Sensor multi-target optimization arrangement method for health monitoring of bridge girder erection machine and application of sensor multi-target optimization arrangement method

Abstract

The invention belongs to the related technical field of sensor optimization arrangement, and discloses a bridge girder erection machine health monitoring-oriented sensor multi-target optimization arrangement method and application thereof, and the method comprises the following steps: (1) carrying out modal analysis on a finite element model of a bridge girder erection machine to obtain a modal shape matrix, and determining the number of modals; (2) training neural network models corresponding to the three objective functions by adopting a data set; and (3) taking the three trained neural network models as fitness functions of a Hemma optimization algorithm, and improving the Hemma optimization algorithm by combining a Pareto theory to obtain an improved Hemma optimization algorithm, thereby obtaining a final Pareto leading edge solution set and further obtaining an optimal solution. According to the method, the He-Marine optimization algorithm is combined with the neural network, and the trained neural network model is used as the fitness function of the He-Marine optimization algorithm, so that the calculation speed of a fitness objective function value in the optimization algorithm is greatly improved, and the optimization efficiency of the method is further improved.

Description

Technical Field

[0001] The present invention belongs to the technical field related to sensor optimization layout, and more specifically, relates to a sensor multi-objective optimization layout method for bridge erection machine health monitoring and its application. Background Art

[0002] With the acceleration of urbanization, the demand for urban infrastructure construction is increasing. As the key link of the urban transportation network, the construction and maintenance of bridges are crucial to ensure the smoothness and safety of urban transportation. As an important equipment in bridge construction, the stability and safety of the performance of bridge erection machines are directly related to the quality of the project and the safety of construction workers.

[0003] During the bridge construction process, the structural health monitoring of the bridge erecting machine becomes particularly important. Sensor monitoring is an effective means. By installing sensors at key parts of the bridge erecting machine, its operating status can be monitored in real time, and the changing trends of key parameters can be captured and analyzed, thereby achieving comprehensive monitoring of the bridge erecting machine status. The selection, quantity and optimal layout of sensors are key links to ensure the monitoring effect.

[0004] At present, the criteria for sensor layout are relatively mature, and commonly used methods include modal confidence, kinetic energy, and information matrix. These methods help determine the optimal layout position of sensors through different evaluation indicators. Considering multiple layout criteria, multi-objective optimization is performed. One is to obtain a new objective function by performing a simple weighted assignment operation on two objective functions. Among them, the human subjective influence factor is large and not scientific and reasonable. The other is the Pareto multi-objective sensor layout criterion. The Pareto frontier solution set is first calculated by the "calculate first, decide later" method, and then the final optimal solution is selected from the Pareto frontier solution set. However, in most sensor layout optimization methods, a single objective function or a single objective function composed of multiple objectives with simple weighting is used to optimize sensor layout, which makes it difficult to comprehensively consider various factors. At the same time, with the increase in the number of measuring points and sensors, the amount of calculation will increase sharply. The calculation time of the objective function will continue to increase with the increase in the number of measuring points and sensors, resulting in low optimization efficiency. Summary of the invention

[0005] In view of the above defects or improvement needs of the prior art, the present invention provides a multi-objective optimization layout method of sensors for bridge erection machine health monitoring and its application, which aims to solve the problem of low layout efficiency of existing sensors.

[0006] To achieve the above object, according to one aspect of the present invention, a multi-objective optimization arrangement method of sensors for health monitoring of a bridge erecting machine is provided, the method comprising the following steps:

[0007] Step 1: Perform modal analysis on the finite element model of the bridge erecting machine to obtain the modal vibration matrix and determine the number of modes;

[0008] Step 2: Using a data set to train a neural network model corresponding to three objective functions; wherein the three objective functions are modal independence, vibration response intensity, and damage sensitivity; the data set is composed of a plurality of sensor arrangement points and their corresponding three objective function values, and the three objective function values ​​are calculated based on the obtained modal vibration matrix and the number of modes;

[0009] Step three, the three trained neural network models are used as the fitness functions of the Hippo optimization algorithm respectively, and the Hippo optimization algorithm is improved in combination with the Pareto theory to obtain the improved Hippo optimization algorithm; the final Pareto frontier solution set is obtained by the improved Hippo optimization algorithm, and then the optimal solution is obtained, that is, the optimal layout plan of the sensor points.

[0010] Furthermore, when the final Pareto frontier solution set is obtained by improving the Hippo optimization algorithm, all solutions are first calculated based on the initial sensor layout points in one cycle, and then the Pareto frontier solution set of the current cycle is obtained through the Pareto theory. In the next cycle, the currently obtained Pareto frontier solution set is used as the initial solution to obtain the final Pareto frontier solution set.

[0011] Furthermore, the standardized absolute difference method is used to measure the final Pareto frontier solution set to obtain the optimal solution.

[0012] Furthermore, the Fisher information matrix Q is obtained based on the mode shape matrix:

[0013]

[0014] Where m is the selected modal order;

[0015] The ROC rate of change of the 2-norm (∥Q∥2) of the Fisher information matrix Q is used as the evaluation criterion, and a reasonable number of modes is selected according to the trend of the change rate.

[0016] Furthermore, the modal vector space intersection angle is used to evaluate the modal independence index; the vibration response intensity index reflects the sensor's response ability to structural vibration at the monitoring point; and the damage sensitivity index reflects the sensor's sensitivity to structural damage.

[0017] Furthermore, in the evaluation of modal independence index, the modal assurance criterion MAC effectively measures the independence between the modes measured at the measuring point by evaluating the size of the intersection angle of the modal vector space:

[0018]

[0019] Among them, MAC ij represents the off-diagonal elements in the MAC matrix, and i≠j, Φ i With Φ j Represents the i-th and j-th order mode vibration vectors; when MAC ij = 0, indicating that the i-th and j-th order mode vibration vectors are completely orthogonal; when MAC ij When =1, it indicates that the i-th and j-th order mode vibration vectors completely overlap.

[0020] Furthermore, based on the damage sensitivity Fisher information matrix Q s for:

[0021] Q s =S T S

[0022] in,

[0023]

[0024] Where S i represents the i-th order damage sensitivity, L represents the total number of structural elements, Kt represents the element stiffness matrix of the t-th element, λ i Represents the i-th order eigenvalue of the structure; decompose according to the degree of freedom so that Q s It is equal to the sum of the damage Fisher information matrices corresponding to each degree of freedom:

[0025]

[0026] Among them, the size of the matrix trace value is used to evaluate the damage sensitivity of each degree of freedom:

[0027]

[0028] When the trace value trace(Qsj) of the j-th degree of freedom corresponding to the measuring point is larger, it means that the measuring point is more sensitive to damage.

[0029] The present invention also provides a health monitoring method for a bridge-building machine. The health monitoring method first adopts the above-mentioned multi-objective optimization layout method of sensors for health monitoring of the bridge-building machine to obtain the optimal layout plan of sensor points; then, multiple sensors are arranged according to the optimal layout plan, and then the health monitoring of the bridge-building machine is performed through the multiple sensors.

[0030] The present invention also provides a sensor multi-objective optimization layout system for bridge-building machine health monitoring, which includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it executes the sensor multi-objective optimization layout method for bridge-building machine health monitoring as described above.

[0031] The present invention also provides a computer-readable storage medium: the computer-readable storage medium stores machine-executable instructions, and when the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the above-mentioned multi-objective optimization arrangement method of sensors for bridge-building machine health monitoring or / and the health monitoring method of the bridge-building machine.

[0032] In general, compared with the prior art, the above technical solution conceived by the present invention, the multi-objective optimization arrangement method of sensors for health monitoring of bridge erecting machines and its application provided by the present invention mainly have the following beneficial effects:

[0033] 1. The present invention combines the Hippo optimization algorithm with a neural network. The input of the neural network is the layout points of the sensors, and the output is the values ​​of the three objective functions. The trained neural network model is used as the fitness function of the Hippo optimization algorithm, which greatly improves the calculation speed of the fitness objective function value in the optimization algorithm, thereby improving the optimization efficiency of the present invention.

[0034] 2. The present invention adopts Pareto theory to deploy multi-target sensors, calculates the Pareto frontier solution set through the method of "calculate first, decide later", and then selects the final optimal solution in the Pareto frontier, thereby improving the optimization efficiency and quality.

[0035] 3. The present invention takes into account the Pareto optimality theory, combines the Hippo optimization algorithm and the neural network to perform multi-objective optimization, and improves the optimization quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of a multi-objective optimization arrangement method of sensors for health monitoring of bridge erecting machines provided by the present invention;

[0037] Figure 2 Schematic diagram of a finite element model of a bridge erecting machine in an embodiment of the present invention;

[0038] Figure 3 It is a calculation flow chart of performing multi-objective optimization arrangement of sensors according to the multi-objective improved Hippo optimization algorithm in an embodiment of the present invention;

[0039] Figure 4 is a schematic diagram of the Pareto front calculated by the multi-objective optimization method of sensors for structural health monitoring of bridge erection machines in an embodiment of the present invention;

[0040] Figure 5 It is a schematic diagram of a multi-objective optimization arrangement scheme of sensors for structural health monitoring of a bridge erecting machine in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0042] See also Figure 1 The present invention establishes a finite element model of a bridge erection machine, determines the candidate sensor layout points, and performs modal analysis to determine the number of modes. Then, the three objective functions of modal independence, vibration response intensity, and damage sensitivity are determined, and a neural network is trained through a training data set of the layout points. The neural network is combined with the Hippo optimization algorithm, and then combined with the Pareto theory to perform multi-objective optimization of sensors for bridge erection machine structural health monitoring, and obtain the Pareto frontier solution set. Finally, the solution set is measured by the standardized absolute difference method, the optimal solution is selected, and the sensor layout position and number for bridge erection machine structural health monitoring are determined.

[0043] The multi-objective optimization arrangement method of sensors for health monitoring of bridge erecting machines provided by the present invention mainly comprises the following steps:

[0044] Step 1: Perform modal analysis on the finite element model of the bridge erection machine to obtain the modal vibration matrix and determine the number of modes.

[0045] Specifically, a finite element model of the bridge-building machine is established: the layout positions of the candidate sensors are determined, a modal analysis is performed on the finite element model of the bridge-building machine to obtain the modal vibration matrix, and the number of modes is selected.

[0046] In this implementation, according to the needs of structural health monitoring of the bridge crane, three key evaluation indicators, namely modal independence, vibration response intensity and damage sensitivity, are determined as the objective function. Among them, the modal independence index is evaluated by the modal vector space intersection angle. When the modal vector space intersection angle is larger, the modal independence is better and the original modal characteristics can be retained more; the vibration response intensity index reflects the sensor's ability to respond to structural vibration at the monitoring point. The higher the response intensity, the stronger the sensor's sensitivity to vibration; the damage sensitivity index reflects the sensor's sensitivity to structural damage. The higher the sensitivity, the more effectively the sensor can detect structural damage.

[0047] Step 2: Use the data set to train the neural network model corresponding to the three objective functions; the three objective functions are modal independence, vibration response intensity and damage sensitivity; the data set is composed of multiple sensor layout points and their corresponding three objective function values, and the three objective function values ​​are calculated based on the obtained modal vibration matrix and the number of modes.

[0048] A data set containing 10,000 sensor deployment points is generated based on the expressions of the above three objective functions. A neural network model is trained based on these data, with the input being the sensor deployment location and the output being the values ​​of the three objective functions; the neural network model is validated to ensure that it can accurately predict the effect of sensor deployment, and the validated neural network model is saved.

[0049] Step three, use the three trained neural network models as the fitness function of the Hippo optimization algorithm respectively, and improve the Hippo optimization algorithm in combination with the Pareto theory to obtain the improved Hippo optimization algorithm; obtain the final Pareto frontier solution set through the improved Hippo optimization algorithm, specifically: first calculate all solutions based on the initial sensor layout points in one cycle, and then obtain the Pareto frontier solution set of the current cycle through the Pareto theory, and then use the current Pareto frontier solution set as the initial solution in the next cycle to obtain the final Pareto frontier solution set.

[0050] Step 4: Use the standardized absolute value method of difference to measure the final Pareto frontier solution set to obtain the optimal solution, that is, the optimal layout of the sensor points.

[0051] Specifically, the standardized absolute difference method calculates the comprehensive difference of each solution by standardizing each objective function value, and selects the solution with the smallest difference as the optimal solution.

[0052] The present invention also provides a multi-objective optimization arrangement system of sensors for monitoring the structural health of a bridge erecting machine, which comprises:

[0053] The structural response data acquisition module is used to establish a finite element model of a preset bridge crane structure where sensors can be deployed. Through modal analysis, the modal vibration matrix is ​​obtained to understand the natural vibration characteristics of the structure, including natural frequency and vibration shape data.

[0054] The objective function construction module is used to use the three key evaluation indicators to perform optimization analysis and determine the three key evaluation indicators of modal independence, vibration response intensity and damage sensitivity as the objective function.

[0055] The multi-objective Hippo optimization algorithm solution module uses the neural network model as the fitness function of the optimization algorithm and combines it with the Pareto theory to obtain the Pareto frontier solution set.

[0056] The optimal solution selection module is used to measure the solution set using the standardized absolute difference method, select the optimal solution, and determine the sensor layout location and quantity for bridge crane structural health monitoring.

[0057] The present invention is further described in detail below with reference to specific embodiments.

[0058] In this embodiment, a multi-objective optimization arrangement method of sensors for structural health monitoring of a bridge erection machine is proposed, comprising the following steps:

[0059] S1: Establish the finite element model of the bridge erecting machine: The finite element model of the bridge erecting machine is as follows: Figure 2 As shown, the candidate sensor layout points are determined, the modal analysis is performed on the finite element model of the bridge erection machine, the modal vibration matrix is ​​obtained, the natural vibration characteristics of the structure, including the natural frequency and vibration shape, are understood, and the number of modes is selected.

[0060] Among them, the undamped free vibration equation of the n-degree-of-freedom system is:

[0061] (K-λ i M)Φ i =0

[0062] Among them, K represents the structural stiffness matrix, M represents the structural mass matrix, and λ i Represents the i-th order eigenvalue of the structure, whose value is equal to the i-th order free vibration frequency ω i square, Φ is the mode shape matrix, Φ i Represents the i-th order eigenvector or mode shape of the structure.

[0063] Determine the candidate sensor deployment points, such as Figure 2 As shown in the finite element model of the bridge erecting machine in the embodiment of the present invention, there are 7598 units and 16230 nodes in the model. In the specific embodiment of the present invention, the sensor layout of the main beam of the bridge erecting machine is optimized.

[0064] Considering the number of nodes, 40 measuring points were selected along the upper side of the flange of the main beam of the bridge crane and the bottom plate of the main beam. Each node has three degrees of freedom in the X, Y, and Z directions. A total of 80 candidate sensor layout points with 240 degrees of freedom were determined.

[0065] The number of sensors determined in the specific embodiment of the present invention is 15.

[0066] Select the number of modes: perform modal analysis on the established finite element model to determine the first 30 natural frequencies and modal vibration shapes Φ of the finite element model. 30 .

[0067] Among them, the Fisher information matrix Q is obtained based on the mode shape matrix:

[0068]

[0069]

[0070] Where m is the selected mode order.

[0071] The rate of change (ROC) of the 2-norm (∥Q∥2) of the Fisher information matrix Q is used as the evaluation criterion, and a reasonable number of modes is selected according to the changing trend of the rate of change.

[0072] in,

[0073]

[0074] The ROC value is calculated, and the first 20 modes are finally selected to form the modal vibration matrix according to the changing trend of ROC.

[0075] S2: Determine the objective function. According to the requirements of the bridge crane structural health monitoring, three key evaluation indicators, namely modal independence, vibration response intensity and damage sensitivity, are determined as the objective function.

[0076] The modal vector space intersection angle is used to evaluate the modal independence index. The larger the modal vector space intersection angle, the better the modal independence and the more the original modal characteristics can be retained. The vibration response intensity index reflects the sensor's ability to respond to structural vibration at the monitoring point. The higher the response intensity, the stronger the sensor's sensitivity to vibration. The damage sensitivity index reflects the sensor's sensitivity to structural damage. The higher the sensitivity, the more effectively the sensor can detect structural damage.

[0077] In the evaluation of modal independence index, the Modal Assurance Criterion (MAC) effectively measures the independence between the modes measured at the measuring point by evaluating the size of the intersection angle of the modal vector space:

[0078]

[0079] Among them, MAC ij represents the off-diagonal elements in the MAC matrix, and i≠j, Φ i With Φ j represents the i-th and j-th order mode vibration vectors. ij = 0, indicating that the i-th and j-th order mode vibration vectors are completely orthogonal; when MAC ij = 1, indicating that the i-th and j-th order mode vibration vectors completely overlap. So MAC ij The smaller the value, the better the independence between the i-th and j-th order modes, and the easier it is to distinguish the two. Therefore, when arranging sensors, the maximum value of the non-diagonal element of the MAC matrix should be minimized as much as possible. At this time, the intersection angle of the most relevant two-order modal vector space is the largest, and the independence of each order mode is also maximized.

[0080] The magnitude of the modal kinetic energy can reflect the contribution of the node to the modal vibration mode. The greater the contribution, the greater the structural dynamic response and the higher the response signal-to-noise ratio. The vibration response intensity index reflects the sensor's ability to respond to structural vibration at the monitoring point. The higher the response intensity, the more sensitive the sensor is to vibration.

[0081] The modal kinetic energy criterion (MKE) is:

[0082] MKE=Φ T MΦ

[0083] Where Φ represents the modal vibration matrix and M represents the structural mass matrix.

[0084] Among them, the average modal kinetic energy (Average Modal Kinetic Energy, AMKE j ) The modal kinetic energy contributed by all vibration modes corresponding to the j-th degree of freedom of the modal kinetic energy matrix is ​​averaged to reflect the magnitude of the average modal kinetic energy corresponding to this degree of freedom:

[0085]

[0086] Where N is the total modal order, AMKE j Represents the average modal kinetic energy corresponding to the jth degree of freedom, MKE ji Represents the modal kinetic energy corresponding to the i-th vibration mode of the j-th degree of freedom.

[0087] The damage sensitivity index reflects the sensitivity of the sensor to structural damage. The higher the sensitivity, the more effectively the sensor can detect structural damage.

[0088] Among them, the Fisher information matrix Q based on damage sensitivity s for:

[0089] Q s =S T S

[0090] in,

[0091]

[0092] Where S i represents the i-th order damage sensitivity, L represents the total number of structural elements, Kt represents the element stiffness matrix of the t-th element, λ i Represents the i-th order eigenvalue of the structure.

[0093] Decompose according to the degree of freedom so that Q s It is equal to the sum of the damage Fisher information matrices corresponding to each degree of freedom:

[0094]

[0095] Among them, the size of the matrix trace value is used to evaluate the damage sensitivity of each degree of freedom:

[0096]

[0097] When the trace value trace(Qsj) of the j-th degree of freedom corresponding to the measuring point is larger, it means that the measuring point is more sensitive to damage.

[0098] According to the establishment of the three key evaluation indicators of modal independence, vibration response intensity and damage sensitivity, the maximization problem is transformed into a minimization problem and the objective function is constructed.

[0099] minf=(f1,f2,f3)

[0100] in,

[0101]

[0102] S3: Establish a neural network model and obtain a data set containing 10,000 deployment points through calculation. According to the expressions of the above three objective functions, establish a neural network model with the input as the sensor deployment location and the output as the values ​​of the three objective functions. Verify the validity of the model to ensure that it can accurately predict the effect of sensor deployment, and save the verified model.

[0103] The code in the present invention is written in MATLAB language to construct a BP neural network model. The neural network layer input adopts a 0-1 numbering method. If there are n sensor positions to be measured, n units are generated in the input layer. When inputting data information, the value at the sensor position is set to 1, and the rest is 0, such as inputting 1 in the 1st, 2nd, 5th, 6th, and 8th units of the input layer, and inputting 0 in other units.

[0104] A three-layer feedforward BP neural network model was used. The model started with n input neurons, contained two hidden layers, and an output layer with 1 neuron. Each layer contained 60 neurons. The features were extracted by Sigmoid activation function and nonlinear transformation was performed. The mean square error (MSE) was used as the loss function, the learning rate was 0.01, and the root mean square error was 1×10 -10 The number of iterations is 1000. After the neural network model is trained, the model is evaluated by the model credibility index R2. The closer R2 is to 1, the better the performance of the neural network model, the higher the credibility, and the higher the accuracy.

[0105] As shown in the construction of the objective function in the previous steps, three BP neural network models corresponding to the objective functions f1, f2, and f3 are constructed to calculate the objective function values, which are T1, T2, and T3 respectively.

[0106] By randomly selecting and calculating 10,000 data of the finite element model of the bridge erection machine, the obtained data set is randomly shuffled, 8,000 data are used as the training set, and 2,000 data are used as the test set. The three constructed BP neural network models are trained. After the training is completed, the performance of the three neural network models is tested by 2,000 test sets respectively, and the accuracy rate is 95.05%. The model credibility index R 2 The model was evaluated and the credibility index value was 0.9808. The performance of the neural network model was good, and then the model was saved.

[0107] S4: Optimization algorithm selection, introducing the trained neural network model as the fitness function, improving the Hippo optimization algorithm, and using the Hippo optimization algorithm combined with the neural network model to solve the multi-objective function established in S3. Combined with the concept of Pareto optimal solution, the population is quickly non-dominated sorted, the algorithm is iterated multiple times, and the final Pareto frontier solution set is calculated.

[0108] Specifically, Figure 3 The calculation flow chart of the multi-objective optimization arrangement of sensors by the multi-objective improved Hippo optimization algorithm is shown in the following specific steps:

[0109] S41: Generate an initial population, which is encoded in floating point form. The population size is 50, the search agent dimension is 240, each candidate position is represented by a floating point number between 0 and 1, and then the positions corresponding to the first 15 largest values ​​are selected as sensor placement positions.

[0110] The three neural networks trained by S3 are used to calculate the values ​​of the three fitness functions: modal independence, vibration response intensity, and damage sensitivity. The Pareto front is calculated using fast dominance sorting.

[0111] S42: Perform exploration phase, update the position of the search agent, and update the position from search agent 1 to search agent 25.

[0112] Furthermore, the values ​​of parameters for position update such as I1, I2, Alfa, A, B are calculated. X_P1(i,:), X_P2(i,:) are compared with X(i,:). If X_P1(i,:), X_P2(i,:) dominates X(i,:), or complementarily dominates, X_P1(i,:), X_P2(i,:) are retained.

[0113] S43: Search agent defense against predators stage, defense is performed from search agent 26 to search agent 50.

[0114] Among them, the positions of predators are randomly generated and dominance comparisons are performed.

[0115] S44: The search agent moves away from the predator stage, looping from search agent 1 to search agent 50.

[0116] S45: Update the initial population, calculate the Pareto frontier based on the initial population and the agents retained in this cycle, and retain the first 50 search agents as the initial population for the next cycle. Set the number of iterations to 500, execute step S42 until the number of iterations is reached, save the optimized population, and calculate the Pareto frontier results as follows: Figure 4 As shown, the Pareto front solution set is obtained.

[0117] S5: Select the optimal solution. After obtaining the Pareto front solution set, the standardized absolute difference method is used to measure the solution set and select the optimal solution. The standardized absolute difference method calculates the comprehensive difference of each solution by standardizing the values ​​of each objective function, and selects the solution with the smallest difference as the optimal solution.

[0118] For each objective function, its value in the Pareto front solution set is normalized, where:

[0119]

[0120] Here, each objective function value is subtracted from the minimum value of the objective among all solutions, and then divided by the difference between the maximum and minimum values ​​of the objective among all solutions.

[0121]

[0122] Where n is the number of objective functions, is the average of the normalized values ​​of the ith objective in all solutions.

[0123] Select the comprehensive difference D(x j ) is the smallest solution as the optimal solution.

[0124] The multi-objective optimization arrangement scheme of sensors for monitoring the structural health of a bridge erecting machine obtained through the above steps in the specific embodiment of the present invention is as follows: Figure 5 shown.

[0125] The present invention also provides a health monitoring method for a bridge-building machine. The health monitoring method first adopts the above-mentioned multi-objective optimization layout method of sensors for health monitoring of the bridge-building machine to obtain the optimal layout plan of sensor points; then, multiple sensors are arranged according to the optimal layout plan, and then the health monitoring of the bridge-building machine is performed through the multiple sensors.

[0126] The present invention also provides a sensor multi-objective optimization layout system for bridge-building machine health monitoring, which includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it executes the sensor multi-objective optimization layout method for bridge-building machine health monitoring as described above.

[0127] The present invention also provides a computer-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the multi-objective optimization arrangement method of sensors for bridge-building machine health monitoring or / and the health monitoring method of the bridge-building machine as described above.

[0128] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-objective optimization arrangement method of sensors for health monitoring of bridge erection machines, characterized in that: The method comprises the following steps: Step 1: Perform modal analysis on the finite element model of the bridge erecting machine to obtain the modal vibration matrix and determine the number of modes; Step 2: Using a data set to train a neural network model corresponding to three objective functions; wherein the three objective functions are modal independence, vibration response intensity, and damage sensitivity; the data set is composed of a plurality of sensor arrangement points and their corresponding three objective function values, and the three objective function values ​​are calculated based on the obtained modal vibration matrix and the number of modes; Step three, the three trained neural network models are used as the fitness functions of the Hippo optimization algorithm respectively, and the Hippo optimization algorithm is improved in combination with the Pareto theory to obtain the improved Hippo optimization algorithm; the final Pareto frontier solution set is obtained by the improved Hippo optimization algorithm, and then the optimal solution is obtained, that is, the optimal layout plan of the sensor points.

2. The multi-objective optimization arrangement method of sensors for health monitoring of bridge erecting machines according to claim 1 is characterized in that: When the final Pareto frontier solution set is obtained by improving the Hippo optimization algorithm, all solutions are first calculated based on the initial sensor layout points in one cycle, and then the Pareto frontier solution set of the current cycle is obtained through the Pareto theory. In the next cycle, the current Pareto frontier solution set is used as the initial solution to obtain the final Pareto frontier solution set.

3. The multi-objective optimization arrangement method of sensors for health monitoring of bridge erection machines according to claim 1, characterized in that: The standardized absolute difference method is used to measure the final Pareto frontier solution set to obtain the optimal solution.

4. The multi-objective optimization arrangement method of sensors for health monitoring of bridge erecting machines according to claim 1 is characterized in that: The Fisher information matrix Q is obtained based on the mode shape matrix: Where m is the selected modal order; The ROC rate of change of the 2-norm (∥Q∥2) of the Fisher information matrix Q is used as the evaluation criterion, and a reasonable number of modes is selected according to the trend of the change rate.

5. The multi-objective optimization arrangement method of sensors for health monitoring of bridge erection machines according to claim 1, characterized in that: The modal vector space intersection angle is used to evaluate the modal independence index; the vibration response intensity index reflects the sensor's ability to respond to structural vibration at the monitoring point; and the damage sensitivity index reflects the sensor's sensitivity to structural damage.

6. The multi-objective optimization arrangement method of sensors for health monitoring of bridge erecting machines according to any one of claims 1 to 5, characterized in that: In the evaluation of modal independence index, the modal assurance criterion MAC effectively measures the independence between the modes measured at the measuring point by evaluating the size of the intersection angle of the modal vector space: Among them, MAC ij represents the off-diagonal elements in the MAC matrix, and i≠j, Φ i With Φ j Represents the i-th and j-th order mode vibration vectors; when MAC ij = 0, indicating that the i-th and j-th order mode vibration vectors are completely orthogonal; when MAC ij When =1, it indicates that the i-th and j-th order mode vibration vectors completely overlap.

7. The multi-objective optimization arrangement method of sensors for health monitoring of bridge erecting machines according to claim 6 is characterized in that: Fisher information matrix Q based on damage sensitivity s for: Q s =S T S in, Where S i represents the i-th order damage sensitivity, L represents the total number of structural elements, Kt represents the element stiffness matrix of the t-th element, λ i Represents the i-th order eigenvalue of the structure; decompose according to the degree of freedom so that Q s It is equal to the sum of the damage Fisher information matrices corresponding to each degree of freedom: Among them, the size of the matrix trace value is used to evaluate the damage sensitivity of each degree of freedom: When the trace value trace(Qsj) of the j-th degree of freedom corresponding to the measuring point is larger, it means that the measuring point is more sensitive to damage.

8. A health monitoring method for a bridge erecting machine, characterized in that: The health monitoring method first adopts the sensor multi-objective optimization layout method for bridge-building machine health monitoring described in any one of claims 1-7 to obtain the optimal layout plan of sensor points; then, multiple sensors are set according to the optimal layout plan, and then the health of the bridge-building machine is monitored by multiple sensors.

9. A multi-objective optimization arrangement system of sensors for health monitoring of bridge erecting machines, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the multi-objective optimization arrangement method of sensors for health monitoring of bridge erecting machines as described in any one of claims 1 to 7 is executed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores machine-executable instructions. When the machine-executable instructions are called and executed by the processor, the machine-executable instructions prompt the processor to implement the multi-objective optimization arrangement method of sensors for bridge-building machine health monitoring as described in any one of claims 1-7 or / and the health monitoring method for bridge-building machine as described in claim 8.

Images