Deformation estimation device, deformation estimation system, and deformation estimation method

Abstract

This deformation estimation device (60) is provided with a control unit (50) for estimating deformation occurring in a structure (1) on the basis of a measurement value indicating deformation of a measurement surface on the surface of a structure (1). The control unit (50): predicts predicted deformation information indicating, by a probability distribution, deformation occurring on the measurement surface of the structure (1) in a plurality of boundary conditions applied to the structure (1); sets a first observation equation indicating the correlation between a first observation variable indicating the measurement value and the error of the measurement value by a probability distribution and the predicted deformation information serving as a first state variable; and estimates deformation occurring in the structure on the basis of the first observation equation.

Description

Deformation estimation device, deformation estimation system, and deformation estimation method

[0001] The present application relates to a deformation estimation device, a deformation estimation system, and a deformation estimation method.

[0002] For example, accurately estimating and understanding the deformation that occurs in various structures such as electric power equipment, steel towers, and steel bridges is important for assessing the life span of these structures and for performing maintenance and conservation work. However, prediction errors, measurement errors, etc. occur in the process of predicting and measuring the deformation of a structure. In order to reduce such errors and accurately estimate the deformation that occurs in a structure, a material testing machine as a deformation estimation device with the following configuration has been disclosed.

[0003] That is, in a conventional measurement method for measuring displacement and strain fields using a materials testing machine as a deformation estimation device, fixed noise in a grayscale image of a test piece captured by a video camera is probabilistically modeled based on darkness, and this probabilistic model of fixed noise is used to evaluate the uncertainty of the measurement of the center of gravity of a dot affixed to the surface of the test piece when a load is applied. Then, dot displacement, displacement field, and strain field are calculated probabilistically from the measurement results of the dot center of gravity using the law of uncertainty propagation. By performing measurements probabilistically using the dot center of gravity tracking method, uncertainty is propagated to each physical quantity derived by calculation from the observed value, and the reliability of the measurement value is presented to the user (see, for example, Patent Document 1).

[0004] International Publication No. WO2015 / 049757 (paragraphs

[0028] to

[0032] )

[0005] However, in such a conventional measurement method for measuring displacement fields and strain fields using a material testing machine as a deformation estimation device, it is not possible to sufficiently reduce errors that occur in the process of measuring a structure, and there are cases in which the deformation occurring in the structure cannot be accurately estimated. The present application discloses technology for solving the above-mentioned problems, and aims to provide a deformation estimation device, a deformation estimation system, and a deformation estimation method that can estimate the deformation occurring in a structure with high accuracy.

[0006] The deformation estimation device disclosed in the present application is a deformation estimation device equipped with a control unit that estimates deformation occurring in a structure based on measurement values ​​that indicate deformation on a measurement surface of the surface of the structure, wherein the control unit predicts predicted deformation information that indicates, by a probability distribution, deformation occurring on the measurement surface of the structure under a plurality of boundary conditions added to the structure, sets a first observation equation that indicates the correlation between the measurement values ​​and a first observation variable that indicates, by a probability distribution, an error in the measurement values, and the predicted deformation information as a first state variable, and estimates the deformation occurring in the structure based on the first observation equation. The deformation estimation system disclosed in the present application is a deformation estimation system comprising a deformation estimation device configured as described above, and an output unit that displays output from the control unit, wherein the control unit sets at least the longitudinal elastic modulus and Poisson's ratio of the structure as material characteristic values ​​of the structure in predicting the predicted deformation information, among the longitudinal elastic modulus, density, and Poisson's ratio, and performs a soundness assessment to determine whether the estimated deformation occurring in the structure and the probability distribution of its error exceed a first threshold value as an index value indicating soundness, and displays the assessment result on the output unit. The deformation estimation method disclosed in the present application is a deformation estimation method using the deformation estimation device configured as described above, wherein the control unit includes the steps of: predicting predicted deformation information that indicates, by a probability distribution, the deformation that will occur on the measurement surface of the structure under a plurality of boundary conditions added to the structure; and setting a first observation equation that indicates the correlation between the measurement value and a first observation variable (u*) that indicates, by a probability distribution, the error in the measurement value, and the predicted deformation information as a first state variable, and estimating the deformation that will occur in the structure based on the first observation equation.

[0007] The deformation estimation device and deformation estimation method disclosed herein provide a deformation estimation device and deformation estimation method that can estimate deformation occurring in a structure with high accuracy. Furthermore, the deformation estimation system disclosed herein estimates deformation occurring in a structure with high accuracy and then determines whether the deformation occurring is within an allowable value, allowing managers to appropriately evaluate the lifespan of the structure, and perform maintenance and conservation work, etc.

[0008] FIG. 6A is a block diagram showing a schematic configuration of a deformation estimation device according to embodiment 1 and a deformation estimation system including this deformation estimation device. FIG. 6B is a block diagram showing a schematic configuration of a measurement unit included in the deformation estimation device according to embodiment 2. FIG. 6C is a block diagram showing a schematic configuration of a measurement unit included in the deformation estimation device according to embodiment 2. FIG. 6D is a block diagram showing a schematic configuration of a measurement unit included in the measurement unit according to embodiment 2. 1 is a flow chart showing the flow of control processing of a model generation unit provided in a measurement unit according to embodiment 2. FIG. 2 is a flow chart showing the flow of control processing of a measurement unit according to embodiment 2. FIG. 3 is a conceptual diagram for explaining control processing of a measurement unit for a structure that is an inspection target according to embodiment 2. FIG. 4 is a block diagram showing a schematic configuration of a deformation estimation device according to embodiment 3 and a deformation estimation system including this deformation estimation device. FIG. 5 is a flow chart showing the flow of control processing of an estimating unit and a crack estimating unit provided in a deformation estimation device according to embodiment 3. FIG. 6 is a block diagram showing a schematic configuration of a deformation estimation system according to embodiment 4. FIG. 7 is a diagram showing an example of the hardware configuration of a control unit. FIG. 8 is a diagram showing an example of the hardware configuration of a control unit.

[0009] Embodiment 1. Preferred embodiments of a deformation estimation device, a deformation estimation system, and a deformation estimation method according to the present embodiment will be described below with reference to the drawings. Note that the same contents and corresponding parts are assigned the same reference numerals, and detailed descriptions thereof will be omitted. Similarly, in the following embodiments, redundant descriptions of components assigned the same reference numerals will be omitted.

[0010] FIG. 1 is a block diagram showing a schematic configuration of a deformation estimation device 60 according to embodiment 1 and a deformation estimation system 100 including the deformation estimation device 60. FIG. 2 is a flowchart showing the control process flow of a variable output unit 20 included in the deformation estimation device 60 shown in FIG. 1. FIG. 3 is a flowchart showing the control process flow of a model generation unit 30 included in the deformation estimation device 60 shown in FIG. 1. FIG. 4 is a flowchart showing the control process flow of an estimation unit 40 included in the deformation estimation device 60 shown in FIG. 1. FIG. 5 is a cutaway perspective view showing a portion of a structure 1 to be inspected. FIG. 6A is a diagram for explaining the process of setting boundary conditions at a cross section 1A of the structure shown in FIG. 5. FIG. 6B is a diagram for explaining the process of setting boundary conditions at a cross section 1B of the structure shown in FIG. 5.

[0011] 1 estimates deformation occurring in a structure 1 that is an object of inspection. The deformation estimation system 100 of this embodiment includes the deformation estimation device 60 and an output unit 5 that outputs the estimation results of the deformation estimation device 60, and notifies the estimated deformation via the output unit 5 to a manager who performs maintenance and preservation work.

[0012] The output unit 5 that notifies the administrator may have a display function such as a display, or may have an audio output function such as a speaker. The configuration of the output unit 5 is not limited as long as it has a function that can notify the administrator of the estimation result in the deformation estimation device 60.

[0013] 5 shows the surfaces 1S, 1A, 1B, 1C, and 1D of the structure 1, the surface 1S is the surface that can be observed directly, and the surfaces that cannot be observed directly are cross sections 1A, 1B, 1C, and 1D obtained by cutting the structure 1 parallel to the X-axis and Y-axis directions. As will be described below, the deformation estimation device 60 measures a measurement surface 1M, which is a set area within the observable surface 1S.

[0014] The configuration of the deformation estimation device 60 will be described. As shown in FIG. 1 , the deformation estimation device 60 includes a measurement unit 10 that measures deformation on a measurement surface 1M of a surface 1S of a structure 1 as a measurement value, and a control unit 50 that estimates deformation occurring in the structure 1 based on this measurement value. The detailed configuration of the measurement unit 10 is not limited, but for example, the measurement unit 10 may acquire the position of a point on the measurement surface 1M before deformation occurs, acquire the position of this point after deformation occurs, and calculate deformation from the displacement of the position. The control unit 50 includes a variable output unit 20, a model generation unit 30, and an estimation unit 40. The control processing in each of these units will be described below.

[0015] First, the control processing of the variable output unit 20 will be described based on the flow diagram of the control processing of the variable output unit 20 shown in Fig. 2. The variable output unit 20 generates a shape model that models the structure and shape of the structure 1, and sets multiple boundary conditions for this shape model. Here, the variable output unit 20 sets boundary conditions f together with load conditions that indicate the applied state of external forces, forced displacements, temperature changes, etc. applied to the structure 1, and constraint conditions that constrain the displacement of the structure 1 in each direction and rotational direction (step S21). In the process of step S21, the variable output unit 20 assumes various stresses, loads, and constraint conditions that will be applied to the structure 1, and sets multiple boundary conditions fi (i = 1, 2, 3, ... n).

[0016] Next, the variable output unit 20 selects a boundary condition fi from the plurality of boundary conditions fi that have been set (step S22). Next, the variable output unit 20 performs a numerical analysis to analyze the structure of the structure 1 based on the selected boundary condition fi (step S23). In this numerical analysis, the variable output unit 20 sets cross sections 1A and 1B shown in Fig. 5, which are surfaces that provide the boundary condition fi to the shape model. The variable output unit 20 performs a numerical analysis using the finite element method as follows, for example.

[0017] As shown in FIG. 6A , the variable output unit 20 divides the cross section 1A of the structure 1 into n units in the X-axis direction and m units in the Y-axis direction, forming a grid pattern, into a plurality of rectangular unit surfaces 2. Also, as shown in FIG. 6B , the variable output unit 20 divides the cross section 1B into c units in the X-axis direction and d units in the Y-axis direction, forming a grid pattern, into a plurality of rectangular unit surfaces 2. Note that, although the unit surfaces 2 are shown to be rectangular, they are not limited thereto and may be trapezoidal, for example. In this manner, if the intersection points of the parallel lines dividing the cross sections 1A and 1B into a plurality of unit surfaces 2, i.e., the vertices of each unit surface 2, are defined as nodes 2P, then each node 2P in the cross section 1A is represented by a number from (0,0) to (n,m), and each node 2P in the cross section 1B is represented by a number from (0,0) to (c,d).

[0018] The variable output unit 20 then assigns the selected boundary conditions fi to each node 2P of the cross sections 1A and 1B, and performs a numerical analysis to predict a deformation vector ui as a first predicted deformation that indicates the deformation that occurs on the measurement surface 1M under these boundary conditions fi.

[0019] In this numerical analysis, the variable output unit 20 sets and stores the boundary condition vector fIi shown in the following (Equation 1), which collectively represents the boundary conditions of each node 2P of these cross sections 1A and 1B as a vector.

[0020]

[0021] In setting the boundary condition vector fIi in the above equation (1), for example, in cross section 1A, the boundary condition applied to node 2P(0,0) is boundary condition fIi,A(0,0), and the boundary condition applied to node 2P(n,m) is boundary condition fIi,A(n,m). Also, in cross section 1B, the boundary condition applied to node 2P(0,0) is boundary condition fIi,B(0,0), and the boundary condition applied to node 2P(n,m) is boundary condition fIi,B(c,d).

[0022] Next, the variable output unit 20 saves the deformation vector ui of the measurement surface 1M predicted by the above numerical analysis (step S24). Note that the deformation vector ui may be a vector of displacement change amount indicating the difference before and after deformation of each node occurring on the measurement surface 1M.

[0023] Next, the variable output unit 20 performs the numerical analysis shown in steps S22 to S24 for all the boundary conditions fi set in step S21 and determines whether the deformation vector ui has been predicted (step S25). If there is a boundary condition fi that has not been numerically analyzed (step S25, NO), in step S25A, the number i of the boundary condition fi is incremented by 1, and the processes of steps S22 to S25 are repeated.

[0024] Once the deformation vectors ui (i = 1, 2 ... n) have been found for all the set boundary conditions fi (i = 1, 2 ... n) (step S25, YES), matrices F[fi1, ... fii, ... fIn] and U[u1, ... ui, ... un] are created from the boundary condition vectors fii and the deformation vectors ui, and matrix A shown in the following equation (2), which indicates the correlation between matrix F and matrix U, is derived (step S26). Note that matrix F in equation (2) is a pseudo-inverse matrix.

[0025]

[0026] Next, the variable output unit 20 derives basis vectors Ur of the linear space of the deformation field of the measurement surface 1M based on the matrix A. The basis vectors Ur may be subjected to singular value decomposition to reduce their rank so that the deformation field can be expressed with the minimum necessary information (step S27).

[0027] Next, the variable output unit 20 derives a coefficient vector fvr as an element constituting the linear space of the deformation field by superposing the basis vector Ur, and a probability distribution of uncertainty including a prediction error of the initial value of the coefficient vector fvr (step S28). Note that although the coefficient vector fvr is derived as a coefficient for superposing the basis vector Ur, it may be a scalar instead of a vector.

[0028] In parallel with step S28, the variable output unit 20 acquires a displacement vector u* as a measurement value indicating the actual deformation on the measurement surface 1M of the structure 1, which is input from the measurement unit 10 via the network. Then, the variable output unit 20 derives the probability distribution of the displacement vector u* and the uncertainty ε of this displacement vector u*, which includes measurement errors and the like (step S29).

[0029] Next, the variable output unit 20 outputs the derived basis vector Ur, the probability distribution including the uncertainty of the derived coefficient vector fvr, and the probability distribution including the uncertainty of the displacement vector u* as variables (step S30). Each output variable is input to the subsequent model generation unit 30.

[0030] Next, the control process of the model generation unit 30 will be described with reference to the flowchart of the control process of the model generation unit 30 shown in Fig. 3. First, the model generation unit 30 acquires the basis vector Ur as a variable input from the variable output unit 20 in the preceding stage, a probability distribution including the uncertainty of the coefficient vector fvr, and a probability distribution including the uncertainty ε of the displacement vector u* (step S31).

[0031] Next, the model generation unit 30 determines a first state equation that represents the time transition of the coefficient vector fvr, using the coefficient vector fvr and the uncertainty of its initial value (step S32), with the coefficient vector fvr as the first state variable. As an example, the state equation for the case where the deformation of the structure does not change over time is shown below (Equation 3). However, it is assumed that the uncertainty of the initial value of the coefficient vector fvr is included in {fvr}.

[0032]

[0033] Next, the model generation unit 30 determines a first observation equation that indicates the correlation between the displacement vector u* and the coefficient vector fvr (step S33), using the displacement vector u* as a first observation variable. The first observation equation is shown below (Equation 4).

[0034]

[0035] As shown in the above (Equation 4), the displacement vector u* as the first observation variable in the first observation equation is expressed by multiplying the coefficient vector fvr as the first state variable in the above first state equation by the basis vector Ur that constitutes the deformation field.

[0036] Next, the model generation unit 30 uses the first state equation and the first observation equation to generate a first state space model for estimating deformation occurring in the structure 1 (step S34). The generated first state space model is input to the subsequent estimation unit 40.

[0037] Next, the control process of the estimation unit 40 will be described with reference to the flowchart of the control process of the estimation unit 40 shown in Fig. 4. First, the estimation unit 40 acquires the first state space model input from the model generation unit 30 at the preceding stage (step S41).

[0038] Next, the estimation unit 40 obtains a first prior distribution based on the first state space model (step S42), that is, a probability distribution including uncertainty of the coefficient vector fvr, which is the first state variable in the first state equation.

[0039] Next, the estimation unit 40 updates the first prior distribution with the first likelihood function to derive the first posterior distribution (step S43). This first likelihood function is derived from the actual deformation of the measurement surface 1M measured by the measurement unit 10 (step S45), as a function indicating the likelihood distribution of the displacement vector u* and the uncertainty ε, which is the error of the displacement vector u*, from the measured values ​​(step S46).

[0040] Next, the estimation unit 40 determines whether the measurement of the measurement surface 1M in step S45 has been performed a set first number of times (step S44). If the measurement has not been performed the set first number of times (step S44, NO), the posterior distribution derived in step S43 is set as a new prior distribution (step S47).

[0041] After the measurement surface 1M has been measured a set first number of times and the prior distribution has been updated using the first likelihood function a set first number of times (step S44, YES), the estimation unit 40 estimates the deformation occurring in the structure 1 based on the updated posterior distribution (step S48). The deformation of the structure 1 can be derived by multiplying the coefficient vector fvr derived based on the posterior distribution by the basis vector Ur.

[0042] Methods for determining the posterior distribution include sequential Bayesian estimation, particle filters, MCMC (Markov chain Monte Carlo) methods, etc. When the deformation to be predicted is elastic and it can be assumed that the measurement error and the system error assumed in the state space model follow a Gaussian distribution, the use of a Kalman filter allows estimation with reduced computational cost.

[0043] Next, the estimation unit 40 outputs the estimated deformation (step S49). The deformation estimated by the estimation unit 40 is input to the output unit 5 shown in Fig. 1 via a network, and is displayed on a display or the like in the output unit 5, thereby notifying the manager. This allows the manager to carry out appropriate maintenance and conservation work for the structure 1 based on the estimated deformation.

[0044] Below, a description will be given of control processing that differs from the control processing of the estimation unit 40 shown in Fig. 4. Fig. 7 is a flow diagram showing another example of the flow of control processing of the estimation unit 40 of this embodiment. Steps S48-A1, S48-A2, and S48-A3 are added to the control processing of the estimation unit 40 shown in Fig. 4 described above. The control operations of steps S41 to S48 are the same as those in Fig. 4, and therefore will not be described again.

[0045] When the deformation of the structure 1 is estimated by multiplying the coefficient vector fvr by the basis vector Ur in step S48, the estimation unit 40 calculates an error, which is the difference between this estimated deformation and the displacement vector u* measured by the measurement unit 10 (step S48-A1). This calculated error is recorded in the estimation unit 40.

[0046] Next, the estimation unit 40 determines whether the calculated error is the minimum value among the recorded past errors (step S48-A2). If the calculated error is not the minimum value (step S48-A2, NO), the value of the parameter that was set as a constant in the first observation equation (Equation 4) relating the displacement vector u* and the coefficient vector fvr is changed (step S48-A3). Then, the control process from steps S41 to S48-A1 is performed again using the first state space model constructed using the new parameter value, and the estimation unit 40 changes the parameter value until the calculated error becomes the minimum value.

[0047] When the error reaches a minimum value (step S48-A2, YES), the estimation unit 40 outputs the deformation of the structure 1 estimated in step S48 when the error is minimum (step S49). In this way, since the deformation is estimated based on the first state space model in which optimal values ​​are set for the unknown parameters in the first state space model, it is possible to estimate the deformation with high accuracy and reduced error.

[0048] In determining the minimum error value in step S48-A2, since no past errors to be compared with are recorded when the error is calculated for the first time, the minimum value may not be determined. Instead, the parameters may be changed a set number of times, the error may be calculated each time, and the errors for each parameter may be recorded. After that, the minimum value may be determined in step S48-A2. Furthermore, if the displacement vector u* measured in step S45 is recorded, it is not necessary to measure it again when determining the minimum error value in step S48-A2. The determination of the minimum error in step S48-A2 can be performed using the recorded displacement vector u*.

[0049] In the above, the deformation is estimated based on the first state space model configured using the first observation equation and the first state equation, but it is also possible to estimate the deformation using only the first observation equation. In this case, step S32 for determining the first state equation and step S32 for generating the first state space model are not performed.

[0050] In the above description, the displacement vector u* is used as the first observation variable, and the first observation equation shows the correlation between this displacement vector u* and the coefficient vector fvr as the first state variable. However, this is not limited to this. As the first state variable, a deformation vector ui other than the coefficient vector fvr, or another element constituting the linear space of the deformation field of the measurement surface 1M, such as a basis vector Ur, may be used. That is, at least one of the coefficient vector fvr, the deformation vector u*i, and the basis vector Ur may be used as the first state variable as predicted deformation information indicating a prediction of deformation occurring on the measurement surface of the structure under multiple boundary conditions applied to the structure. The first observation equation may show the correlation between the first observation variable, which is the displacement vector u*, and the predicted deformation information as the first state variable. Furthermore, when the basis vector Ur is used as the first state variable, the basis vector Ur may be expressed by a probability distribution.

[0051] Below, a control process different from the control process of the estimation unit 40 shown in Fig. 4 will be described. Fig. 8 is a flow diagram showing another example of the flow of the control process of the estimation unit 40 of this embodiment. Step S48-B1 is added to the control process of the estimation unit 40 shown in Fig. 4 described above. In addition, an imaging device capable of acquiring an image of the measurement surface 1M of the structure 1 is used as the measurement unit 10.

[0052] Here, as a preliminary step of the control processing of the estimating unit 40, when setting the boundary conditions of the variable output unit 20 shown in Fig. 2, at least the longitudinal elastic modulus and Poisson's ratio are set as the material characteristic values ​​of the structure 1. The estimating unit 40 estimates the deformation, which will be described below, based on a first state space model using variables reflecting these material characteristic values.

[0053] In step S42, the estimation unit 40 derives, as a first prior distribution, at least one of displacement, tilt, and distortion as deformations occurring in the structure 1, and a probability distribution of the error thereof, based on the first state variable estimated by the first state space model and the probability distribution of the error thereof. In addition, in step S46, the estimation unit 40 derives, as a likelihood distribution of the first likelihood function, at least one of displacement, tilt, and distortion of the structure 1, and a probability distribution of the error thereof, based on an image of the measurement surface 1M acquired by the measurement unit 10. Note that, for example, in measuring distortion, the displacement of the measurement surface 1M can be measured and the distortion can be visualized by analyzing and calculating images of the measurement surface 1M before and after deformation.

[0054] In step S43, the estimation unit 40 updates the first prior distribution, which indicates a probability distribution of at least one of displacement, tilt, and distortion, with a first likelihood function, which indicates a probability distribution of at least one of displacement, tilt, and distortion derived from the image, to derive a first posterior distribution.

[0055] The estimation unit 40 then determines the soundness of the structure 1 based on the deformation occurring in the structure estimated based on the first posterior distribution (S48-B1). As the determination of soundness, the estimation unit 40 determines whether the estimated deformation and the probability distribution of its error exceed a set first threshold. This first threshold is set as an index value indicating whether the state of the structure 1 is in a state where soundness can be ensured, and a tolerance is set for at least one of the strain, tilt, and displacement estimated as deformation.

[0056] When strain is set as the first threshold, the estimation unit 40 outputs a probability distribution indicating the distribution of strain as the deformation of the structure estimated based on the first posterior distribution. Because the strain distribution is a differential value of displacement, it can capture minute displacements, allowing for a more detailed understanding of the deformation distribution on the measurement surface 1M, and enabling a more accurate assessment of the soundness of the structure 1. Furthermore, because the soundness of at least one of the strain, tilt, and displacement estimated as deformation is assessed based on a probability distribution, it is possible to output a reliability that takes into account uncertainties such as external forces acting on the structure 1, and the actual soundness of the structure 1 can be quantitatively determined based on the assessed soundness.

[0057] The estimation unit 40 outputs the deformation estimated in step S48 and the soundness of the structure 1 determined in step S48-B1 (step S49). This allows the manager to appropriately perform lifespan assessment, maintenance, conservation work, etc. of the structure 1 based on the output soundness.

[0058] The deformation estimation device of this embodiment, configured as described above, is a deformation estimation device that includes a control unit that estimates deformation that will occur in a structure based on measurement values ​​that indicate deformation on a measurement surface of the surface of the structure, wherein the control unit predicts predicted deformation information that indicates, by a probability distribution, deformation that will occur on the measurement surface of the structure under a plurality of boundary conditions added to the structure, sets a first observation equation that indicates the correlation between the measurement values ​​and a first observation variable that indicates, by a probability distribution, the error in the measurement values, and the predicted deformation information, and estimates the deformation that will occur in the structure based on the first observation equation.

[0059] In this way, the control unit predicts, as predicted deformation information, deformation occurring on the measured surface of the structure under multiple boundary conditions, including external forces acting on the structure. That is, the control unit sets multiple boundary conditions and performs a numerical analysis in advance to predict predicted deformation information for each of the set boundary conditions. Then, the control unit sets a first observation equation that correlates the predicted deformation information obtained for each boundary condition with a first observation variable that indicates deformation obtained from the measured values. The control unit then estimates the deformation occurring on the structure based on this first observation equation. That is, the predicted deformation information used as the first state variable in the first observation equation is derived based on the diverse and multiple boundary conditions assumed to be acting on the structure. This enables accurate estimation of the deformation of the structure. Furthermore, because the first state variable is represented by a probability distribution that includes errors, the value takes into account uncertainties such as external forces acting on the structure.

[0060] Furthermore, in the deformation estimation device of this embodiment configured as described above, the control unit: sets a first state equation representing the time transition of the first state variable using the predicted deformation information, and sets a first observation equation indicating the correlation between the first observation variable and the first state variable, and generates a first state space model that estimates the deformation of the structure based on the first state equation and the first observation equation; sets the first state variable estimated by the first state space model and a probability distribution of an error therein as a first prior distribution; derives a first likelihood function for the probability distribution of the displacement of the structure and an error therein obtained from the measurement values; derives a first posterior distribution based on the first prior distribution and the first likelihood function, and estimates the deformation that will occur in the structure based on the first posterior distribution.

[0061] In this way, the variable set as the first state variable in the first state equation is derived based on the diverse and multiple boundary conditions assumed to be applied to the structure. This enables accurate estimation of the deformation of the structure. Furthermore, since the first state variable is expressed using a probability distribution that includes error, the value takes into account the uncertainty of external forces acting on the structure, etc.

[0062] Furthermore, a first likelihood function is derived for the probability distribution of deformation of the structure and its error obtained from the measurement values ​​of the measurement unit, a first posterior distribution is derived based on the first prior distribution and the first likelihood function, and deformation occurring in the structure is estimated based on this first posterior distribution. In this way, the first prior distribution is updated using the first likelihood function that indicates the probability distribution of deformation that most closely matches the information included in the first prior distribution, thereby enabling even more accurate deformation estimation.

[0063] Furthermore, in the deformation estimation device of this embodiment configured as described above, the control unit derives, as the first state variable, a first predicted deformation indicating the difference before and after the deformation occurring on the measurement surface under a plurality of boundary conditions applied to the structure, or a probability distribution indicating the correlation between the boundary conditions and the first predicted deformation.

[0064] In this way, the first state variable may be the deformation vector ui as the first predicted deformation, which is obtained by numerical analysis and indicates the deformation occurring on the measured surface of the structure using a probability distribution, or an element indicating the correlation between this deformation vector ui and the boundary condition f. By using the deformation vector ui obtained from the measurement value or another element related to this deformation vector ui as the first state variable used in this way, it becomes possible to estimate multifaceted deformation according to the conditions of the structure, boundary conditions, etc., and to estimate deformation with even greater accuracy.

[0065] Furthermore, in the deformation estimation device of this embodiment configured as described above, the control unit derives a probability distribution of elements constituting the deformation field of the measurement surface with respect to the boundary conditions and the errors of the elements as a probability distribution showing the correlation between the boundary conditions and the first predicted deformation.

[0066] That is, the elements constituting the deformation field of the structure for a plurality of diverse and multiple boundary conditions are used as the elements that indicate the correlation between the boundary conditions and the deformation vector ui, which is the first predicted deformation, and are set as the first state variable. This enables more accurate estimation of the deformation of the structure. Furthermore, because the elements constituting the deformation field are expressed using a probability distribution that includes errors, the values ​​take into account the uncertainty of external forces acting on the structure, etc.

[0067] In addition, in the deformation estimation device of this embodiment configured as described above, the control unit derives coefficients of each basis vector that constitutes a linear space in the deformation field as the elements that constitute the deformation field.

[0068] In this way, the coefficients of each basis vector constituting the linear space of the deformation field and the probability distribution of the errors of the coefficients are derived as elements constituting the deformation field of the measurement surface, which become the first state variables in the first state equation. Since the probability distribution of the coefficients of each basis vector constituting the linear space of the deformation field is set as the first state variable in this way, the deformation field occurring in the structure can be expressed with high accuracy, and therefore it is possible to derive the deformation with high accuracy.

[0069] In the modified estimation device of this embodiment configured as described above, the control unit derives the probability distribution of the first observation variable in the first observation equation by multiplying the first state variable in the first state equation by the basis vectors.

[0070] In this way, the actual displacement of the structure, which is the first observation variable, is expressed by multiplying the coefficient vector, which is the first state variable, by the basis vector. In this way, by minimizing the information included in the prior distribution and expressing the first observation variable by multiplying the elements that make up the deformation field, it is possible to improve the calculation accuracy while shortening the calculation time for setting the variables used in the first state space model.

[0071] In addition, a Kalman filter is used to estimate the deformation of the structure from which the first posterior distribution is derived, and measurements are taken multiple times for the deformation of the same structure a set first number of times, and the estimated deformation is updated sequentially in accordance with the measurements, thereby improving the estimation accuracy.

[0072] Embodiment 2. Hereinafter, Embodiment 2 of the present application will be described with reference to the drawings, focusing on differences from Embodiment 1 above. Portions similar to those in Embodiment 1 above will be assigned the same reference numerals and description thereof will be omitted. FIG. 9 is a block diagram showing the schematic configuration of a deformation estimation device 260 according to Embodiment 2 and a deformation estimation system 200 including this deformation estimation device 260. FIG. 10 is a block diagram showing the schematic configuration of a measurement unit 210 included in the deformation estimation device 260 shown in FIG. 9. FIG. 11 is a flow diagram showing the flow of control processing of a variable output unit 270 included in the measurement unit 210 shown in FIG. 10. FIG. 12 is a flow diagram showing the flow of control processing of a model generation unit 280 included in the measurement unit 210 shown in FIG. 10. FIG. 13 is a flow diagram showing the flow of control processing of the measurement unit 210 shown in FIG. 10. FIG. 14 is a conceptual diagram for explaining control processing of the measurement unit 210 shown in FIG. 10 with respect to a structure 1 to be inspected.

[0073] The deformation estimation system 200 of this embodiment shown in Figure 9 includes a deformation estimation device 260 and an output unit 5 that outputs the estimation results of this deformation estimation device 260. The deformation estimation device 260 of this embodiment differs from the deformation estimation device 60 of embodiment 1 shown in Figure 1 in the configuration of the measurement unit 210. The measurement unit 10 of embodiment 1 measures the measurement surface 1M of the structure 1 multiple times from a fixed position. The measurement unit 210 of embodiment 2, as shown in Figure 14, measures the measurement surface 1M of the structure 1 multiple times while moving, thereby reducing measurement errors on the measurement surface 1M. It is assumed that the structure 1 does not move. In embodiment 2, the coordinate system of the measurement unit 10 is shown as two-dimensional.

[0074] The configuration of the measurement unit 210 will be described. As shown in Fig. 10, the measurement unit 210 includes a measurement device 211 that measures the position of feature point i on the measurement surface 1M of the structure 1 as a measurement value, and a control unit 212 that accurately estimates the position of feature point i on the measurement surface 1M of the structure 1 based on the measurement value. Thus, the deformation estimation device 260 of this embodiment includes, as control units, the control unit 50 shown in Embodiment 1 and the control unit 212 provided in the measurement unit 210. The control unit 212 includes a variable output unit 270, a model generation unit 280, an estimation unit 290, and an output unit 6. Control processing in each of these units will be described below.

[0075] First, the control processing of the variable output unit 270 will be described based on the flow diagram of the control processing of the variable output unit 270 shown in Fig. 11. The variable output unit 270 sets each variable used in the model generation unit 280 at the subsequent stage.

[0076] In step S71, the variable output unit 270 sets, as variables, first coordinates (Xw,i, Yw,i, Zw,i) which are the position of feature point i on the measurement plane 1M of the structure 1 in the three-dimensional coordinate system. Here, the first coordinates are defined in the world coordinate system.

[0077] Furthermore, the variable output unit 270 indicates, as a vector indicating conversion information between the first coordinates (Xw,i, Yw,i, Zw,i) of the feature point i in the world coordinate system and the second coordinates (ui, vi) in the two-dimensional coordinate system on the measurement device 211, a quaternion q indicating rotation in the three-dimensional coordinate system and a three-axis vector s indicating translation, as shown in the following (Equation 5), for example, and sets these as variables. Note that, as shown in FIG. 14 , a function for calculating a parameter for rotating the coordinates from the quaternion q is denoted as E.

[0078]

[0079] Furthermore, the variable output unit 270 sets, as a variable, a movement vector shown in the following equation (6) using the vector of the above conversion information, which indicates the movement speed of the measuring device 211 and its uncertainty.

[0080]

[0081] Here, hg and hs represent errors that are uncertainties in rotation and velocity.

[0082] Next, the variable output unit 270 sets the position (ui, vi) of feature point i in the two-dimensional coordinate system on the measurement device 211 and the probability distribution of its uncertainty (εui, εvi) as variables (step S72). Next, the variable output unit 270 outputs the derived variables (step S73). The output variables are input to the model generation unit 280 at the subsequent stage.

[0083] Next, the control processing of the model generation unit 280 will be described based on the flowchart of the control processing of the model generation unit 280 shown in Fig. 12. First, the model generation unit 280 acquires, as variables input from the variable output unit 270 in the preceding stage, the first coordinate of feature point i in a three-dimensional coordinate system, a quaternion q and a vector s which are transformation information vectors, movement vectors (d{q} / dt, d{s} / dt) indicating the movement speed of the measurement device 211 and their uncertainties (ηq, ηs), and the position of feature point i in a two-dimensional coordinate system measured by the measurement device 211 and the probability distribution (ui, vi) of its uncertainty (u, vi) (step S81).

[0084] Next, the model generation unit 280 defines the first coordinate of the feature point i in the three-dimensional coordinate system, the quaternion q, the vector s, and the moving speed (d{q} / dt, d{s} / dt) of the measuring device 211 as second state variables, and calculates the second state equations shown in the following equations that represent the time transition of these second state variables (step S82).

[0085]

[0086] In the second state equation, the position of the feature point in the world coordinate system (first coordinate) does not change even if the measurement time changes. In contrast, the rotational and translational vectors (q, s) and the rotational and translational change rates (d{q} / dt, d{s} / dt), which indicate the movement of the measurement device 211, change over time.

[0087] Next, the model generation unit 280 sets the position of feature point i in the two-dimensional coordinate system on the measurement device 211 and the probability distributions ui and vi, which include the uncertainty of the error, as second observation variables. The model generation unit 280 then calculates a second observation equation that indicates the correlation between this position (ui, vi) and the first coordinate of feature point i in the three-dimensional coordinate system, which are the second state variables, the quaternion q, the vector s, and the movement speed (d{q} / dt, d{s} / dt) of the measurement device 211 and its uncertainty (ηq, ηs) (step S83). The second observation equation is shown below.

[0088]

[0089] Here, hu and hv are functions that convert points in the world coordinate system into points on the two-dimensional coordinate system on the measurement device 211.

[0090] Next, the model generation unit 280 generates a second state space model for estimating the first coordinate, which is the position of the feature point i of the structure 1 on the three-dimensional coordinate system, using the second state equation and the second observation equation (step S84). The generated second state space model is input to the subsequent estimation unit 290.

[0091] Next, the control processing of the estimation unit 290 will be described with reference to the flowchart of the control processing of the estimation unit 290 shown in Fig. 13. First, the estimation unit 290 acquires the second state space model input from the model generation unit 280 at the preceding stage (step S91).

[0092] Next, the estimation unit 290 calculates a second prior distribution based on the second state space model (step S92). That is, a probability distribution is calculated including the first coordinates (Xw,i, Yw,i, Zw,i) of the feature point i in the three-dimensional coordinate system, the quaternion q, the vector s, and the moving speeds (d{q} / dt, d{s} / dt) of the measuring device 211 and their uncertainties (ηq, ηs), which are set as second state variables in the second state equation.

[0093] Next, the estimation unit 290 updates the second prior distribution with the second likelihood function to derive a second posterior distribution (step S93). This second likelihood function is derived as a function indicating the position of the feature point i in the two-dimensional coordinate system on the measurement device 211 and the probability distribution (ui, vi) of its uncertainty based on the measured value by measuring the feature point i with the measurement device 211 (step S95) (step S96).

[0094] Next, the estimation unit 290 determines whether the measurement of the measurement surface 1M in step S95 has been performed a set second number of times (step S94). If the measurement has not been performed the set second number of times (step S94, NO), the estimation unit 290 updates the second posterior distribution derived in step S93 as a new second prior distribution (step S97), and uses this as the prior distribution in step S92.

[0095] Methods for calculating the second posterior distribution include sequential Bayesian estimation, particle filters, etc. If the measurement error and the system error assumed in the state space model can be assumed to be Gaussian distributed, an ensemble Kalman filter can be used to reduce the computational cost and make the estimation.

[0096] After the measurement of the measurement surface 1M is performed the set second number of times and the prior distribution is updated using the second likelihood function the set second number of times (step S94, YES), the estimation unit 290 estimates and outputs the position of the feature point i of the structure 1 in the world coordinate system based on the updated second posterior distribution (steps S98, S99).

[0097] The first coordinates (Xw,i, Yw,i, Zw,i) of feature point i of structure 1 defined on the world coordinate system will have different values ​​in the second coordinates Co1, Co2 on the two-dimensional coordinate system on the measurement device 211, but in this way, it is possible to accurately estimate the first coordinate on the world coordinate system from the second coordinates Co1, Co2 on the measurement device 211. In this way, the measurement unit 210 derives the position of feature point i on the structure 1 with high accuracy, with measurement errors and the like reduced. Therefore, the deformation estimation device 260 of this embodiment can accurately estimate the deformation of the structure 1 based on the position information of feature point i of the structure 1 that has been measured with high accuracy.

[0098] The measuring device 211 that measures the feature point may be a strain gauge, or a device such as a laser displacement meter or an optical interferometer. The measuring device 211 may also be one that measures the feature point i from an image. Measuring the position of the feature point i from an image enables non-contact measurement over a wide area. Furthermore, since multiple points can be measured at once without contact, the measurement time can be shortened. A specific method for measuring the position of the feature point i from an image is a method using a digital image correlation method. With the digital image correlation method, the measuring device 211 can be performed using only a digital camera, allowing the measuring device 211 to be miniaturized.

[0099] Furthermore, if an unsampled Kalman filter is used in the estimation unit 290, measurements are taken multiple times while moving the measurement device in response to deformation of the same structure, and the positions of the estimated feature points are updated sequentially in response to the measurements, thereby improving the estimation accuracy.

[0100] In the deformation estimation device of this embodiment configured as described above, the control unit sets a second state equation that represents a time transition of the second state variables, using as second variables first coordinates of feature points on the measurement surface of the structure in a three-dimensional coordinate system, a transformation information vector that transforms the first coordinates into second coordinates in a two-dimensional coordinate system held by the measurement unit, and a movement vector using the transformation information vector of the measurement unit that measures the structure while moving within the three-dimensional coordinate system and a probability distribution of an error therein, and sets a second observation equation that represents a correlation between the second observation variable and the second state variable, using as a second observation variable the second coordinates of the structure in the two-dimensional coordinate system of the measurement unit, and generates a second state space model that estimates the first coordinates in the three-dimensional coordinate system of the feature points of the structure based on the second state equation and the second observation equation; The first coordinate of the feature point in the three-dimensional coordinate system estimated by the second state space model and a probability distribution of an error therebetween are used as a second prior distribution, a second likelihood function is derived for the second coordinate of the structure measured by the measurement unit in the two-dimensional coordinate system and a probability distribution of an error therebetween, a second posterior distribution is derived based on the second prior distribution and the second likelihood function, and the first coordinate of the feature point in the three-dimensional coordinate system of the structure is estimated based on the second posterior distribution.

[0101] In this way, by assuming the movement of the measurement unit when measuring feature points and including this in the second prior distribution in the control unit of the measurement unit, it is possible to reduce measurement errors in the positions of feature points due to positional deviation of the measurement unit.

[0102] Furthermore, for example, when deformation is measured from images captured by a camera fixed to a human hand or a robot, errors in the measurement position occur due to three-dimensional misalignment between the measured images. However, the measurement unit 210 of this embodiment can reduce such errors. Furthermore, when capturing images multiple times, there is no need to fix the camera or use multiple cameras to reduce measurement errors. Fixing the camera leads to the need to create jigs, which increases the size and complexity of the device. Using multiple cameras also poses the same problem of increasing the size of the device. According to the measurement unit of this embodiment, even when an administrator photographs a structure while moving around using a single camera without fixing the camera, the error between the images, even if there is a three-dimensional misalignment, can be reduced and deformation of the structure 1 being photographed can be measured with high accuracy.

[0103] In this way, the deformation estimation device of this embodiment can acquire the displacement of a structure with high accuracy, and therefore can accurately estimate the deformation occurring in the structure.

[0104] Embodiment 3. Hereinafter, Embodiment 3 of the present application will be described with reference to the drawings, focusing on the differences from Embodiment 1 above. Portions similar to those in Embodiment 1 above will be assigned the same reference numerals and description thereof will be omitted. Fig. 15 is a block diagram showing the schematic configuration of a deformation estimation device 360 ​​according to Embodiment 3 and a deformation estimation system 300 including this deformation estimation device 360. Fig. 16 is a flow diagram showing the flow of control processing of the deformation estimation device 360 ​​shown in Fig. 15.

[0105] As shown in FIG. 15, a deformation estimation device 360 ​​of this embodiment differs from the deformation estimation device 60 shown in the first embodiment in that it includes a crack estimation unit 341 .

[0106] The control process of the variable output unit 20 differs slightly from that of the first embodiment. When setting the boundary conditions f, which include load conditions indicating the state of external forces, etc., applied to the structure 1 and constraint conditions that constrain the displacement of the structure 1 in each direction and rotational direction, the variable output unit 20 sets an estimated crack within the structure 1. Specifically, if a crack is assumed to have occurred on the cross section 1A shown in FIG. 5, the variable output unit 20 sets the cross section 1A, which is an area including the location where the crack is expected to occur, as a crack candidate surface. Then, the variable output unit 20 sets the constraint condition for a node 2P, among multiple nodes 2P on the cross section 1A, that is set to have a crack, to no constraint on displacement. The variable output unit 20 then performs a numerical analysis based on this boundary condition to determine a deformation vector ui on the measurement surface 1M. Then, the positions of the nodes 2P that are set to have a crack train are sequentially changed, and the deformation vector ui is determined and saved for each node 2P. As a result, the coefficient vector fvr and the basis vector Ur, which are elements constituting the linear space of the deformation field of the measurement surface 1M and which are derived by the variable output unit 20, contain information about the crack in the cross section 1A.

[0107] Since the coefficient vector fvr and the basis vector Ur, which are variables acquired by the model generation unit 30, contain crack information, this information is reflected in the first space model. Except for the fact that the basis vector Ur and the coefficient vector fvr contain crack information, the first space model is the same as that of the first embodiment.

[0108] As in the first embodiment, the estimation unit 40 estimates and outputs the deformation on the measurement surface 1M (step S49). The crack estimation unit 341 selects constraint conditions that will produce deformation closest to the estimated deformation on the measurement surface 1M estimated by the estimation unit 40. That is, since the constraint conditions in the variable output unit 20 include information on cracks occurring at each node 2P on the cross section 1A, the shape of the crack can be obtained from the selected constraint conditions.

[0109] The crack estimation unit 341 outputs the obtained crack information (step S349A-2). The output crack information indicates the shape and position of the crack. The crack information may be displayed numerically or as an image.

[0110] In this way, cross section 1A, which is an area including a location where a crack is predicted to occur inside structure 1, and data on the deformation at measurement surface 1M that occurs when a crack exists at node 2P of cross section 1A are prepared in advance for each node through numerical analysis, and learning data showing the relationship between cross section 1A and measurement surface 1M are prepared.

[0111] In the deformation estimation device of this embodiment configured as described above, the control unit sets an area including a location where a crack is expected to occur inside the structure as a crack candidate surface under the constraint conditions that constitute the boundary conditions of the numerical analysis, divides the crack candidate surface into a plurality of unit surfaces, and sets nodes among the plurality of nodes that constitute each unit surface where a crack is set to exist as having no constraint on displacement; predicts the first predicted deformation on the measurement surface by numerical analysis based on the boundary conditions in which the constraint conditions are set; and estimates a crack existing on the crack candidate surface based on the constraint conditions corresponding to the first predicted deformation and the deformation that will occur in the structure estimated based on the first posterior distribution.

[0112] In this way, cracks occurring inside a structure can be estimated from deformation on the measurement surface, allowing for appropriate lifespan assessments, maintenance, and conservation work to be carried out even on structures that require careful management.

[0113] Embodiment 4. Hereinafter, embodiment 4 of the present application will be described with reference to the drawings, focusing on the differences from embodiment 1 above. The same parts as embodiment 1 above will be assigned the same reference numerals and description thereof will be omitted. FIG. 17 is a block diagram showing a schematic configuration of a deformation estimation system 400 according to embodiment 4. The deformation estimation system 400 of this embodiment is an inspection system for electric power equipment using the deformation estimation device 60 of embodiment 1. An electric power equipment connected to a power line that supplies electric power is set as the structure 1 to be inspected. This embodiment differs from the deformation estimation system 100 shown in embodiment 1 in that the control unit 50 includes a diagnosis unit 407 and a structure control unit 408.

[0114] Here, as a preliminary step to the control of the estimation unit 40, when setting the boundary conditions, the load conditions indicating the state of application of external forces, etc., to the electric power equipment as the structure 1 and the constraint conditions that constrain the displacement of the electric power equipment in each direction and in the rotational direction are collectively set as boundary conditions f. In addition, the longitudinal elastic modulus, density, and Poisson's ratio of the electric power equipment are set as material characteristic values.

[0115] The estimation unit 40 estimates the deformation based on a first state space model using variables reflecting the material property values. The diagnosis unit 407 determines the soundness and reliability of the electric power equipment based on the deformation occurring in the electric power equipment estimated based on the first posterior distribution.

[0116] As a determination of soundness, the diagnoser 407 determines whether the probability distribution of the deformation and its error estimated based on the first posterior distribution exceeds a set first threshold. This first threshold is set as an index value indicating whether the state of the electric power equipment is in a state where soundness can be ensured, and for example, a tolerance value for strain, tilt, displacement, etc. is set. As a determination of reliability, the diagnoser 407 determines whether the probability distribution of the deformation occurring in the structure and its error estimated based on the first posterior distribution exceeds a second threshold as an index value indicating reliability. As this second threshold, for example, the output, operating temperature, operating period, or operating frequency of the electric power equipment at which the electric power equipment can be used while remaining sound is set.

[0117] Next, the structure control unit 408 controls the electric power equipment based on the result of the diagnosis unit 407. For example, if it is necessary to reduce the output of the electric power equipment, the structure control unit 408 controls the output to be reduced.

[0118] In the above, the soundness of the electric power equipment is judged based on the deformation of the measurement surface 1M of the electric power equipment, but it is also possible to estimate a crack occurring inside the electric power equipment by using a configuration including the crack estimation unit 341 shown in embodiment 3. For example, the presence or absence of a crack may be determined by setting at least one of a displacement that restrains the electric power equipment when no crack has occurred in the electric power equipment with respect to the boundary conditions, and a displacement that restrains the electric power equipment when a crack has occurred in the electric power equipment.

[0119] Furthermore, while electric power equipment connected to a power line is shown as structure 1, an outdoor building such as a steel bridge or steel tower may also be set as structure 1. In this case, the boundary conditions include the displacement that constrains the steel bridge or steel tower, the shape of the steel bridge or steel tower, and the material property values ​​of the steel bridge or steel tower, such as the modulus of longitudinal elasticity, density, and Poisson's ratio. When inspecting for the presence or absence of cracks, the boundary conditions may include at least one of the constraint conditions for when there are no cracks inside the steel bridge or steel tower, and the constraint conditions for when there are cracks in the steel bridge or steel tower.

[0120] The diagnosis unit 407 may determine the range of use, period of use, frequency of use, or repair range of the railway bridge or steel tower after the estimation based on the estimated deformation, and may notify the manager by displaying the range of use, period of use, or repair range of the railway bridge or steel tower, sounding an alarm if the frequency of use is exceeded, and outputting the time to repair the repair range. For example, a drone may be set as the structure control unit 408, and the repair range may be output to the drone, and the repair may be performed by the drone.

[0121] The estimated information is used to output the conditions under which a steel bridge or tower can be used in a healthy state, as well as the repair scope, and the range of use, period, and repair scope, along with the timing of repairs and alarms for repair frequency, which are then output and reported to the administrator.This allows steel bridges or steel towers, which require careful management, to be controlled so that they can be used in a healthy state for a long period of time.In addition, because the system can diagnose the health of structures using a probability distribution, it is possible to determine the control output for structures that also take into account control uncertainty, improving control accuracy.

[0122] Furthermore, when applying this deformation estimation device to pressure vessels and pressure piping used in electric power equipment, the external forces set as boundary conditions may also include internal pressure, and the design pressure or pressure from past inspection records may be set.

[0123] Furthermore, if the structure is electric power equipment, the structure control unit may perform power control to adjust the output of the monitored electric power equipment based on the results of the reliability determination. Furthermore, if the structure is electric power equipment such as a power plant, the structure control unit may control the temperature of water, coolant (e.g., hydrogen), etc. supplied to the power plant as environmental elements for adjusting the operating environment of the power plant. Furthermore, the environmental temperature where the electric power equipment is installed and a temperature control device inside the electric power equipment may be controlled as environmental elements for adjusting the operating environment of the electric power equipment.

[0124] The deformation estimation system of this embodiment configured as described above is a deformation estimation system comprising: a deformation estimation device configured as described above; and an output unit that displays output from the control unit, wherein the control unit performs the numerical analysis by setting at least the longitudinal elastic modulus and Poisson's ratio of the structure as material property values ​​of the structure in the numerical analysis; performs a soundness assessment to determine whether the probability distribution of the deformation and its error occurring in the structure estimated based on the first posterior distribution exceeds a first threshold value as an index value indicating soundness; and displays the assessment result on the output unit.

[0125] This allows managers to confirm the integrity of structures based on accurately estimated deformations that occur in the structures, allowing them to carry out appropriate lifespan assessments, maintenance, and conservation work, even for structures that require careful management, enabling long-term operation.

[0126] Furthermore, in the deformation estimation system of this embodiment configured as described above, the measurement unit acquires an image of the measurement surface of the structure, and the control unit derives, as the first prior distribution, at least one of displacement, tilt, and distortion occurring in the structure and a probability distribution of an error thereof based on the first state variable estimated by the first state space model and a probability distribution of an error thereof, and derives, as the likelihood distribution of the first likelihood function, a probability distribution of at least one of displacement, tilt, and distortion of the structure based on the image and a probability distribution of an error thereof, and derives the first posterior distribution based on the first prior distribution and the first likelihood function, and performs the soundness assessment based on the first posterior distribution.

[0127] Furthermore, in the deformation estimation system of this embodiment configured as described above, the control unit sets, as the structure, an electric power equipment or an outdoor building connected to a power line that supplies electric power, performs the numerical analysis by setting at least the Young's modulus and Poisson's ratio of the electric power equipment or the outdoor building as the material characteristic values ​​in the numerical analysis, and performs a reliability determination to determine whether the probability distribution of the deformation and error occurring in the structure estimated based on the first posterior distribution exceeds a second threshold value as an index value indicating reliability, and displays the determination result on the output unit.

[0128] In this way, the soundness and reliability of a structure are judged based on deformation estimated by a state space model based on numerical analysis that reflects material property values, so that managers can confirm the soundness of a structure based on the deformation that occurs in the structure, which is accurately estimated, and can therefore perform appropriate lifespan assessments, maintenance, conservation work, etc., even for structures that require careful management. Furthermore, long-term operation is possible by controlling the structure based on the results of this soundness assessment or reliability assessment.

[0129] Furthermore, in the deformation estimation system of this embodiment configured as described above, in a configuration in which the electric power equipment is set as the structure, the control unit controls at least one of power control in the electric power equipment and environmental elements that adjust the operating environment of the electric power equipment based on the reliability judgment.

[0130] In this way, based on the results of the reliability assessment, the power control of the power equipment or the environmental factors that adjust the operating environment of the power equipment are adjusted, making it possible to prevent excessive use of the power equipment or ensure an appropriate temperature environment, thereby enabling long-term operation.

[0131] The hardware configuration of the control unit will be described below. Each embodiment is provided with a processing circuit for executing processing in the control unit. The processing circuit may be dedicated hardware or a CPU (Central Processing Unit, also referred to as a central processing unit, processing unit, arithmetic unit, microprocessor, microcomputer, processor, or DSP) that executes programs stored in memory.

[0132] Fig. 18 is a diagram illustrating an example of the hardware configuration of the control unit. In Fig. 18, a processing circuit 52 is connected to a bus B1. When the processing circuit 52 is dedicated hardware, the processing circuit 52 corresponds to, for example, a single circuit, a composite circuit, a programmed processor, an ASIC, an FPGA, or a combination thereof. Each function of each unit, including the control unit, of the deformation estimation device and deformation estimation system shown in each embodiment may be realized by the processing circuit 52, or the functions of each unit may be realized collectively by the processing circuit 52.

[0133] FIG. 19 is a diagram illustrating an example of the hardware configuration of the control unit. In FIG. 19, a processor 54 and a memory 53, which is a storage device, are connected to a bus B2. When the processing circuit is a CPU, the functions of each unit of the estimation device are realized by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in the memory 53. The processing circuit realizes the functions of each unit by reading and executing the program stored in the memory 53. In other words, the estimation device includes the memory 53 for storing a program whose steps are executed when executed by the processing circuit. Furthermore, these programs can be said to cause a computer to execute a procedure or method to be executed. Here, the memory 404 corresponds to a non-volatile or volatile semiconductor memory such as a RAM, a ROM, a flash memory, an EPROM, an EEPROM, or a magnetic disk, a flexible disk, an optical disk, a compact disk, a minidisk, a DVD, or the like.

[0134] The functions of each unit of the deformation estimation device may be partially realized by dedicated hardware and partially realized by software or firmware. For example, the model generation unit among the functions may be realized by a processing circuit as dedicated hardware. Furthermore, the processing circuit may read and execute a program stored in the memory 53 to realize the diagnosis unit that diagnoses the crack state among the functions.

[0135] Thus, the processing circuitry can implement each of the above functions by hardware, software, firmware, or a combination of these.

[0136] Although various exemplary embodiments and examples are described in this application, the various features, aspects, and functions described in one or more embodiments are not limited to the application of a particular embodiment, but may be applied to the embodiments alone or in various combinations. Therefore, countless modifications not illustrated are contemplated within the scope of the technology disclosed in this application. For example, this includes cases where at least one component is modified, added, or omitted, or where at least one component is extracted and combined with components of another embodiment.

[0137] 1 Structure, 10, 210 Measurement unit, 50, 212 Control unit, 60, 260, 360 Deformation estimation device, 1M Measurement surface, 100, 200, 300, 400 Deformation estimation system.

Claims

1. 1. A deformation estimation device including a control unit that estimates deformation occurring in a structure based on measurement values ​​indicating deformation in a measurement plane of a surface of the structure, The control unit is predicting predicted deformation information that indicates, by a probability distribution, deformations occurring on the measurement surface of the structure under a plurality of boundary conditions applied to the structure; setting a first observation equation indicating a correlation between a first observation variable indicating the measurement value and an error of the measurement value by a probability distribution and the predicted deformation information as a first state variable, and estimating a deformation occurring in the structure based on the first observation equation; Deformation estimation device.

2. The control unit is a first state equation expressing a time transition of a first state variable using the predicted deformation information as a first state variable, and a first observation equation expressing a correlation between the first observation variable and the first state variable; and generating a first state space model that estimates deformation of the structure based on the first state equation and the first observation equation; a probability distribution of the first state variable and an error thereof estimated by the first state space model is set as a first prior distribution, a first likelihood function is derived for a probability distribution of the displacement of the structure and an error thereof obtained from the measurement values, a first posterior distribution is derived based on the first prior distribution and the first likelihood function, and a deformation occurring in the structure is estimated based on the first posterior distribution; The deformation estimation device according to claim 1 .

3. The control unit is As the first state variable, deriving a first predicted deformation indicating a difference between before and after deformation occurring on the measurement surface under a plurality of boundary conditions applied to the structure, or a probability distribution indicating a correlation between the boundary conditions and the first predicted deformation; The deformation estimation device according to claim 2 .

4. The control unit is deriving a probability distribution of elements constituting a deformation field of the measurement surface with respect to the boundary condition and a probability distribution of errors of the elements as a probability distribution indicating a correlation between the boundary condition and the first predicted deformation; The deformation estimation device according to claim 3 . Deformation estimation device.

5. The control unit is Deriving coefficients of each basis vector constituting a linear space in the deformation field as the elements constituting the deformation field; The deformation estimation device according to claim 4 .

6. The control unit is deriving a probability distribution of the first observation variable in the first observation equation by multiplying the first state variable in the first state equation by the basis vector; The deformation estimation device according to claim 5 .

7. The control unit is setting the first observation equation including a parameter relating the first observation variable and the first state variable; changing values ​​of the parameters so as to minimize a difference between a deformation of the structure estimated from the first posterior distribution and a deformation of the structure obtained from the measured values, thereby estimating a deformation occurring in the structure; The deformation estimation device according to claim 2 .

8. The measurement of the measurement value on the measurement surface is performed a first number of times, the control unit derives the first likelihood function for a probability distribution of a displacement of the structure and an error therein based on the measurement value each time the measurement is performed, and derives the first posterior distribution based on the first prior distribution and the first likelihood function the first number of times. The deformation estimation device according to claim 2 .

9. The control unit is a second state equation is set to represent a time transition of a first coordinate of a feature point on the measurement surface of the structure in a three-dimensional coordinate system, a transformation information vector that transforms the first coordinate into a second coordinate in a two-dimensional coordinate system owned by a measurement unit that measures the measurement value, and a movement vector using the transformation information vector of the measurement unit that measures the structure while moving within the three-dimensional coordinate system, and a probability distribution of an error of the movement vector; and a second observation equation is set to represent a correlation between the second observation variable and the second state variable, and a second state space model is generated that estimates the first coordinate in the three-dimensional coordinate system of the feature point of the structure based on the second state equation and the second observation equation; a probability distribution of the first coordinate of the feature point in the three-dimensional coordinate system estimated by the second state space model and an error therebetween as a second prior distribution, a second likelihood function is derived for the probability distribution of the second coordinate of the structure measured by the measurement unit in the two-dimensional coordinate system and an error therebetween, a second posterior distribution is derived based on the second prior distribution and the second likelihood function, and the first coordinate of the feature point in the three-dimensional coordinate system of the structure is estimated based on the second posterior distribution. The deformation estimation device according to claim 2 .

10. The measurement unit acquires an image of the measurement surface of the structure, The control unit derives the second coordinates of the feature points in the two-dimensional coordinate system from the image. The deformation estimation device according to claim 9 .

11. the measurement unit measures the feature points a set second number of times while moving relative to the structure; the control unit derives the second coordinate in the two-dimensional coordinate system and the second likelihood function for a probability distribution of an error thereof based on the measurement value each time the measurement is performed by the measurement unit, and derives the second posterior distribution based on the second prior distribution and the second likelihood function the second number of times. The deformation estimation device according to claim 9 .

12. The control unit is In the constraint condition constituting the boundary condition, A region including a location where a crack is expected to occur within the structure is set as a crack candidate surface, the crack candidate surface is divided into a plurality of unit surfaces, and among a plurality of nodes constituting each of the unit surfaces, a node where a crack is set to exist is set as not being restrained with respect to displacement; predicting the first predicted deformation on the measurement surface based on the boundary condition in which the constraint condition is set; estimating a crack existing on the crack candidate surface based on the constraint condition corresponding to the first predicted deformation and the deformation occurring in the structure estimated based on the first posterior distribution; The deformation estimation device according to claim 3 .

13. A deformation estimation device according to any one of claims 1 to 12, An output unit that displays an output from the control unit, The control unit is setting at least a longitudinal elastic modulus and a Poisson's ratio of the structure as material characteristic values ​​of the structure in the prediction of the predicted deformation information, among the longitudinal elastic modulus, density, and Poisson's ratio of the structure; performing a soundness determination to determine whether or not the estimated deformation occurring in the structure and a probability distribution of the error therein exceed a first threshold value as an index value indicating soundness, and displaying the determination result on the output unit; Deformation estimation system.

14. an image of the measurement surface of the structure is acquired as the measurement value; The control unit is a first state equation expressing a time transition of a first state variable using the predicted deformation information as a first state variable, and a first observation equation expressing a correlation between the first observation variable and the first state variable; and generating a first state space model that estimates deformation of the structure based on the first state equation and the first observation equation; a probability distribution of the first state variable and an error thereof estimated by the first state space model is defined as a first prior distribution, a first likelihood function is derived for a probability distribution of the displacement of the structure and an error thereof obtained from the measurement values, a first posterior distribution is derived based on the first prior distribution and the first likelihood function, and a deformation occurring in the structure is estimated based on the first posterior distribution; deriving, as the first prior distribution, at least one of a displacement, a tilt, and a strain occurring in the structure and a probability distribution of an error thereof based on the first state variable estimated by the first state space model, and deriving, as a likelihood distribution of the first likelihood function, a probability distribution of at least one of a displacement, a tilt, and a strain of the structure based on the image and a probability distribution of an error thereof; deriving the first posterior distribution based on the first prior distribution and the first likelihood function, and performing the soundness determination based on the first posterior distribution; The deformation estimation system of claim 13.

15. The control unit is As the structure, an electric power device or an outdoor building connected to a power line that supplies electric power is set, As the material characteristic value, at least the longitudinal elastic modulus and the Poisson's ratio of the electric power equipment or the outdoor building are set, among the longitudinal elastic modulus, density, and Poisson's ratio; performing a reliability determination to determine whether or not the estimated deformation occurring in the structure and a probability distribution of the error thereof exceed a second threshold value as an index value indicating reliability, and displaying the determination result on the output unit; The deformation estimation system of claim 13.

16. The control unit is In a configuration in which the electric power device is set as the structure, Controlling at least one of power control in the electric power device and an environmental element for adjusting an operating environment of the electric power device based on the reliability determination. The deformation estimation system of claim 15.

17. A deformation estimation method using a deformation estimation device including a control unit that estimates deformation occurring in a structure based on a measurement value indicating deformation in a measurement plane of a surface of the structure, the method comprising: The control unit is A step of predicting predicted deformation information indicating, by a probability distribution, deformation occurring on the measurement surface of the structure under a plurality of boundary conditions applied to the structure; setting a first observation equation indicating a correlation between a first observation variable indicating the measurement value and an error of the measurement value by a probability distribution and the predicted deformation information as a first state variable, and estimating a deformation occurring in the structure based on the first observation equation. Deformation estimation methods.