A method for optimizing the arrangement of sensors in a tensile membrane structure
By optimizing the sensor positions of tensile membrane structures using the Fisher information matrix and an improved discrete particle swarm optimization algorithm, the problem of a lack of systematic theoretical basis for sensor placement was solved, enabling efficient information acquisition and structural health monitoring, and ensuring the accuracy and safety of measurements.
Patent Information
- Application Number
- CN202211570647.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-12-08
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2042-12-08
AI Technical Summary
Existing technologies lack a systematic theoretical basis for sensor placement in the health monitoring of tensile membrane structures, resulting in redundant measuring points or missing key locations, which affects the accuracy and efficiency of information acquisition. Furthermore, existing optimization methods do not fully utilize the characteristics of sensor placement optimization.
We employ Fisher's information matrix eigenvalue matrix theory, an improved discrete particle swarm optimization algorithm, and MAC modal confidence matrix theory, combined with cooling and Lanzos methods, to optimize the sensor arrangement of tensile membrane structures. The optimal number and location of sensors are obtained through nonlinear analysis and modal analysis.
This approach ensures the effectiveness and accuracy of sensor measurement signals, avoids redundant measurements, improves the signal-to-noise ratio of information, guarantees the accuracy of structural health monitoring and the effectiveness of damage identification, and reduces the risk of safety accidents.
Smart Images

Figure CN116432480B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of damage diagnosis of tensile membrane acquisition response, and particularly relates to a method for optimizing the arrangement of sensors in tensile membrane structures. Background Technology
[0002] Membrane structures, a new type of lightweight and flexible spatial structure system that emerged in the mid-to-late 20th century, possess numerous advantages such as large span, light weight, diverse artistic designs, environmental friendliness and energy efficiency, good mechanical properties, strong self-cleaning ability, short construction period, and excellent economic efficiency, earning them the reputation of "modern green buildings for the 21st century." Like other commonly used building structures, large-span cable-membrane spatial structures are also susceptible to damage accumulation and resistance decay due to external environmental loads, such as environmental erosion, material aging, load fatigue effects, and sudden overload effects. This reduces their service life, increases the risk of damage from natural disasters, and can even lead to catastrophic accidents. Large-span cable-membrane spatial structures are often local landmarks; their destruction can cause significant casualties and huge economic losses, as well as extremely negative social impacts. Optimizing the placement of sensors for structural health monitoring can address these problems and risks.
[0003] Optimized sensor placement is a crucial aspect of structural health monitoring, as it collects essential data for structural condition assessment. Inappropriate sensor placement can lead to ineffective or redundant data acquisition, hindering the acquisition of comprehensive structural response information and preventing the effective extraction of useful structural information from noise. This directly impacts the accuracy of structural health assessment. Sensor optimization techniques aim to construct an optimal scheme for sensor types, quantities, and locations to better acquire dynamic response information from the structure, thereby achieving an optimal balance between sensor cost and system monitoring performance.
[0004] Currently, the placement of sensors in membrane structure health monitoring is mainly determined based on the understanding and experience of structural vibration characteristics. This qualitative approach, lacking systematic theoretical basis and analysis, easily leads to resource waste due to redundant sensors and information omissions and inaccurate measurements due to missing key sensor locations. Therefore, to improve the signal-to-noise ratio of collected information and the efficiency of structural modal parameter identification, it is necessary to use reasonable modeling and appropriate optimization methods to determine the optimal number and placement of sensors for tensile membrane structures. Furthermore, regarding the optimization of sensor placement using intelligent algorithms, existing methods mostly directly use the basic formulas of optimization algorithms for calculation, rarely considering algorithm improvements based on the characteristics of the sensor optimization placement problem model to obtain better placement results. Based on this, this patent proposes a method for optimizing the placement of sensors in tensile membrane structures. Summary of the Invention
[0005] The purpose of this invention is to provide a method for optimizing the arrangement of sensors in a tensile membrane structure, thereby solving the problems mentioned in the background art.
[0006] This invention protects a method for optimizing the arrangement of sensors in a tensile membrane structure, the method comprising the following steps:
[0007] S01: Establish a finite element model of the tensile membrane structure, and perform nonlinear analysis on it using the cooling method and the support lifting method to complete the application of prestress to achieve the overall tensioning process of the membrane surface; use the Lanzos method and mass normalization theory to perform modal analysis on the tensile membrane structure to obtain the modal vibration data of the tensile membrane structure for the optimization of the arrangement of membrane structure sensors.
[0008] S02: Based on the Fisher information matrix eigenvalue matrix theory, two methods are used to determine the number of modes for optimal sensor arrangement in a tensile membrane structure: the Fisher information matrix trace change rate and the information matrix L2 norm difference.
[0009] S03: Determine the number of sensors for optimal sensor arrangement in tensile membrane structures based on the mode-independent correlation criterion in MAC mode confidence matrix theory.
[0010] S04: The sensor position optimization of the tensile membrane structure is completed by using an improved discrete particle swarm optimization algorithm based on a random strategy and combined with the MAC off-diagonal element root mean square fitness function.
[0011] S05: Evaluate the quality of sensor optimization layout results for tensile membrane structures using the MAC modal confidence criterion.
[0012] As a further aspect of the present invention, in S01, it is assumed that the mode shape after the model mass normalization is as follows: The initial mode shape is , We can obtain:
[0013] ;
[0014] when At that time, the quality normalization is satisfied, therefore:
[0015] ;
[0016] Given all degrees of freedom of the structure, the constructed first n order Fisher information matrix can be expressed as:
[0017] ;
[0018] In the formula: The Fisher information matrix is based on the mode shape; The modal shape matrix of the structure;
[0019] Considering the orthogonality of the mode shapes, the above equation can be rewritten as:
[0020] ;
[0021] in where 1≤n≤i;
[0022] Considering the normalized mode shape, integrating the above equations, we get:
[0023] .
[0024] As a further embodiment of the present invention, in S02, it is assumed that the structure has n modal shape vectors (1≤i≤n-1), where the L2 norm is... The modal proximity is defined as the absolute value of the difference between the L2 norms of two adjacent mode shape vectors, i.e.:
[0025] ;
[0026] The distance between matrices is measured by norm. When the modal closeness is lower than the preset closeness, one of the modes is retained as the mode shape to be used for sensor optimization, and it is compared with the next mode until the optimization requirements are met.
[0027] The number of target modes is selected by the rate of change of the Fisher information matrix trace, i.e.:
[0028] .
[0029] As a further aspect of the present invention, in step S03, according to matrix theory, the modal vector matrix is... The column pivots are used for QR decomposition, and the MAC criterion is used as the judgment standard:
[0030]
[0031] In the formula: E is the permutation matrix; n is the number of degrees of freedom to be measured; M is the order of the monitored mode shape; ;and , The structural node positions corresponding to the non-zero degrees of freedom of the matrix can be used as the initial measurement points of the sensor; when R is decomposed into When combining column vectors, ( (corresponding to) The rows (columns) are A subset of row vectors with a larger norm degree of freedom;
[0032] Suppose that the QR decomposition yields The subset of the modal matrix composed of the residual degrees of freedom of the structure is , record Received The largest off-diagonal element of the matrix is ,Increase After the kth degree of freedom Become :
[0033] ;
[0034] ;
[0035] ;
[0036] The above calculations are from Find one that can make Reduce the maximum degree of freedom and place its corresponding value into In the end, the graph of the change between the maximum off-diagonal element value of the MAC matrix and the number of measurement points is obtained, and the number of sensors is determined based on the graph.
[0037] As a further aspect of the present invention, the optimization of the sensor placement location for the tensile membrane structure using an improved discrete particle swarm optimization algorithm based on a random strategy combined with the MAC off-diagonal root mean square fitness function specifically includes:
[0038] Initialize the basic parameters of the particle swarm, setting the population size, maximum number of iterations, inertia weight, learner factor, and the maximum and minimum values of velocity and position;
[0039] Initialize the local optimum and the historical optimum;
[0040] Particle swarm velocity update;
[0041] Replace the sensor node number in the velocity update based on the weight judgment condition rand[0,1) < w;
[0042] Based on the self-learning factor judgment condition rand[0,1) < c1, replace the sensor node number in the velocity update;
[0043] Based on the social learning factor judgment condition rand[0,1) < c2, replace the sensor node number in the velocity update;
[0044] Compare the local optimal solution with the global optimal solution among all particles. If the local optimal solution is better than the global optimal solution, replace the sensor node number in the position update and perform the position update.
[0045] The computation stops when the number of iterations reaches the preset maximum number of iterations, and the global optimal solution and the optimal sensor placement node positions are output; otherwise, the process proceeds to the particle swarm velocity update step.
[0046] As a further aspect of the present invention, in step S05, when all off-diagonal positions of the calculated MAC mode confidence matrix are less than 0.25, it is determined that the sensor optimization arrangement scheme meets the set optimization requirements.
[0047] The beneficial effects of this invention are:
[0048] (1) This invention takes the optimized arrangement of sensors in tensile membrane structures as the starting point. This method can quickly determine the optimal layout scheme of sensors, ensure the effectiveness of the measurement signals of each sensor, enable the limited number of sensors to obtain effective and non-redundant measurement information, avoid the omission and mismeasurement of local modes, and improve the accuracy of sensor measurement and the effectiveness of information.
[0049] (2) This invention minimizes the maximum value of the off-diagonal elements of the MAC mode confidence matrix, which can distinguish different modes to the greatest extent, ensure that the requirements for online monitoring and damage identification and diagnosis of tensile membrane structures are met to the maximum extent, and prevent the risk of safety accidents caused by damage to tensile membrane structures. It can be widely applied in engineering practice and has good applicability to such membrane structures. Attached Figure Description
[0050] Figure 1 This is a main flowchart of a sensor optimization arrangement method for a tensile membrane structure.
[0051] Figure 2 This is a flowchart illustrating the determination of modal order and number of measuring points in a sensor optimization arrangement method for a tensile membrane structure.
[0052] Figure 3 This is a flowchart of sensor position optimization for tensile membrane structures based on an improved discrete particle swarm optimization algorithm;
[0053] Figure 4 This is an optimal measurement point distribution diagram of the sensor optimization arrangement method for a tensile membrane structure according to the present invention in a planar biaxial tensile membrane structure. Detailed Implementation
[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0055] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.
[0056] This invention provides a sensor optimization arrangement method for tensile membrane structures, specifically a sensor optimization arrangement method for planar bidirectional tensile membrane structures based on Fisher's information matrix and a stochastic strategy-improved discrete particle swarm optimization algorithm. Taking the optimization arrangement of sensors in tensile membrane structures as a starting point, this method proposes a sensor optimization arrangement method for tensile membrane structures based on Fisher's information matrix and a stochastic strategy-improved discrete particle swarm optimization algorithm. This method can quickly determine the optimal sensor layout scheme, ensuring the effectiveness of the measurement signals from each sensor. It allows a limited number of sensors to obtain effective, non-redundant measurement information, avoiding missed and false measurements of local modes, and improving the accuracy and effectiveness of sensor measurements. Simultaneously, it minimizes the maximum value of the off-diagonal elements of the MAC mode confidence matrix, maximizing the differentiation of different modes and ensuring that the requirements for online monitoring and damage identification and diagnosis of tensile membrane structures are met to the greatest extent. This helps prevent safety risks caused by damage to tensile membrane structures and can be widely applied in engineering practice, showing good applicability to such membrane structures.
[0057] like Figure 1 and Figure 2 The diagram shown is a main flowchart of a sensor optimization arrangement method for a tensile membrane structure according to an embodiment of the present invention. Step 01: Establish a finite element model of the planar bidirectional tensile membrane structure, and perform nonlinear analysis on it using a cooling method to complete the application of prestress to the membrane structure, thereby realizing the overall tensioning process of the membrane surface. Next, modal analysis of the tensile membrane structure is performed using the Lanzos method and mass normalization theory, and finally, the modal vibration data of the tensile membrane structure are obtained for the optimized arrangement of the membrane structure sensors.
[0058] Specifically, the planar biaxially tensioned membrane structure is drawn using ABAQUS finite element simulation software. This discussion focuses on planar biaxially tensioned membrane structures with four free sides; therefore, only the membrane material properties need to be set, including material density, elastic modulus, and membrane thickness. Next, triangular membrane elements are used to mesh the planar biaxially tensioned membrane structure, and rigid constraint boundary conditions are set at the four corner nodes. Prestress is applied to the membrane surface using a cooling method, and calculations are performed using nonlinear statics and general methods. Then, modal analysis is performed on the prestressed planar biaxially tensioned membrane structure, selecting the Lanzos method and mass normalization method, and setting the number of modal orders to the first 100. Finally, the calculation is submitted. After the calculation, the first 100 modal shape data of the planar biaxially tensioned membrane structure are exported, with each modal shape saved as a separate TXT file, totaling 100 modal shape text data. Of course, the implementation is not limited to ABAQUS software; other higher versions or other finite element analysis software can also be used.
[0059] Step 02: Based on the Fisher information matrix eigenvalue matrix theory, the modal order of the optimized arrangement of sensors in the tensile membrane structure is determined by two methods: the difference of the second norm of the Fisher information matrix and the rate of change of the trace of the information matrix.
[0060] Specifically, using Matlab software, formulas (6) and (7) in the technical solution are programmed. Using the text of the first 100 modal shapes of the aforementioned planar biaxial tensile membrane structure as the data source, the variation law between the Fisher information matrix trace change rate and the Fisher information matrix L2 norm difference and the first 100 modal orders of the planar biaxial tensile membrane structure can be calculated. The inflection points of these two curves are marked and used as the modal orders selected for the optimal arrangement of sensors in the tensile membrane structure. Of course, modal orders after the inflection points can also be flexibly selected according to research needs to study the influence of different modal order selections on the optimal arrangement of sensors.
[0061] Step 03: Based on the mode shape independent correlation criterion in the MAC modal confidence matrix theory, determine the number of sensors for optimal sensor arrangement in the tensile membrane structure.
[0062] Step 04: Optimize the sensor placement location of the tensile membrane structure by using an improved discrete particle swarm optimization algorithm based on a stochastic strategy and combining it with the MAC off-diagonal element root mean square fitness function.
[0063] Specifically, to avoid the limitations of the original particle swarm optimization algorithm in sensor optimization, this patent proposes an improved discrete particle swarm optimization algorithm based on a stochastic strategy, such as... Figure 3 As shown. In this improved discrete particle swarm optimization algorithm, variables and are respectively the particles used for parallel optimization. Assuming that for a structure with m degrees of freedom, there are n sensor points, then the dimensions of variables and particles are (1, 2, 3, ..., n), and the search domain of the particles is [1, m]. During initialization, the improved discrete particle swarm optimization algorithm first randomly selects n integer solutions from the integer domain [1, m] to form the initial search particle positions. To avoid duplicate nodes when selecting measurement points for each particle initialization, a statement to delete duplicate nodes is set here, and the sampling is repeated. Based on the initialized particles, the fitness function is used to calculate the local optimum and global optimum of the algorithm at this time, thus completing the initialization settings of the improved discrete particle swarm optimization algorithm.
[0064] Furthermore, the particle swarm optimization algorithm is improved so that when optimizing sensor positions in a discrete model of a membrane structure, different judgment statements are used to update velocity and position. The sensor position optimization follows these steps:
[0065] Step 11: Initialize the basic parameters of the particle swarm, setting the population size, maximum number of iterations, inertia weight, learner factor, and the maximum and minimum values of velocity and position;
[0066] Step 12: Initialize the local optimum and historical optimum;
[0067] Step 13: Particle swarm velocity update;
[0068] Step 131: Replace the sensor node number in the velocity update according to the weight judgment condition;
[0069] Step 132: Replace the sensor node number in the speed update based on the self-learning factor judgment condition;
[0070] Step 133: Replace the sensor node number in the speed update based on the social learning factor judgment condition;
[0071] Step 134: Submit the particle into the fitness function to calculate the particle function value, compare the local optimal solution with the global optimal solution among all particles, and if the local optimal solution is better than the global optimal solution, replace the sensor node number in the position update and perform position update;
[0072] Step 135: Stop the operation when the number of iterations reaches the preset maximum number of iterations, and output the global optimal solution and the optimal sensor placement node position; otherwise, go to step 13.
[0073] Based on the above analysis steps, Matlab code was written to perform iterative calculations using the first 100 modal vibration modes of the aforementioned planar biaxial tensile membrane structure as the data source. When the improved discrete particle swarm optimization algorithm reaches its maximum iteration count, the code terminates and outputs the optimal sensor placement for this optimization. The sensor placement effect for the planar biaxial tensile membrane structure is as follows: Figure 4 As shown.
[0074] Step S05: Use the MAC modal confidence criterion to evaluate the quality of the sensor optimization arrangement results for the tensile membrane structure. When the off-diagonal positions of the calculated MAC modal confidence matrix are all less than 0.25, it indicates that the sensor optimization arrangement scheme is better.
[0075] Specifically, by substituting the number and location information of sensor placement points for the tensile membrane structure obtained from the sensor placement method based on the Fisher information matrix and a stochastic strategy-improved discrete particle swarm optimization algorithm into the MAC modal confidence criterion evaluation code written in MATLAB, the structural MAC modal confidence matrix values corresponding to the placement locations of the sensor points can be calculated. When all off-diagonal values of the MAC matrix are less than 0.25, it indicates that the sensor placement optimization scheme is effective. Of course, theoretical analysis shows that the smaller the off-diagonal values of the MAC matrix, the better the sensor placement optimization scheme is.
[0076] 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 within the protection scope of the present invention.
Claims
1. A method of sensor optimization placement for a tensile membrane structure, characterized by, The method includes: S01: Establish a finite element model of the tensile membrane structure, and perform nonlinear analysis on it using the cooling method and the support lifting method to complete the application of prestress to achieve the overall tensioning process of the membrane surface; use the Lanzos method and mass normalization theory to perform modal analysis on the tensile membrane structure to obtain the modal vibration data of the tensile membrane structure for the optimization of the arrangement of membrane structure sensors. S02: Based on the Fisher information matrix eigenvalue matrix theory, two methods are used to determine the number of modes for optimal sensor arrangement in a tensile membrane structure: the Fisher information matrix trace change rate and the information matrix L2 norm difference. S03: Determine the number of sensors for optimal sensor arrangement in tensile membrane structures based on the mode-independent correlation criterion in MAC mode confidence matrix theory. S04: The sensor position optimization of the tensile membrane structure is completed by using an improved discrete particle swarm optimization algorithm based on a random strategy and combined with the MAC off-diagonal element root mean square fitness function. S05: Evaluate the quality of sensor optimization layout results for tensile membrane structures using the MAC modal confidence criterion.
2. The method for optimizing the arrangement of sensors in a tensile membrane structure according to claim 1, characterized in that, In S01, assume the mass matrix is... The mode shape after model mass normalization is The initial mode shape is , We can obtain: ; when At that time, the quality normalization is satisfied, therefore: ; Given all degrees of freedom of the structure, the first n order Fisher information matrix is represented as follows: ; In the formula: The Fisher information matrix is based on the mode shape; The modal shape matrix of the structure; Considering the orthogonality of the mode shapes, the above equation can be rewritten as: ; in where 1≤n≤i; Considering the normalized mode shape, integrating the above equations, we get: 。 3. The method for optimizing the arrangement of sensors in a tensile membrane structure according to claim 1, characterized in that, In S02, assume the structure has n modal shape vectors (1≤i≤n-1), where the L2 norm is... The absolute value of the L2 norm difference between two adjacent modal shape vectors is defined as the modal closeness. ,Right now: ; The distance between matrices is measured by norm. When the modal closeness is lower than the preset closeness, one of the modes is retained as the mode shape to be used for sensor optimization, and it is compared with the next mode until the optimization requirements are met. Rate of change of Fisher's information matrix trace The number of target modes is selected, where the Fisher information matrix trace is simply referred to as... ,Right now: 。 4. The method for optimizing the arrangement of sensors in a tensile membrane structure according to claim 1, characterized in that, In step S03, according to matrix theory, the modal vector matrix is... The column pivots are used for QR decomposition, and the MAC criterion is used as the judgment standard: ; In the formula: E is the permutation matrix; n is the number of degrees of freedom to be measured; m is the order of the monitored mode shape; ;and , The structural node positions corresponding to the non-zero degrees of freedom of the matrix are used as the initial measurement points of the sensor; when R is decomposed into When combining column vectors, ( (corresponding to) The rows or columns are A subset of row vectors with a larger norm degree of freedom; Suppose that the QR decomposition yields The subset of the modal matrix composed of the residual degrees of freedom of the structure is The element in the i-th row and J-th column of the mode shape matrix is , record Received The largest off-diagonal element of the matrix is ,Increase After the kth degree of freedom Become : ; ; ; The above calculations are from Find one that can make Reduce the maximum degree of freedom and place its corresponding value into In the end, the graph of the change between the maximum off-diagonal element value of the MAC matrix and the number of measurement points is obtained, and the number of sensors is determined based on the graph.
5. A method for optimizing the arrangement of sensors in a tensile membrane structure according to any one of claims 1-4, characterized in that, The optimization of sensor placement in tensile membrane structures using an improved discrete particle swarm optimization algorithm based on a stochastic strategy, combined with the MAC off-diagonal root mean square fitness function, specifically includes: Initialize the basic parameters of the particle swarm, setting the population size, maximum number of iterations, inertia weight, learner factor, and the maximum and minimum values of velocity and position; Initialize the local optimum and the historical optimum; Particle swarm velocity update; Based on the weight w, the sensor node number in the velocity update is replaced according to the condition rand[0,1) < w. Based on the self-learning factor c1, the sensor node number in the velocity update is replaced according to the condition rand[0,1) < c1. Based on the social learning factor c2, the sensor node number in the velocity update is replaced according to the condition rand[0,1) < c2. Compare the local optimal solution with the global optimal solution among all particles. If the local optimal solution is better than the global optimal solution, replace the sensor node number in the position update and perform the position update. The computation stops when the number of iterations reaches the preset maximum number of iterations, and the global optimal solution and the optimal sensor placement node positions are output; otherwise, the process proceeds to the particle swarm velocity update step.
6. The method for optimizing the arrangement of sensors in a tensile membrane structure according to claim 1, characterized in that, In step S05, when all off-diagonal positions of the calculated MAC mode confidence matrix are less than 0.25, it is determined that the sensor optimization arrangement scheme meets the set optimization requirements.
Citation Information
Patent Citations
Optimal arrangement method of coal mining machine rocker arm vibration sensors
CN103413185A
Gait recognition method based on bilinear joint sparsity discrimination analysis
CN105224918A
