The “△” in the graph shown is the time series data of the observation value measured in the measurement step (step S1). Figure 3 The curve in the graph shown is the analysis result calculated in step S2. Figure 3 The "X" in is the value to be estimated, and the "i" indicates the trial number.

[0094] In step S4, it is determined whether the error calculated in step S3 is sufficiently small. If the error is sufficiently small, the process proceeds to step S11, and the initial value set in step S2 is determined as the estimated value.

[0095] When the error is large, steps S5 and S6 are performed.

[0096] In steps S5 and S6, disturbances are applied to the estimated object (parameters), and FEM analysis, differential analysis, boundary element analysis, particle analysis, meshless analysis, or analysis based on theoretical formulas are performed to calculate the analysis results, and a sensitivity matrix is ​​generated based on the analysis results. These analyses can be performed through a server on the Internet.

[0097] The sensitivity matrix represents the influence of each estimation object (parameter) on the analysis result.

[0098] Figure 4 and Figure 5 It is a graph and mathematical formula used to explain the sensitivity matrix.

[0099] In step S5, for all estimated objects, a value to which a disturbance is applied to one estimated object included in a plurality of estimated objects (parameters) is set as the value of the estimated object, and FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on a theoretical formula is performed to obtain an analysis result. In addition, the analysis in step S5 may be the same analysis as step S2 except that the value to which the disturbance is applied is set as the value of the estimated object.

[0100] For example, Figure 4 Instructions for use include X1-X m A method for generating a sensitivity matrix when performing FEM heat conduction analysis under the analytical conditions of m estimation objects (parameters). In addition, the superscript i of X represents the number of the trial. n is the temperature time series data Y. δX is the disturbance added to X. δT is the difference between the temperature calculated in step S5 and the temperature calculated in step S2. α is the update ratio of the estimation object (parameter).

[0101] First, a disturbance is applied to X1 among the m parameters, and FEM heat conduction analysis is performed to calculate temperature time series data Y. Parameters other than X1 are set to the values ​​of the parameters used in step S2. The leftmost column of the sensitivity matrix is ​​generated based on the analysis results.

[0102] Next, a disturbance is applied to X2 among the m parameters, and FEM heat conduction analysis is performed to calculate the temperature time series data Y. Parameters other than X2 are set to the values ​​of the parameters used in step S2. The second column from the left end of the sensitivity matrix is ​​generated based on the analysis results.

[0103] Then, for X3-X m Perturbations are applied sequentially, FEM heat conduction analysis is performed, and columns of the sensitivity matrix are generated sequentially.

[0104] Figure 5 The sensitivity matrix is ​​described when FEM heat conduction analysis (Y=F(X)) is performed using analysis conditions including standard deviation σ, thermal efficiency η, time offset t0, and specific heat c as estimation target X (parameter). In addition, in the measurement step of step S1, time series data of temperature are measured at measurement point 1 and measurement point 2. Therefore, the sensitivity matrix is ​​generated using the analysis results (analysis results Y) at measurement point 1 and measurement point 2.

[0105] First, a disturbance is applied to σ among σ, η, t0, and c, and FEM heat conduction analysis is performed to calculate the temperature time series data Y. Parameters other than σ are set to the values ​​of the parameters used in step S2. The leftmost column of the sensitivity matrix is ​​generated based on the analysis results.

[0106] Next, a disturbance is applied to η among σ, η, t0, and c, and FEM heat conduction analysis is performed to calculate the time series data of temperature. The parameters other than η are set to the values ​​of the parameters used in step S2. The second column from the left end of the sensitivity matrix is ​​generated based on the analysis results.

[0107] Next, a disturbance is applied to t0 among σ, η, t0, and c, and FEM heat conduction analysis is performed to calculate the temperature time series data Y. Parameters other than t0 are set to the values ​​of the parameters used in step S2. The third column from the left end of the sensitivity matrix is ​​generated based on the analysis results.

[0108] Next, a disturbance is applied to c among σ, η, t0, and c, and FEM heat conduction analysis is performed to calculate the temperature time series data Y. Parameters other than c are set to the values ​​of the parameters used in step S2. The rightmost column of the sensitivity matrix is ​​generated based on the analysis results.

[0109] In this way, a sensitivity matrix can be generated.

[0110] In step S7 , the amount by which the value of the estimation target (parameter) is changed is calculated based on the sensitivity matrix generated in step S6 and the error calculated in step S3 .

[0111] Figure 6 These are mathematical formulas and explanatory diagrams for calculating the amount by which the value of the estimation object (parameter) is changed.

[0112] In the formula, X i is the value of the estimated object (parameter) used in the FEM analysis, differential analysis, boundary element analysis, particle analysis, meshless analysis, or analysis based on a theoretical formula in step S2 (i is the trial number), [[MW X ] T MW X ] -1 [MWX ] T W y R(X i , k) is the amount by which the value of the estimated object (parameter) changes, X i+1 is the value of the estimated object (parameter) used in the FEM analysis, differential analysis, boundary element analysis, particle analysis, meshless analysis, or analysis based on a theoretical formula in step S8. M is the sensitivity matrix calculated in step S6, R(X i , k) is the error calculated in step S3. X , W Y is the weighting matrix.

[0113] By using such an equation, the value of the estimation object (parameter) can be changed so that the error R(X i , k) becomes smaller, and the value (updated value of the estimated object) used in the FEM analysis, differential method analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on the theoretical formula in step S8 can be calculated.

[0114] In step S8, the value calculated in step S7 (the updated value of the estimated object (parameter)) is set in the estimated object (parameter) included in the analysis condition, and FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on a theoretical formula is performed to calculate the analysis result Y (Y = F (X)). In addition, the analysis in step S8 can be set to the same analysis as step S2 except that the value calculated in step S7 is set as the estimated object. In addition, these analyses can also be performed by a server on the Internet.

[0115] In step S9, the error between the measurement data of the measurement point measured in the measurement step of step S1 and the analysis result Y corresponding to the measurement point among the analysis results calculated in step S8 is calculated. "Calculation of error" is applicable unless there is any contradiction in the description of step S3.

[0116] In step S10, it is determined whether the error calculated in step S9 is sufficiently small. If the error is sufficiently small, step S11 is performed to determine the value of the estimation object (parameter) set in step S8 (the value of the estimation object calculated in step S7) as the estimated value.

[0117] If the error is large, the process returns to step S5 and steps S5 to S10 are performed again. In the above steps S5 to S7, although the analysis conditions, analysis results, and error calculated in step S3 of the FEM analysis, difference analysis, boundary element analysis, particle analysis, meshless analysis, or analysis based on a theoretical formula in step S2 are used, in the steps S5 to S7 performed again, the analysis conditions, analysis results, and error calculated in step S9 of the FEM analysis, difference analysis, boundary element analysis, particle analysis, meshless analysis, or analysis based on a theoretical formula in the previous step S8 are used instead of these to perform steps S5 to S7.

[0118] Steps S5-S10 are repeated until the error becomes small. By this repetition, the value of the estimation object (parameter) is updated so that the error becomes small. Therefore, by repeating steps S5-S10, the estimation accuracy of the estimation object is also improved.

[0119] In step S10 , if it is determined that the error calculated in step S9 is sufficiently small, the value of the estimation target (parameter) set in the previous step S8 (the value of the estimation target calculated in step S7 ) is determined as the estimated value.

[0120] In this way, the value of the estimation object (parameter) can be estimated with high accuracy.

[0121] By performing heat conduction analysis (FEM analysis, differential analysis, boundary element analysis, particle analysis, meshless analysis, or analysis based on a theoretical formula) using the values ​​estimated by the estimation method of this embodiment, it is also possible to calculate the time series data of the temperature distribution of the area other than the measurement point with high accuracy. Therefore, it is possible to obtain a high-accuracy temperature distribution / time history change that is expanded three-dimensionally. In addition, by this heat conduction analysis, it is also possible to calculate the penetration shape parameters (penetration width B a and penetration depth B b etc.) or melt pool shape parameters (thickness, width, length of the melt pool).

[0122] In addition, by performing thermo-elastoplastic analysis (FEM analysis, differential method analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on theoretical formulas) using the values ​​estimated by the estimation method of the present embodiment, it is possible to calculate with high precision the displacement distribution, displacement history, stress history, stress distribution, strain history, strain distribution, or residual stress distribution.

[0123] In addition, by using the values ​​estimated by the estimation method of this embodiment to perform electric field analysis (spot welding analysis) (FEM analysis, differential method analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on theoretical formulas), the heating history, heat generation distribution, current density history, current density distribution, potential history or potential distribution can be calculated with high precision.

[0124] Estimation of thermal efficiency, standard deviation, time offset and specific heat

[0125] Using the above-mentioned estimation method, thermal efficiency, standard deviation, time shift, and specific heat are estimated.

[0126] Five measuring points 1-5 are set in the center of a metal plate (square with a side of 400 mm and a thickness of 22 mm) made of general structural rolled steel SS400, and a thermocouple is installed at each measuring point. The detailed positions of measuring points 1-5 are as follows: Figure 7 At each measuring point, a thermocouple was inserted into a hole opened on the back of the metal plate, and the thermocouple was set at a position 2 mm away from the surface of the metal plate.

[0127] like Figure 7 As shown in FIG. 1 , the surface of the metal plate is heated from one end to the other end by gas along a straight line connecting the midpoint of one side of the metal plate and the midpoint of the side opposite to the one side, and the time series data of the temperature of the measuring points 1-5 is measured using the thermocouple (step S1). The heating conditions are: the gas flow rate of the fuel gas (city gas 13A) is 2.3m 3 / h, oxygen gas flow rate is 4.0m 3 / h, torch feed speed is 160mm / min, and torch distance is 35mm.

[0128] Figure 8 Indicates the temperature changes of the measured measurement points 1-5.

[0129] Next, an FEM heat conduction analysis is performed using an analytical model formed based on the shape and size of the above-mentioned metal plate (number of nodes: 128982, number of elements: 106860), a heat source model corresponding to the above-mentioned gas heating, and material constants of the general structural rolled steel SS400 as analytical conditions, and the analytical results (time series data of the temperatures of measuring points 1-5 (0.1-500sec is 5000 at intervals of 0.1 seconds)) are calculated (step S2). In the analytical conditions, thermal efficiency η, standard deviation σ, time offset t0 and specific heat c are set as estimated objects (parameters). The initial values ​​of the parameters are set to η=0.1, σ=10mm, t0=0sec, c=0.3kJ / kgK.

[0130] Then, follow Figure 1, steps S3 and S4 are performed, and steps S5 to S10 are repeated. The number of times (trial times) of FEM analysis (step S8) performed by updating the estimated target (parameter) is nine.

[0131] The estimated values ​​determined in step S11 are η=0.301, σ=31.4 mm, t0=-12.8 seconds, and c=0.487 kJ / kgK. Fig. 9 is a graph showing the change in the value of thermal efficiency η with respect to the number of trials. Fig.10 is a graph showing the change in the value of the standard deviation σ with respect to the number of trials. Fig.11 is a graph showing the change in the time offset t0 with respect to the number of trials. Fig.12 It is a graph showing the change in the value of specific heat c with respect to the number of trials.

[0132] As can be seen from these graphs, when the number of trials exceeds 4, the value of each estimation target (parameter) hardly changes.

[0133] Then, FEM heat conduction analysis is performed using the estimated values ​​determined in step S11 (η=0.301, σ=31.4 mm, t0=-12.8 seconds, c=0.487 kJ / kgK), and the analysis results (time series data of the temperatures at the measurement points 1-5) are calculated.

[0134] The analysis results are as follows Fig.13 (a)-(e) are shown. In addition, Fig.13 In (a) to (e), for comparison, the temperature changes (measurement data) at the measurement points 1 to 5 measured in step S1 are also shown.

[0135] from Fig.13 As can be seen from (a) to (e), the temperature changes at the measurement points 1 to 5 calculated by the FEM heat conduction analysis are substantially consistent with the temperature changes at the measurement points 1 to 5 actually measured. Therefore, it is considered that the values ​​of the estimation objects (parameters) can be estimated with high accuracy.

[0136] Estimation of Temperature Dependence of Yield Stress

[0137] Using the above-mentioned estimation method, the temperature dependence of the yield stress at a plurality of temperatures was estimated.

[0138] The surface of a metal plate (a square with a side of 400 mm and a thickness of 22 mm) made of general structural rolled steel SS400 was gas heated from one end to the other along a straight line connecting the midpoint of one side and the midpoint of the side opposite to the one side (with the gas heating being conducted along the straight line). Figure 7 The heating conditions are: the gas flow rate of the fuel gas (city gas 13A) is 2.3m 3 / h, oxygen gas flow rate is 4.0m 3 / h, the torch feed speed was set to 160 mm / min, and the torch distance was set to 35 mm. The shapes of the metal plate before and after the heat treatment were measured using a laser tracker, and the out-of-plane displacement of the metal plate due to the heat treatment was calculated (step S1). Fig.14 The calculated distribution of the out-of-plane displacement of the metal plate is shown. In steps S3 and S9, the out-of-plane displacement distribution is used instead of the measurement data.

[0139] Next, the FEM thermoelastic-plastic analysis is performed using the analytical model (number of nodes: 128982, number of elements: 106860) formed based on the shape and size of the metal plate, the heat input distribution calculated by the FEM heat conduction analysis, the yield stress at multiple temperatures (temperature dependence of the yield stress), etc. as analytical conditions, and the analytical results (out-of-plane displacement distribution) are calculated (step S2). In the analytical conditions, the initial values ​​of the yield stress at multiple temperatures are set to Fig.15 (a) shows the values ​​in the graph.

[0140] Then, follow Figure 1 , perform steps S3 and S4, and repeat steps S5-S10. The number of trials is 40.

[0141] Fig.15 The graph of (b) shows the estimated values ​​(yield stress at a plurality of temperatures, temperature dependence of yield stress) determined in step S11. The estimated value of yield stress at room temperature is 270 MPa.

[0142] Next, the FEM thermoelastic analysis is performed using the estimated values ​​determined in step S11, and the analysis results (out-of-plane displacement distribution) are calculated.

[0143] Fig.16 (a) shows the analytical model as the analytical result. Fig.16 In (b)-(d), it means Fig.16 The analysis results (out-of-plane displacement distribution) of the dashed lines AA', BB', CC' shown in (a) and the out-of-plane displacement distribution of the dashed lines AA', BB', CC' measured in step S1.

[0144] from Fig.16 As can be seen from (b) to (d), the out-of-plane displacement distribution calculated by the FEM thermoelastic analysis is substantially consistent with the actually measured out-of-plane displacement distribution. Therefore, it is considered that the value of each estimation object (parameter) can be estimated with high accuracy.

[0145] A tensile test was performed on a metal plate (square with a side of 400 mm, thickness: 22 mm) of general structural rolled steel SS400 to measure the yield stress at room temperature. The measured value of the yield stress at room temperature of the metal plate was 263 MPa. The estimated value of the yield stress at room temperature of the metal plate was 270 MPa, which can be considered to be able to be estimated with high accuracy.

[0146] Estimation of torch speed

[0147] Using the above estimation method, a linear heating method was used to eliminate the Fig.17 Estimation of torch speed in the reverse strain process of out-of-plane deformation generated in assembly welding of complex structures shown.

[0148] The out-of-plane displacement appears significantly Fig.18 The dotted line AA' of the complex structure in the curve diagram is set as the evaluation position, and the out-of-plane displacement that eliminates the out-of-plane displacement is calculated based on the distribution of the out-of-plane displacement at the dotted line AA', and the calculated out-of-plane displacement is used as the target deformation. Fig.18 The distribution of out-of-plane displacement at the dashed line AA' of the complex structure is shown.

[0149] exist Figure 1 In steps S3 and S9 of the flowchart, the target deformation is used instead of the measurement data.

[0150] generate Fig.17 Analytical model of the complex structure shown.

[0151] Set the number of joints in the analytical model to 669990, the number of elements to 610190, and the material of the metal plate to SS400. Figure 1 In steps S2, S5 and S8 of the flowchart of Fig.17 The position indicated by the arrow is used as a linear heating position to perform FEM thermo-elastic analysis, and the analysis result (out-of-plane displacement distribution at the dotted line AA') is calculated (step S2).

[0152] In the FEM thermo-elastoplastic analysis, the heat input of the front torch in the series heating, in which two torches are arranged at an interval of 55 mm, is set to 26250 W, and the heat input of the rear torch is set to 23750 W. The thermal efficiency η is set to 0.293 in both the front and rear torches, and the standard deviation σ when assuming a normal distribution of heat input is set to 29.3 mm in both the front and rear torches. Both the front and rear torches are heated at a torch speed v0. When the heated length of the front torch reaches 602 mm, the torch speeds of the front and rear torches are changed from v0 to v1. The initial values ​​of the torch speeds v0 and v1 are set to 4.0 mm / sec to reduce the deviation between the target deformation at the evaluation position (dashed line A-A') and the out-of-plane displacement as the analysis result, and an analysis is performed. Figure 1 The torch speed is optimized as shown in the flowchart of FIG. X and W Y is the identity matrix.

[0153] Fig.19 The horizontal axis represents the number of repetitions of parameter updates, and the vertical axis represents the speed of the welding torch. Fig.19 It can be seen that the torch speeds v0 and v1 hardly change after the fifth time and converge to a constant value. In addition, the torch speeds v0 and v1 converge to 5.23 mm / sec and 3.59 mm / sec, which are taken as the optimal solutions and used as estimated values ​​(step S11).

[0154] Using the estimated torch speed (v0 = 5.23 mm / sec, v1 = 3.59 mm / sec), Fig.17 The position indicated by the arrow is used as a linear heating position for FEM thermoelastic analysis (forward analysis), and the analysis result (out-of-plane displacement distribution at the dotted line AA') is calculated.

[0155] Fig. 20 It represents the amount of deformation caused by assembly welding and the hypothetical out-of-plane displacement after assembly welding, which is the out-of-plane displacement measured during actual assembly welding, added to the out-of-plane displacement after linear heating at the evaluation position (dashed line A-A') using the estimated welding torch speed (v0=5.23mm / sec, v1=3.59mm / sec).

[0156] like Fig. 20 As shown, at the evaluation position (dashed line A-A'), it can be confirmed in the analysis that the out-of-plane displacement after assembly welding during linear heating under heating conditions including the estimated welding torch speed (v0=5.23mm / sec, v1=3.59mm / sec) is largely eliminated by assembly welding.

Claims

1. An estimation method, comprising estimating at least one estimation object included in an analysis condition based on an analysis result and measurement data corresponding to a part of the analysis result or data calculated based on the measurement data, characterized in that: The method includes: a measuring step of measuring the measurement data of at least one measuring point; A first analysis step, using the analysis conditions including the estimated object, performs FEM analysis, differential analysis, boundary element analysis, particle analysis, meshless analysis or analysis based on a theoretical formula to obtain an analysis result; a second analysis step, wherein a value obtained by applying a disturbance to the value of the estimated object used in the first analysis step is set as the value of the estimated object, and FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on a theoretical formula is performed to obtain an analysis result; and a calculation step of calculating a sensitivity matrix based on the analysis result of the first analysis step and the analysis result of the second analysis step, calculating an error between the measurement data of the measurement point or data calculated based on the measurement data and the analysis result of the first analysis step, and calculating an amount of change in the value of the estimation object based on the sensitivity matrix and the error, The first analytical step, the second analytical step and the calculating step are repeated until the error becomes smaller. In the first analysis step, the estimation object is analyzed by setting an initial value. In the second and subsequent first analysis steps, a value obtained by changing the value of the estimation target used in the previous first analysis step by the amount of change calculated in the previous calculation step is set as the estimation target for analysis.

2. The estimation method according to claim 1, wherein: The analysis condition includes a plurality of the estimation objects. The second analysis step is a step of performing FEM analysis, differential analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on a theoretical formula on all types of the estimated objects, setting a value to which a disturbance is applied to one of the multiple estimated objects as a value of the estimated object, and obtaining an analysis result. The calculation step is a step of calculating the amount of change in the value of each type of estimation target.

3. The estimation method according to claim 1, wherein: The measurement data is time series data.

4. The estimation method according to claim 1, wherein: The measuring step is a step of measuring the measurement data at a plurality of measurement points. The calculation step is a step of calculating a sensitivity matrix based on the analysis results of the first analysis step and the analysis results of the second analysis step, calculating the error between the measurement data of each measuring point or the data calculated based on the measurement data and the analysis result of the first analysis step, and calculating the amount of change in the value of the estimated object based on the sensitivity matrix and the error.

5. The estimation method according to claim 1, wherein: The error is a k-th power error.

6. The estimation method according to claim 1, wherein: In the calculation step, the parameter of the analysis result and the parameter of the estimation target are weighted to calculate the amount of change in the value of the estimation target.

7. The estimation method according to claim 1, wherein: The first analysis step and the second analysis step are steps for performing heat conduction analysis. The measuring step is a step of measuring a temperature change at the measuring point when the target object is heated.

8. The estimation method according to claim 7, wherein: The estimated object includes the thermal efficiency of the heat source, the heat transfer rate, the standard deviation of the heat input distribution, the Goldak heat source parameter, the offset of the heat input distribution in the time direction, the specific heat or heat capacity of the object, the density of the object, the thermal conductivity of the object and at least one of the actual coordinates (x, y, z) of the measuring point.

9. The estimation method according to claim 1, wherein: The first analysis step and the second analysis step are steps for performing thermoelastic-plastic analysis. The estimated object includes at least one of the parameters related to the construction conditions, such as Young's modulus, Poisson's ratio, yield stress, linear expansion coefficient, annealing temperature, processing solidification parameters, creep constant, time offset variable, space offset variable, mechanical melting temperature and heat input rate of linear heating.

10. The estimation method according to claim 9, wherein: The measuring step is a step of measuring the displacement of the measuring point caused by heating or cooling the target object.

11. The estimation method according to claim 1, wherein: The first analysis step and the second analysis step are steps for performing electric field analysis. The estimation object includes at least one of contact resistance, conductivity, a time offset variable, and a space offset variable.

12. A method of analysis, characterized in that: The method uses an analytical condition including a value of the estimation object estimated by the estimation method according to any one of claims 1 to 11 to perform FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on a theoretical formula.

13. A method of analysis, characterized in that: The method comprises the step of performing heat conduction analysis using an analysis condition including the value of the estimation object estimated by the estimation method according to claim 7 or 8, The output of the heat conduction analysis includes at least one of temperature history, temperature distribution, fusion depth shape parameters and molten pool shape parameters.

14. A program for causing a computer to execute an estimating method for estimating at least one estimation object included in an analysis condition based on an analysis result and measurement data corresponding to a part of the analysis result or data calculated based on the measurement data, characterized in that: The measurement data is data measured at at least one measurement point. The inference method includes: A first analysis step, which uses FEM analysis, differential analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on a theoretical formula to obtain an analysis result using an analysis condition including the estimated object; a second analysis step, which obtains an analysis result of FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on a theoretical formula by setting a value obtained by applying a disturbance to the value of the estimated object used in the first analysis step as the value of the estimated object; and a calculation step of calculating a sensitivity matrix based on the analysis result of the first analysis step and the analysis result of the second analysis step, calculating an error between the measurement data of the measurement point or data calculated based on the measurement data and the analysis result of the first analysis step, and calculating an amount of change in the value of the estimation object based on the sensitivity matrix and the error, The first analytical step, the second analytical step and the calculating step are repeated until the error becomes smaller. The FEM analysis, the difference method analysis, the boundary element method analysis, the particle method analysis, the meshless method analysis or the analysis based on the theoretical formula in the first analysis step is an analysis performed by setting initial values ​​for the estimation object. The FEM analysis, differential method analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on theoretical formula in the first analysis step after the second time is an analysis in which the value of the estimated object used in the previous first analysis step is changed only by the amount of change calculated in the previous calculation step, and the value obtained is set as the estimated object.

15. An estimation device comprising a control unit, characterized in that: The control unit is configured to execute an estimating method for estimating at least one estimation object included in the analysis condition based on the analysis result and measurement data corresponding to a part of the analysis result or data calculated based on the measurement data. The measurement data is data measured at at least one measurement point. The inference method includes: A first analysis step, which uses FEM analysis, differential analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on a theoretical formula to obtain an analysis result using an analysis condition including the estimated object; a second analysis step, which obtains an analysis result of FEM analysis, difference method analysis, boundary element method analysis, particle method analysis, meshless method analysis, or analysis based on a theoretical formula by setting a value obtained by applying a disturbance to the value of the estimated object used in the first analysis step as the value of the estimated object; and a calculation step of calculating a sensitivity matrix based on the analysis result of the first analysis step and the analysis result of the second analysis step, calculating an error between the measurement data of the measurement point or data calculated based on the measurement data and the analysis result of the first analysis step, and calculating an amount of change in the value of the estimation object based on the sensitivity matrix and the error, The first analytical step, the second analytical step and the calculating step are repeated until the error becomes smaller. The FEM analysis, the difference method analysis, the boundary element method analysis, the particle method analysis, the meshless method analysis or the analysis based on the theoretical formula in the first analysis step is an analysis performed by setting initial values ​​for the estimation object. The FEM analysis, differential method analysis, boundary element method analysis, particle method analysis, meshless method analysis or analysis based on theoretical formula in the first analysis step after the second time is an analysis in which the value of the estimated object used in the previous first analysis step is changed only by the amount of change calculated in the previous calculation step, and the value obtained is set as the estimated object.

16. The estimation device according to claim 15, wherein: The control unit is configured to transmit at least a part of the analysis conditions to the outside of the estimation device and receive at least a part of the analysis results from the outside of the estimation device in the first analysis step or the second analysis step.

Images