Building structure load identification method and equipment

Abstract

The invention discloses a building structure load identification method and equipment. The method comprises the steps of determining an initial state quantity and an initial state quantity covariance of a building structure; based on the UT transformation principle of an unscented Kalman filter, generating a group of Sigma points according to the state quantity and covariance at the previous moment, and writing the Sigma points into a matrix form; calculating information of each column of Sigma points in the matrix on the basis of the Alembert principle, the structural quality and the matrix information to obtain a plurality of accelerated speed predicted values with the number equal to that of the columns of the matrix; calculating a weighted average value of all acceleration predicted values based on a weight value of UT (Unscented Kalman Filter); constructing a target function about the unknown load of the structure based on the acceleration measurement value measured by the sensor and the weighted average value; based on a Levenberg-Marquardt optimization method, calculating a corresponding load estimation value when the target function obtains a minimum value; updating the state quantity and the state quantity covariance of the unscented Kalman filter based on the load estimation value; and circularly executing the steps until all time steps are calculated.

Description

TECHNICAL FIELD

[0001] The application relates to a building structure load identification method and device, and belongs to the technical field of building structure health monitoring. BACKGROUND

[0002] As an important carrier of human production and life, the safety and reliability of a building structure are very important. With the continuous expansion of modern building scale and the increase of service life, the structure is inevitably affected by various factors, such as changes in use load, earthquakes, wind load, material degradation, construction quality defects and environmental effects. These factors may cause the structure to have a decreased bearing capacity, a degraded stiffness and a damage accumulation, and further cause a loss of function or even a structural failure.

[0003] Accurate identification of the load acting on a building structure in actual use is an important prerequisite for mastering the real working state of the structure and evaluating the safety and durability of the structure. At present, structure load identification methods mainly include direct measurement and indirect inference. The direct measurement method needs to arrange a large number of sensors at key positions of the structure to monitor the acting load in real time, but the arrangement cost is high, the maintenance is difficult, and the reliability of the sensors is easily affected by environmental interference. The indirect inference method is usually based on the dynamic response of the structure, and the acting load of the structure is identified through inversion calculation, but this method is limited by the ill-posedness of the inverse problem, and is highly dependent on the measurement accuracy, noise level and mathematical model, so there are still deficiencies in the identification accuracy and applicability.

[0004] Therefore, there is an urgent need for a building structure load identification method with higher precision, lower cost and wider application range. SUMMARY

[0005] To solve the problems mentioned in the background, the purpose of the application is to provide a building structure load identification method, which comprises the following steps: determining the initial state quantity and the initial state quantity covariance of the building structure; generating a set of Sigma points according to the state quantity and the covariance at the previous moment based on the UT transformation principle of the unscented Kalman filter; writing the Sigma points into a matrix form; calculating each column of Sigma point information in the matrix based on the D'Alembert principle and the mass and matrix information of the structure, to obtain a plurality of acceleration prediction values equal to the number of columns of the matrix; calculating the weighted average value of all the acceleration prediction values based on the weight value of the UT transformation of the unscented Kalman filter; constructing a target function about the unknown load of the structure based on the acceleration measurement value measured by the sensor and the weighted average value; calculating the load estimation value corresponding to the minimum value of the target function based on the Levenberg-Marquardt optimization method; updating the state quantity and the state quantity covariance of the unscented Kalman filter based on the load estimation value; and cyclically executing the above steps until all the time steps are calculated.

[0006] Preferably, the displacement, velocity, stiffness, and damping of the target structure are selected as the components of the state variables; the acceleration response of the target structure is selected as the observation value in the identification process.

[0007] Preferably, based on the UT transform principle of the unscented Kalman filter, a set of Sigma points is generated according to the state variables and covariance of the previous time step, as shown in the following formula:

[0008] In the above formula, Indicates the first k Time step i One SP point; ,and k ∈[1, N ], N This represents the total number of time steps; i Indicates the first i SP points, and i ∈[0, 8 n ]; σ = 4 n ; λ = φ 2 ( σ + κ )- σ , φ Indicates the scaling factor. φ =0.05, κ Indicates the proportional parameter. κ = 0 or 3- σ ; Preferably, based on d'Alembert's principle and structural mass and matrix information, the formula for calculating the acceleration prediction value by analyzing the Sigma point information in each column of the matrix is ​​as follows:

[0009] In the above formula, They represent the first k The time step is the quantity corresponding to the structural displacement, structural velocity, structural stiffness, and structural damping after calculation, i = 0, 1, 2, 3, ..., 8n.

[0010] Preferably, the formula for calculating the weighted average of all acceleration predictions based on the weight values ​​of the unscented Kalman filter UT transform is as follows:

[0011] In the above formula, Indicates the first i The average weight value of each SP point. Indicates the first k The weighted average of the time-step acceleration predictions.

[0012] Preferably, based on the acceleration measurements obtained from the sensors and the weighted average value, the objective function for the unknown load on the structure is constructed as follows:

[0013] In the above formula, Indicates the number of measurements taken by the sensor k Acceleration measurements at the time step F k Indicates the first k The unknown load on the structure at the time step, where L represents the unknown load. F k The position matrix.

[0014] Preferably, based on the Levenberg-Marquardt optimization method, the load value corresponding to the minimum value of the objective function is calculated. And based on the load value The state variables corresponding to each Sigma point are obtained by solving the state equations. The estimated observation value for each Sigma point is obtained by solving the observation equation. .

[0015] Preferably, the measurement update step based on the unscented Kalman filter algorithm calculates the output of the first step. k Predicted values ​​of time steps Measurement prediction covariance P yy,k and mutual covariance The formula is: ; ; ; In the above formula, For the first k Time step i The mean weight value of each Sigma point, For the first k Time step i Covariance weights of Sigma points For the first k Time step i The estimated observation values ​​corresponding to each Sigma point; For the first k Predicted measurement value at time step, R k For the first k Noise in the time step; Preferably, updating the state variables and state variable covariance of the unscented Kalman filter based on the load estimate further includes: Update #k Kalman gain matrix at time step: ; Update and output the first k State variables at time step: ; Update and output the first k Covariance of state variables at time step: .

[0016] The present invention also provides a building structure load identification device, the device including a processor and a memory, the memory storing at least one instruction, the at least one instruction being loaded and executed by the processor to implement the above-described building structure load identification method.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: I. Simple operation and no damage to the structure: Only acceleration sensors need to be installed in local areas of the structure, without the need for a large number of sensors or damage to the original structure, reducing construction difficulty and cost.

[0018] II. High recognition accuracy: By combining the nonlinear estimation capability of the unscented Kalman filter with the strong convergence of the Levenberg-Marquardt optimization algorithm, the ill-conditioned nature of the inverse problem and the influence of noise are effectively reduced, thereby improving the accuracy of load recognition.

[0019] III. Wide range of applications: It is suitable for complex working conditions such as impact loads and seismic action, and can be adapted to different degrees of freedom and different types of building structures.

[0020] IV. Green and Environmentally Friendly: By accurately identifying loads, timely maintenance and repair of the structure can be achieved, avoiding large-scale damage and reducing resource waste. Attached Figure Description

[0021] For ease of explanation, the present invention will be described in detail below with reference to specific embodiments and accompanying drawings.

[0022] Figure 1 This is a diagram illustrating the effect of identifying impact loads with noise. Figure 2 Image showing the effect of seismic action identification; Figure 3 The diagram shows the building structure of an example. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the invention is described below with reference to specific embodiments shown in the accompanying drawings. However, it should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0024] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0025] Specific Implementation Method 1: This implementation method is a load identification method for building structures. The identification method provided by this invention is a load identification method based on the acceleration response of building structures. Essentially, it is a load identification method that integrates unscented Kalman filters and LM optimization, where the LM optimization method is the Levenberg-Marquardt optimization method. Based on d'Alembert's principle, structural dynamics, and finite element theory, the motion control differential equations of the structure are derived or a finite element model of the structure is established. The structure in question is a building structure. For a building structure, determine the initial values ​​of the corresponding state of the building structure and form the initial state variables. The covariance matrix of the initial state variables, referred to as the initial state variable covariance P0, is determined based on the Kalman filtering principle; where... P0 and P1 are respectively called the 0th time step, i.e., the state quantity and state covariance of the initial step. A set of sigma points, abbreviated as SP points for convenience, is generated based on the UT transform principle of the unscented Kalman filter algorithm. Each SP point represents a state quantity. With state quantity covariance P k The calculated output is then processed. The generated SP points are written in matrix form, and the predicted acceleration values ​​are solved based on d'Alembert's principle, the structural mass M, and the information in each column of the matrix. The number of predicted acceleration values ​​equals the number of columns in the matrix. A weighted average of all predicted acceleration values ​​is calculated based on the weights of the unscented Kalman filter's UT transform. An objective function for solving the unknown load on the structure is constructed based on the acceleration measurements obtained from sensors. The load value corresponding to the minimum value of the objective function is calculated using the LM optimization method. These steps are repeated until all time steps are calculated.

[0026] Specific Implementation Method Two: The content not mentioned in this implementation method is the same as in Specific Implementation Method One, selecting the displacement, velocity, stiffness, and damping of the target structure as components of the state variables; the structural displacement X and structural velocity... The vector consisting of structural stiffness K and structural damping C is called the state variable, denoted by the symbol... This indicates that the structural degrees of freedom are... n That is, X, The dimensions of K and C are all n State variables The dimension is 4 n，The dimension of the state quantity determines the number of sigma points; the acceleration response of the target structure is selected as the observation value in the identification process. The observation value generally refers to a series of physical quantities that can be measured by sensors, including displacement, velocity, acceleration, strain, stress, and force. For building structures, the acceleration measurement value is used as the observation value in this implementation method.

[0027] Specific Implementation Method Three: Content not mentioned in this implementation method is the same as in Specific Implementation Method One or Two. Based on the UT transform principle of the unscented Kalman filter, the ( ) k -1) State variables at time step and state quantity covariance P k-1 Generate the first k time step (8) n +1) sigma points, referred to as SP points for ease of explanation, are shown in the following formula:

[0028] In the above formula, Indicates the first k Time step i One SP point; ,and k ∈[1, N ], N This represents the total number of time steps; i Indicates the first i SP points, and i ∈[0, 8 n ]; σ = 4 n ; λ = φ 2 ( σ + κ )- σ , φ Indicates the scaling factor. φ =0.05, κ Indicates the proportional parameter. κ = 0 or 3- σ ; and the generated (8 n +1) SP points can be written in matrix form as follows:

[0029] In the formula, Representing the first k The time step is a calculated quantity corresponding to structural displacement, structural velocity, structural stiffness, and structural damping, and is not X as mentioned above. K and C; Represents real numbers; Specific Implementation Method Four: Content not mentioned in this implementation method is the same as in Specific Implementation Methods One, Two, or Three. Based on d'Alembert's principle, structural mass, and matrix information, the formula for calculating the acceleration prediction value using the Sigma point information in each column of the matrix is ​​as follows:

[0030] In the above formula, They represent the first k The time step is the quantity corresponding to the structural displacement, structural velocity, structural stiffness, and structural damping after calculation, i = 0, 1, 2, 3, ..., 8n.

[0031] Specific Implementation Method Five: Content not mentioned in this implementation method is the same as in Specific Implementation Methods One, Two, Three, or Four. Based on the weight values ​​of the unscented Kalman filter UT transform, the formula for calculating the weighted average of all acceleration prediction values ​​is as follows:

[0032] In the above formula, Indicates the first i The average weight value of each SP point. Indicates the first k The weighted average of the time-step acceleration predictions.

[0033] Specific Implementation Method Six: Content not mentioned in this implementation method is the same as in Specific Implementation Methods One, Two, Three, Four, or Five. Based on the acceleration measurements obtained from the sensors and the weighted average value, the objective function for the unknown load on the structure is constructed as follows:

[0034] In the above formula, Indicates the number of measurements taken by the sensor k Acceleration measurements at the time step F k Indicates the first k The unknown load on the structure at the time step, where L represents the unknown load. F k The position matrix.

[0035] Specific Implementation Method Seven: Content not mentioned in this implementation method is the same as in Specific Implementation Methods One, Two, Three, Four, Five, or Six. Based on the Levenberg-Marquardt optimization method, the load value corresponding to the minimum value of the objective function is calculated. And based on the load value The state variables corresponding to each Sigma point are obtained by solving the state equations. The estimated observation value for each Sigma point is obtained by solving the observation equation. In this invention, the state equation is a differential operation of state quantities. State quantities are written in the form of displacement and velocity for ease of derivation, because differentiating displacement yields velocity, and differentiating velocity yields acceleration. The state equation can be easily derived from the structural motion control differential equation. Furthermore, when the observed value is acceleration, it can be derived from the displacement and velocity in the state quantities according to d'Alembert's principle. In summary, the displacement and velocity in the state quantities serve the derivation of the state equation and the observation equation. Since the specific details of this process are common knowledge, they will not be elaborated upon in this invention. It should be noted that velocity in this invention is equivalent to an intermediate quantity and does not need to be output; its existence is only to serve the derivation of the state equation and the observation equation, and its calculation is completed by recursive algorithm solution. In some embodiments, based on the state quantities... Based on the motion control differential equations of the target structure, and using linear algebra matrix operations and numerical differentiation and integration operations of numerical analysis, the equations for solving the state variables are derived; in other embodiments, based on the state variables... The output of state variables is set based on the finite element model of the target structure and used as the state equation. The influence of noise is considered in this process. According to the type of observation, the equation relationship for solving the observation is derived based on the motion control differential equation of the building structure and on knowledge of linear algebra matrix operations, mathematical transposition, and mathematical combination of like terms. Alternatively, according to the type of observation, the output of the corresponding observation is set based on the finite element model of the building structure and used as the observation equation. The influence of noise is considered in this process.

[0036] Specific Implementation Method Eight: Content not mentioned in this implementation method is the same as in Specific Implementation Methods One, Two, Three, Four, Five, Six, or Seven. The time update step based on the unscented Kalman filter algorithm completes from the ( ) k -1) Time step reaches the 1st k The updates to the state variables and state variable covariance at each time step are denoted as follows: and The formula is as follows: , ; In the formula, and The first k Time step i The mean weight value and covariance weight value of each SP point For the first k Time step i The estimated state variables corresponding to each SP point, Q k For the first k Noise in the time step; Measurement update step calculation output based on unscented Kalman filter algorithmk Predicted values ​​of time steps Measurement prediction covariance P yy,k and mutual covariance The formula is: ; ; ; In the above formula, For the first k Time step i The mean weight value of each Sigma point, For the first k Time step i Covariance weights of Sigma points For the first k Time step i The estimated observation values ​​corresponding to each Sigma point; For the first k Predicted measurement value at time step, R k For the first k Noise at time steps, cross-covariance Depend on and constitute.

[0037] Specific Implementation Method Nine: Content not mentioned in this implementation method is the same as in Specific Implementation Methods One, Two, Three, Four, Five, Six, Seven, or Eight. Updating the state variables and state variable covariance of the unscented Kalman filter based on the load estimate includes: Update # k Kalman gain matrix at time step: ; Update and output the first k State variables at time step: ; Update and output the first k Covariance of state variables at time step: 。; After the update, the time step becomes ( k +1), repeat the above steps until the maximum time step. N Complete, meaning until the loop ends.

[0038] Specific Implementation Method Ten: To fully illustrate the present invention, this invention uses building structures as examples for illustration. Content not mentioned in this implementation method is the same as in Specific Implementation Methods One, Two, Three, Four, Five, Six, Seven, Eight, or Nine. This implementation method first... Figure 3 The six-story building structure shown is explained below: The building structure includes mass, stiffness, damping, seismic action, and ground.

[0039] The quality mentioned includes m 1. m 2. m 3. m 4. m 5、 and m 6.

[0040] The stiffness mentioned includes k 1. k 2. k 3. k 4. k 5、 and k 6.

[0041] The damping includes c 1. c 2. c 3. c 4. c 5、 and c 6.

[0042] The implementation process is as follows: First, based on structural dynamics and finite element theory, the differential equations for the motion control of building structures are derived.

[0043] Secondly, based on the finite element theory, the displacement and velocity state of a building structure can be characterized by the displacement and velocity of each floor's degrees of freedom. Before an earthquake, it is assumed that the displacement and velocity of each floor's degrees of freedom are zero, while the initial values ​​of stiffness and damping of each floor can be obtained based on the building structure's material composition and structural dynamics theory. Therefore, the initial state quantities... Given that, based on the principle of the Kalman filter, the initial state quantity covariance P0 is obtained.

[0044] Third, during an earthquake, the acceleration response of a building structure can be collected or calculated using sensors or numerical simulation methods, thus the observed values ​​are known.

[0045] Fourth, based on the above-mentioned building structure load identification method and the above-mentioned initial information, building structure load identification can be performed.

[0046] To facilitate the application of this invention, the key parts of the algorithm are further explained here, including the selection of initial state variables. To further explain, the initial state variable is chosen because it is closely related to the setting of the initial state variable covariance P0 and the finally identified state variable. Based on the preceding explanation... Figure 3As shown, the structural degrees of freedom of this embodiment are 6. Therefore, the displacement and velocity states of this building are characterized by 6 parameters, and considering 6 stiffness and 6 damping parameters, the state quantity dimension of this embodiment is (6+6+6+6=24).

[0047] Specific Implementation Method Eleven: This implementation method is a building structure load identification device. The device includes a processor and a memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement a building structure load identification method. It should be understood that the device described in this implementation method includes, but is not limited to, devices including processors and memory. It may also include devices corresponding to other units or modules with information acquisition, information interaction, and control functions. For example, the device may also include signal acquisition devices. The device includes, but is not limited to, PCs, workstations, and mobile devices.

[0048] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method of identifying a load of a building structure, characterized by, The method comprises the following steps: determining initial state quantity and initial state quantity covariance of the building structure; generating a set of Sigma points according to the state quantity and covariance at the previous time based on UT transformation principle of the unscented Kalman filter; writing the Sigma points in matrix form; calculating each column of Sigma point information in the matrix based on D'Alembert's principle and mass and matrix information of the structure to obtain a plurality of acceleration prediction values equal to the number of columns of the matrix; calculating a weighted average value of all the acceleration prediction values based on weight values of UT transformation of the unscented Kalman filter; constructing a target function about unknown loads of the structure based on the acceleration measurement value measured by the sensor and the weighted average value; calculating a load estimation value corresponding to the minimum value of the target function based on Levenberg-Marquardt optimization method; updating the state quantity and state quantity covariance of the unscented Kalman filter based on the load estimation value; and cyclically executing the above steps until all time steps are calculated.

2. The building structural load identification method according to claim 1, characterized by, Displacement, velocity, stiffness and damping of the target structure are selected as components of the state quantity; and acceleration response of the target structure is selected as the observation value in the identification process.

3. The building structural load identification method according to claim 2, characterized by, generating a set of Sigma points according to the state quantity and covariance at the previous time based on UT transformation principle of the unscented Kalman filter, the formula being as follows: ； In the above formula, denotes the k time step i th SP point; , and k ∈[1, N ], N denotes the total number of time steps; i represents the i-th SP point, and i ∈ [0, 8 i ] ; n = 4 σ ; n λ=φ 2 σ κ σ , φ represents a scaling factor, φ = 0.05, κ represents a scaling parameter, κ = 0 or 3- σ .​​​​ 4. The building structural load identification method according to claim 3, characterized by, calculating each column of Sigma point information in the matrix based on D'Alembert's principle and mass and matrix information of the structure to obtain a formula of the acceleration prediction value being as follows: ； In the above formula, respectively represent the first k The quantities corresponding to the structural displacement, structural velocity, structural stiffness and structural damping after the calculation of the time step, i = 0, 1, 2, 3,..., 8n.

5. The building structural load identification method according to claim 4, characterized by, calculating a weighted average value of all the acceleration prediction values based on weight values of UT transformation of the unscented Kalman filter, the formula being as follows: ； In the above formula, denotes the mean weight value of the i SP point, denotes the weighted average of the k acceleration prediction value of the time step.

6. The building structural load identification method according to claim 5, wherein constructing a target function about unknown loads of the structure based on the acceleration measurement value measured by the sensor and the weighted average value, the formula being as follows: In the above equation, denotes the acceleration measurement at the k time step, and F k denotes the unknown load at the k time step, and L denotes the unknown load F k position matrix.

7. The building structural load identification method according to claim 6, wherein Based on the Levenberg-Marquardt optimization method, the load value corresponding to the minimum value of the target function is calculated , and based on the load value , the state quantity corresponding to each Sigma point is solved by the state equation , and the observation estimation value corresponding to each Sigma point is solved by the observation equation .

8. The building structural load identification method according to claim 7, wherein The measurement update step of the unscented Kalman filter algorithm calculates the output of the time step k The measurement prediction value of the time step , the measurement prediction covariance P yy,k and the cross covariance , and the formula is: ； ； ； In the above formula, is the k mean weight value of the i th Sigma point at the th time step, k is the i covariance weight value of the th Sigma point at the k th time step, i is the observation estimation value corresponding to the th Sigma point at the th time step; k is the measurement prediction value at the k th time step, k is the noise at the k th time step.

9. The building structural load identification method according to claim 8, wherein updating the state quantity and state quantity covariance of the unscented Kalman filter based on the load estimation value comprises: updating the first k Kalman gain matrix of the time step: ; updating and outputting the first k state quantity of the time step: ; update and output the first k state quantity covariance of the time step: .

10. A building structure load identification device characterized by comprising: The device comprises a processor and a memory, and the memory stores at least one instruction which is loaded and executed by the processor to implement the building structure load identification method according to any one of claims 1 to 9.

Images