Measuring point strain-based glass plate random load identification method for glass plank road items
Through Fourier transform and multi-objective optimization algorithm, the load parameters on the glass plate are monitored in real time, solving the problem of difficulty in monitoring the number of people on the glass plank road in the existing technology, and accurately identifying and safe controlling the loads on the glass plate are achieved.
Patent Information
- Application Number
- CN202411255373.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-09
- Publication Date
- 2024-12-06
AI Technical Summary
The prior art is difficult to monitor the number of people on each glass plate on the glass plank road in real time, resulting in insufficient safety, especially when there are many tourists, safety accidents are prone to occur.
The random load recognition method of glass plate based on measuring point strain is adopted, and the load parameters on the tempered laminated glass plate are monitored in real time, including load position and weight, through Fourier transform and multi-objective optimization algorithm, to ensure safety.
Real-time identification of the position and weight of each person on the glass plate is achieved, and the number of people on the glass plank road can be accurately monitored to ensure safety, and the method has high accuracy and robustness.
Smart Images

Figure CN119085977A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a random load identification method for a glass plate in the technical field of glass plate load identification. Background Art
[0002] The glass plank road is a transparent glass plank road for sightseeing, which is made of tempered glass suspended high on cliffs. To ensure safety, each piece of tempered glass will be pre-tested to predict how many people the entire glass plank road can accommodate at the same time. However, in actual operation, due to the large number of tourists, it is often impossible to control the actual number of people standing on the glass plank road in time, and it is also impossible to know how many people are on each glass plate in the glass plank road. This makes the glass plank road less safe, and it is easy to cause safety accidents when there are many tourists. Summary of the invention
[0003] The purpose of the present invention is to provide a random load identification method for glass plates in glass plank road projects based on measuring point strain, so as to obtain the actual load parameters of each person, so as to monitor the number of people standing on each fiberglass plate of the glass plank road in real time to ensure safety.
[0004] To achieve the above object, the present invention provides a method for identifying random loads of glass plates in glass plank projects based on measuring point strain, comprising the following steps:
[0005] Step 1, identification of glass plate load quantity based on Fourier transform;
[0006] Step 2: Identification of glass panel load parameters based on multi-objective optimization.
[0007] As a further improvement of the present invention, the specific content of step 1 is as follows:
[0008] At a detection frequency of 5-100 Hz, strain sensors are installed at four measuring points on the bottom edge of the tempered laminated glass plate to monitor the strain and obtain the strain time history in two directions of the measuring points. ix (t) and e iy (t), where i = 1 to 4, i represents the measurement point number, and there are 4 measurement points in total;
[0009] The strain time history signal e in two directions of the i-th measuring point ix (t) and e iy (t) performing Fourier transform;
[0010]
[0011] In the formula, F ix (ω ix ) and F iy (ω iy) are the Fourier transform functions of the i-th measuring point of the tempered laminated glass plate in the X and Y directions, namely the spectrum density function, referred to as spectrum, ω ix and ω iy are the peak frequencies after Fourier transformation, e is the natural index, j is the jth peak after Fourier transformation, t is the detection time, △t is the load identification window length of the tempered laminated glass panel, and e ix (t) and e iy (t) is the strain time history in two directions of the measuring point.
[0012] As a further improvement of the present invention, the specific content of step 2 is as follows:
[0013] In order to ensure that the strain numerical results of the four strain measuring points on the bottom surface of the tempered laminated glass plate are consistent with the on-site detection, there will be multiple objective functions, which makes the glass plate load identification a multi-objective optimization problem; therefore, the glass plate load identification process includes the selection of load identification parameters, the establishment of objective functions and the selection of robust optimization algorithms.
[0014] As a further improvement of the present invention, the specific contents of the selection of the identification load parameters are as follows:
[0015] The error in the load identification parameters of the glass plate comes from the characteristics of each load. The strain of the measuring point caused by each load at different positions on the glass plate is different. Therefore, each load has three parameters corresponding to it, namely the lateral position x, longitudinal position y, and self-weight w of the load. Then, if there are k loads on the glass plate, the total number of load parameters that need to be identified is 3k, which are
[0016] P=[x1 y1 w1 x2 y2 w2…x k y k w k ] T
[0017] Where P is the glass plate load parameter vector to be identified, x k is the lateral position of the kth load, y k is the longitudinal position of the kth load, w k is the kth load mass, and the superscript T is the matrix transpose.
[0018] As a further improvement of the present invention, the specific contents of establishing the objective function are as follows:
[0019] In the identification of glass plate load parameters, the objective function can be defined according to the difference between the numerical calculation results of the strain at the measuring point and the on-site detection value, and standardized. At the same time, there are multiple objective functions, which need to be controlled. The numerical calculation results of the strain in the X and Y directions of the four measuring points and the on-site detection values are used to construct the standardized objective functions g1(P) and g2(P):
[0020]
[0021] Where g1(P) and g2(P) are the constructed objective functions. is the calculated value of the X-direction strain at the i-th measuring point (i=1 to 4) on the bottom surface of the tempered laminated glass plate. It is the on-site detection value of the X-direction strain at the i-th measuring point (i=1 to 4) on the bottom surface of the tempered laminated glass plate; the superscript c represents the calculated result of the physical quantity, and the superscript m represents the measured result of the physical quantity.
[0022] As a further improvement of the present invention, there are multiple objective functions, which can be expressed as
[0023] minimize G(P)={g1(P),g2(P)} T
[0024] stΩ={P|Pl≤P≤Pu}
[0025] In the formula, minimize represents mathematical optimization, st represents the constraints in mathematical optimization; P is the glass plate load parameter vector to be identified; P l , P u and Ω are the lower limit, upper limit and feasible domain of the glass plate load parameter vector respectively; G(P) is the objective function vector of the multi-objective optimization problem. In the multi-objective optimization problem, there is usually no feasible solution that can minimize all objective functions at the same time; therefore, the multi-objective optimization problem is actually the problem of solving the set of non-inferior solutions.
[0026] As a further improvement of the present invention, a genetic algorithm is used to solve the set of non-inferior solutions, starting from multiple feasible solutions. The set of multiple feasible solutions is called a population, denoted as pop(n), where n represents the current iteration step; each individual in the population represents a feasible solution P of the multi-objective optimization problem and is represented in the form of an encoded binary string. The entire genetic algorithm consists of three genetic operators: selection, crossover and mutation.
[0027] As a further improvement of the present invention, the specific steps of solving the non-inferior solution set by genetic algorithm are as follows:
[0028] Step 1: Use a uniformly distributed random function to generate a random number containing N in the feasible domain Ω. sThe initial population of individuals P is pop(0), the population size Ns is set to 50, and P is a feasible solution;
[0029] Step 2: For the minimization problem in the identification of glass plate load parameters, calculate the fitness function values g1(P) and g2(P) of each individual P in the population pop(n);
[0030] Step 3: When the convergence condition Function tolerance 10-4 is met, the calculation of the entire genetic algorithm is stopped; otherwise, a binary tournament that can ensure the genetic diversity of the population is used to select 2 individuals randomly selected from the current population pop(n), and the tournament size is 2; the individual with the smallest fitness g1(P) and g2(P) is saved in the next generation population pop(n+1); the above process is repeated until the number of individuals in the next generation population pop(n+1) reaches the preset number, forming a new population pop(n+1);
[0031] Step 4: With ratio p c Perform an intermediate crossover between two randomly selected individuals to obtain a new population after crossover, crosspop(n+1); the crossover ratio p c Set to 1.0, so that the genes at all positions of any selected individual are crossovered, providing sufficient variability for the next population;
[0032] Step 5: With mutation probability p m Randomly select some loci of individuals in the population crosspop(n+1), change the gene values of these loci, form mutant individuals, and obtain a new population mutpop(n+1) after mutation; return the new population mutpop(n+1) to Step 2 for calculation; the mutation function depends on the constraints, and the mutation probability p m Set to 1.0.
[0033] As a further improvement of the present invention, a genetic algorithm is used to solve the multi-objective optimization problem derived from the identification of glass plate load parameters, and a Pareto frontier Ω*={P1, P2, …, P l}, Ω* represents the set of non-inferior solutions to the optimization problem obtained by solving; finally, the random load parameter matrix on the glass plate can be obtained, that is, the load position and size P = [x1y1 w1 x2 y2 w2…x k y k w k ] T .
[0034] Compared with the prior art, the beneficial effect of the present invention is that, in the first stage, Fourier transform is performed based on the time-course data of the strain monitoring of the measuring point, and the load quantity on the glass plate is identified from the spectrum diagram of the strain of the measuring point. Subsequently, based on the finite element model of the tempered laminated glass plate after verification, the strain influence surface of the four measuring points on the bottom surface of the glass plate is calculated, and the specific parameters of the load are identified based on the strain influence surface of the measuring point, and the real-time identification of the load size and position of the glass plate is realized. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 This is a diagram showing the arrangement of strain measuring points according to the present invention.
[0036] Figure 2 This is the Fourier time-frequency conversion of the strain signal of the measuring point of the present invention (5Hz sampling frequency).
[0037] Figure 3 This is the Fourier time-frequency conversion of the strain signal of the measuring point of the present invention (10Hz sampling frequency).
[0038] Figure 4 This is the Fourier time-frequency conversion of the strain signal at the measuring point of the present invention (100Hz sampling frequency).
[0039] Figure 5 This is the Pareto frontier for multi-objective optimization of the 5 Hz sampling frequency of the present invention.
[0040] Figure 6 This is the Pareto frontier for multi-objective optimization of the 10 Hz sampling frequency of the present invention.
[0041] Figure 7 This is the Pareto frontier for multi-objective optimization of the 100 Hz sampling frequency of the present invention.
[0042] Figure 8 This is the identification result of the glass plate load parameters according to the present invention. DETAILED DESCRIPTION
[0043] The present invention is further described below in conjunction with the accompanying drawings:
[0044] like Figure 1-8 The random load identification method of glass plate of glass plank road project based on measuring point strain shown in the figure includes the following steps:
[0045] Step 1, identification of glass plate load quantity based on Fourier transform;
[0046] At a detection frequency of 5-100 Hz, strain sensors are installed at four measuring points on the bottom edge of the tempered laminated glass plate to monitor the strain and obtain the strain time history in two directions of the measuring points. ix (t) and e iy(t), where i = 1 to 4, i represents the measurement point number, and there are 4 measurement points in total;
[0047] The strain time history signal e in two directions of the i-th measuring point ix (t) and e iy (t) performing Fourier transform;
[0048]
[0049] In the formula, F ix (ω ix ) and F iy (ω iy ) are the Fourier transform functions of the i-th measuring point of the tempered laminated glass plate in the X and Y directions, namely the spectrum density function, referred to as spectrum, ω ix and ω iy are the peak frequencies after Fourier transformation, e is the natural index, j is the jth peak after Fourier transformation, t is the detection time, △t is the load identification window length of the tempered laminated glass panel, and e ix (t) and e iy (t) is the strain time history in two directions of the measuring point.
[0050] Step 2: Identification of glass panel load parameters based on multi-objective optimization.
[0051] In order to ensure that the strain numerical results of the four strain measuring points on the bottom surface of the tempered laminated glass plate are consistent with the on-site detection, there will be multiple objective functions, which makes the glass plate load identification a multi-objective optimization problem; therefore, the glass plate load identification process includes the selection of load identification parameters, the establishment of objective functions and the selection of robust optimization algorithms.
[0052] The error in the load identification parameters of the glass plate comes from the characteristics of each load. The strain of the measuring point caused by each load at different positions on the glass plate is different. Therefore, each load has three parameters corresponding to it, namely the lateral position x, longitudinal position y, and self-weight w of the load. Then, if there are k loads on the glass plate, the total number of load parameters that need to be identified is 3k, which are
[0053] P=[x1 y1 w1 x2 y2 w2…x k y k w k ] T
[0054] Where P is the glass plate load parameter vector to be identified, x k is the lateral position of the kth load, y k is the longitudinal position of the kth load, w k is the kth load mass, and the superscript T is the matrix transpose.
[0055] In the identification of glass plate load parameters, the objective function can be defined according to the difference between the numerical calculation results of the strain at the measuring point and the on-site detection value, and standardized. At the same time, there are multiple objective functions, which need to be controlled. The numerical calculation results of the strain in the X and Y directions of the four measuring points and the on-site detection values are used to construct the standardized objective functions g1(P) and g2(P):
[0056]
[0057] Where g1(P) and g2(P) are the constructed objective functions. is the calculated value of the X-direction strain at the i-th measuring point (i=1 to 4) on the bottom surface of the tempered laminated glass plate. It is the on-site detection value of the X-direction strain at the i-th measuring point (i=1 to 4) on the bottom surface of the tempered laminated glass plate; the superscript c represents the calculated result of the physical quantity, and the superscript m represents the measured result of the physical quantity.
[0058] There are multiple objective functions, which can be expressed as
[0059] minimize G(P)={g1(P),g2(P)} T
[0060] stΩ={P|Pl≤P≤Pu}
[0061] In the formula, minimize represents mathematical optimization, st represents the constraints in mathematical optimization; P is the glass plate load parameter vector to be identified; P l , P u and Ω are the lower limit, upper limit and feasible domain of the glass plate load parameter vector respectively; G(P) is the objective function vector of the multi-objective optimization problem. In the multi-objective optimization problem, there is usually no feasible solution that can minimize all objective functions at the same time; therefore, the multi-objective optimization problem is actually the problem of solving the set of non-inferior solutions.
[0062] A genetic algorithm is used to solve the set of non-inferior solutions, starting from multiple feasible solutions. The set of multiple feasible solutions is called a population, denoted as pop(n), where n represents the current iteration step; each individual in the population represents a feasible solution P of the multi-objective optimization problem and is represented in the form of an encoded binary string. The entire genetic algorithm consists of three genetic operators: selection, crossover, and mutation.
[0063] The specific steps of genetic algorithm to solve the non-inferior solution set are as follows:
[0064] Step 1: Use a uniformly distributed random function to generate a random number containing N in the feasible domain Ω. sThe initial population of individuals P is pop(0), the population size Ns is set to 50, and P is a feasible solution;
[0065] Step 2: For the minimization problem in the identification of glass plate load parameters, calculate the fitness function values g1(P) and g2(P) of each individual P in the population pop(n);
[0066] Step 3: When the convergence condition Function tolerance 10-4 is met, the calculation of the entire genetic algorithm is stopped; otherwise, a binary tournament that can ensure the genetic diversity of the population is used to select 2 individuals randomly selected from the current population pop(n), and the tournament size is 2; the individual with the smallest fitness g1(P) and g2(P) is saved in the next generation population pop(n+1); the above process is repeated until the number of individuals in the next generation population pop(n+1) reaches the preset number, forming a new population pop(n+1);
[0067] Step 4: With ratio p c Perform an intermediate crossover between two randomly selected individuals to obtain a new population after crossover, crosspop(n+1); the crossover ratio p c Set to 1.0, so that the genes at all positions of any selected individual are crossovered, providing sufficient variability for the next population;
[0068] Step 5: With mutation probability p m Randomly select some loci of individuals in the population crosspop(n+1), change the gene values of these loci, form mutant individuals, and obtain a new population mutpop(n+1) after mutation; return the new population mutpop(n+1) to Step 2 for calculation; the mutation function depends on the constraints, and the mutation probability p m Set to 1.0.
[0069] The genetic algorithm is used to solve the multi-objective optimization problem derived from the identification of glass plate load parameters, and the Pareto frontier Ω*={P1,P2,…,P l}, Ω* represents the set of non-inferior solutions to the optimization problem obtained;
[0070] Finally, the random load parameter matrix on the glass plate can be obtained, that is, the load position and size P = [x1 y1 w1 x2 y2w2…x k y k w k ] T .
[0071] In the present invention, a strain sensor is provided at the center position of each of the four edges of the tempered laminated glass plate to measure the strain of the tempered laminated glass plate. The load on the tempered laminated glass plate is the pedestrian load, and the weight and position coordinates of each person are different. The strain time history curve of the measuring point on the bottom surface of the tempered laminated glass plate is detected by the strain sensor, and the strain time history signal of the measuring point is Fourier transformed, so that the frequency domain curve of the strain signal of the measuring point can reflect the characteristics of the pedestrian load. The strain of the four measuring points on the bottom edge of the tempered laminated glass plate is monitored at a detection frequency of 5-100 Hz to obtain the strain time history curve of the measuring point in two directions (X and Y directions). ix (t) and e iy (t).
[0072] The purpose of load identification is to identify the parameters of random loads on tempered laminated glass panels, that is, the size and position of random loads. Based on the strain influence surface of the measuring point on the bottom surface of the glass plate, the load parameters on the glass plate are calculated to minimize the difference between the numerical strain results of the measuring point obtained by finite element calculation and the actual strain measured on site. In order to ensure that the numerical strain results of the four strain measuring points on the bottom surface of the tempered laminated glass plate are consistent with the on-site detection, there will naturally be multiple objective functions, which makes the glass plate load identification a multi-objective optimization problem. Therefore, in the process of glass plate load identification, the selection of identification load parameters, the establishment of objective functions and the selection of robust optimization algorithms are three key steps.
[0073] The error of glass plate load identification parameters comes from the characteristics of each load. Each load at different positions on the glass plate causes different strains at the measuring point. Therefore, each load has three corresponding parameters, namely, the lateral position x, longitudinal position y, and self-weight w of the load. Then, if there are k loads on the glass plate, the total number of load parameters to be identified is 3k, and the glass plate load parameter vector to be identified is
[0074] P=[x1 y1 w1 x2 y2 w2…x k y k w k ] T
[0075] In order to monitor the safety of tempered laminated glass panels, the strains of the four measuring points on the bottom edge of the tempered laminated glass panels are monitored in real time. Therefore, in the identification of glass panel load parameters, the objective function can be defined according to the difference between the numerical calculation results of the strains at the measuring points and the on-site detection values, and standardized processing is performed. At the same time, there are multiple objective functions, and the number needs to be controlled. The numerical calculation results of the strains in the X and Y directions of the four measuring points and the on-site detection values are used to construct the standardized objective functions g1(P) and g2(P):
[0076]
[0077] There are multiple objective functions. Mathematically, the multi-objective optimization problem derived from the identification of glass plate load parameters can be expressed as
[0078] minimize G(P)={g1(P),g2(P)} T
[0079] stΩ={P|Pl≤P≤Pu}
[0080] minimize represents mathematical optimization, st represents the constraints in mathematical optimization; P is the glass plate load parameter vector to be identified; P l , P u and Ω are the lower limit, upper limit and feasible domain of the glass plate load parameter vector respectively; G(P) is the objective function vector of the multi-objective optimization problem. In the multi-objective optimization problem, there is usually no feasible solution that can minimize all objective functions at the same time; therefore, the multi-objective optimization problem is actually the problem of solving the set of non-inferior solutions.
[0081] Traditional optimization methods such as weighted summation, orthogonal boundary crossover algorithm and gradient-based optimization algorithms are difficult to solve nonlinear multi-objective optimization problems. Genetic algorithms are used to solve the global optimal solution of multi-objective optimization problems. When solving multi-objective optimization problems, genetic algorithms start with multiple feasible solutions. This set of multiple feasible solutions is called a population, denoted as pop(n), where n represents the current iteration step. Each individual in the population represents a feasible solution P to the multi-objective optimization problem and is represented in the form of an encoded binary string. The entire genetic algorithm consists of three genetic operators: selection, crossover and mutation.
[0082] The genetic algorithm is used to solve the multi-objective optimization problem derived from the identification of glass plate load parameters, and the Pareto frontier Ω*={P1,P2,…,P l}, Ω* represents the set of non-inferior solutions to the optimization problem obtained;
[0083] Finally, the random load parameter matrix on the glass plate can be obtained, that is, the load position and size P = [x1 y1 w1 x2 y2w2…x k y k w k ] T In this way, the position and weight of each person on the glass plate can be known.
[0084] In this way, it is possible to monitor in real time how many people are on each glass plate and how much each person weighs, so as to know how many people are on the entire glass plank road, and then control the number of people entering the glass plank road to ensure the safety of pedestrians.
[0085] Let's take individual walking as an example to explain it in detail.
[0086] Phase 1: Load quantity identification
[0087] Under the test condition of a person walking, the strain time history signals in the Y direction of the No. 1 and No. 3 measuring points with sampling frequencies of 5 Hz, 10 Hz, and 100 Hz are transformed by Fourier time-frequency conversion, as shown in Figures 2 to 4 As shown in the figure, it can be found that under the three detection frequencies, the frequency domain curves of the Y-direction strain signals at measuring points 1 and 3 have only one peak. This shows that within the current window duration, only one pedestrian load acts on the glass plate.
[0088] Phase 2: Load parameter identification
[0089] For the multi-objective optimization problem arising from the identification of glass plate loads, after obtaining the number of loads acting on the glass plate in the first stage, the multi-objective optimization method is used to identify the specific parameters of the loads. For the working condition where one person is walking, the multi-objective optimization method is used to obtain the Pareto optimal solution set Ω*={X1,X2,…,X 18}like Figure 5-7 As shown. For each sampling frequency, the best compromise solution still needs to be selected from the Pareto optimal solution set Ω*. According to the measured strain value of the measuring point, each non-inferior solution X is calculated. i (X i ∈Ω*,i=1to18,i∈N + ) corresponds to the objective function g k (X i )(k=1to 2,k∈N + ) value. According to each non-inferior solution X i The corresponding objective function g k (X i ) to the origin measures the quality of each non-inferior solution. In the Pareto frontier of three sampling frequencies, the best compromise solution is determined from the non-inferior solution set Ω* according to the Euclidean distance.
[0090] Under the condition of one person walking, the optimal compromise solution load parameters obtained by the three sampling frequency strain time histories are as follows: Figure 8 As shown in Figure 2, the final results of different sampling frequencies converge to the same set of load parameters, namely X = [0.4723 0.472675.15] T. This shows that the proposed load identification method has good robustness, and the sampling frequency of the strain signal has no effect on the result. Since the moment of strain selection is when the strain in the Y direction of measuring point 1 reaches its peak value, the position parameters of the pedestrian load are x=0.4723m and y=0.4726m. This is because in the walking test, the person's walking direction is from measuring point 4 to measuring point 2, and basically walks along the line connecting measuring points 4 to 2. Therefore, when the person walks to the center of the plate, the strain in the Y direction of measuring point 2 reaches its peak. In addition, under the condition of one person walking, the actual weight of the test pedestrian is 78kg, while the identified pedestrian load weight is 75.15kg. The relative error of load identification is only 3.6%, which shows that the proposed glass plate load identification method has high accuracy.
[0091] The present invention performs Fourier transformation based on the time-course data of strain monitoring at the measuring point, and identifies the amount of load on the glass plate from the spectrum of strain at the measuring point. Subsequently, based on the finite element model of the tempered laminated glass plate after verification, the strain influence surface of the four measuring points on the bottom surface of the glass plate is calculated, and the specific parameters of the load are identified based on the strain influence surface of the measuring points, thereby realizing real-time identification of the load size and position of the glass plate.
[0092] The present invention is not limited to the above-mentioned embodiments. On the basis of the technical solution disclosed herein, technicians in this field can make some substitutions and deformations to some technical features therein according to the disclosed technical content without creative labor, and these substitutions and deformations are all within the protection scope of the present invention.
Claims
1. A random load identification method for glass plates in glass plank road projects based on measuring point strain, characterized by: The following steps are included: Step 1, identification of glass plate load quantity based on Fourier transform; Step 2: Identification of glass panel load parameters based on multi-objective optimization.
2. The random load identification method for glass plates of glass plank road projects based on measuring point strain according to claim 1 is characterized by: The specific contents of step 1 are as follows: At a detection frequency of 5-100 Hz, strain sensors are installed at four measuring points on the bottom edge of the tempered laminated glass plate to monitor the strain and obtain the strain time history in two directions of the measuring points. ix (t) and e iy (t), where i = 1 to 4, i represents the measurement point number, and there are 4 measurement points in total; The strain time history signal e in two directions of the i-th measuring point ix (t) and e iy (t) performing Fourier transform; In the formula, F ix (ω ix ) and F iy (ω iy ) are the Fourier transform functions of the i-th measuring point of the tempered laminated glass plate in the X and Y directions, namely the spectrum density function, referred to as spectrum, ω ix and ω iy are the peak frequencies after Fourier transform, e is the natural index, j is the jth peak after Fourier transform, t is the detection time, △t is the load identification window length of the tempered laminated glass panel, and e ix (t) and e iy (t) is the strain time history in two directions of the measuring point.
3. The random load identification method for glass plates of glass plank road projects based on measuring point strain according to claim 2 is characterized by: The specific contents of step 2 are as follows: In order to ensure that the strain numerical results of the four strain measuring points on the bottom surface of the tempered laminated glass plate are consistent with the on-site detection, there will be multiple objective functions, which makes the glass plate load identification a multi-objective optimization problem; therefore, the glass plate load identification process includes the selection of load identification parameters, the establishment of objective functions and the selection of robust optimization algorithms.
4. The random load identification method for glass plates of glass plank road projects based on measuring point strain according to claim 3 is characterized by: The specific contents of the selection of identification load parameters are as follows: The error in the load identification parameters of the glass plate comes from the characteristics of each load. The strain of the measuring point caused by each load at different positions on the glass plate is different. Therefore, each load has three parameters corresponding to it, namely the lateral position x, longitudinal position y, and self-weight w of the load. Then, if there are k loads on the glass plate, the total number of load parameters that need to be identified is 3k, which are P=[x1 y1 w1 x2 y2 w2…x k y k w k ] T Where P is the glass plate load parameter vector to be identified, x k is the lateral position of the kth load, y k is the longitudinal position of the kth load, w k is the kth load mass, and the superscript T is the matrix transpose.
5. The method for identifying random loads of glass plates in glass plank road projects based on measuring point strain according to claim 4 is characterized by: The specific contents of establishing the objective function are as follows: In the identification of glass plate load parameters, the objective function can be defined according to the difference between the numerical calculation results of the strain at the measuring point and the on-site detection value, and standardized. At the same time, there are multiple objective functions, which need to be controlled. The numerical calculation results of the strain in the X and Y directions of the four measuring points and the on-site detection values are used to construct the standardized objective functions g1(P) and g2(P): Where g1(P) and g2(P) are the constructed objective functions. is the calculated value of the X-direction strain at the i-th measuring point (i=1 to 4) on the bottom surface of the tempered laminated glass plate. It is the on-site detection value of the X-direction strain at the i-th measuring point (i=1 to 4) on the bottom surface of the tempered laminated glass plate; the superscript c represents the calculated result of the physical quantity, and the superscript m represents the measured result of the physical quantity.
6. The method for identifying random loads on glass plates of glass plank road projects based on measuring point strain according to claim 5 is characterized by: There are multiple objective functions, which can be expressed as minimize G(P)={g1(P),g2(P)} T stΩ={P|Pl≤P≤Pu} In the formula, minimize represents mathematical optimization, st represents the constraints in mathematical optimization; P is the glass plate load parameter vector to be identified; P l , P u and Ω are the lower limit, upper limit and feasible domain of the glass plate load parameter vector respectively; G(P) is the objective function vector of the multi-objective optimization problem. In the multi-objective optimization problem, there is usually no feasible solution that can minimize all objective functions at the same time; Therefore, the multi-objective optimization problem is actually the problem of solving a set of non-inferior solutions.
7. The method for identifying random loads on glass plates of glass plank road projects based on measuring point strain according to claim 6 is characterized by: A genetic algorithm is used to solve the set of non-inferior solutions, starting from multiple feasible solutions. The set of multiple feasible solutions is called a population, denoted as pop(n), where n represents the current iteration step; each individual in the population represents a feasible solution P of the multi-objective optimization problem and is represented in the form of an encoded binary string. The entire genetic algorithm consists of three genetic operators: selection, crossover, and mutation.
8. The method for identifying random loads on glass plates of glass plank road projects based on measuring point strain according to claim 7 is characterized by: The specific steps of genetic algorithm to solve the non-inferior solution set are as follows: Step 1: Use a uniformly distributed random function to generate a random number containing N in the feasible domain Ω. s The initial population of individuals P is pop(0), the population size Ns is set to 50, and P is a feasible solution; Step 2: For the minimization problem in the identification of glass plate load parameters, calculate the fitness function values g1(P) and g2(P) of each individual P in the population pop(n); Step 3: When the convergence condition Function tolerance 10-4 is met, the calculation of the entire genetic algorithm is stopped; otherwise, a binary tournament that can ensure the genetic diversity of the population is used to select 2 individuals randomly selected from the current population pop(n), and the tournament size is 2; the individual with the smallest fitness g1(P) and g2(P) is saved in the next generation population pop(n+1); the above process is repeated until the number of individuals in the next generation population pop(n+1) reaches the preset number, forming a new population pop(n+1); Step 4: With ratio p c Perform an intermediate crossover between two randomly selected individuals to obtain a new population after crossover, crosspop(n+1); the crossover ratio p c Set to 1.0, so that the genes at all positions of any selected individual are crossovered, providing sufficient variability for the next population; Step 5: With mutation probability p m Randomly select some loci of individuals in the population crosspop(n+1), change the gene values of these loci, form mutant individuals, and obtain a new population mutpop(n+1) after mutation; return the new population mutpop(n+1) to Step 2 for calculation; the mutation function depends on the constraints, and the mutation probability p m Set to 1.
0.
9. The method for identifying random loads of glass plates in glass plank road projects based on measuring point strain according to claim 8 is characterized by: The genetic algorithm is used to solve the multi-objective optimization problem derived from the identification of glass plate load parameters, and the Pareto frontier Ω*={P1,P2,…,P l }, Ω* represents the set of non-inferior solutions to the optimization problem obtained; Finally, the random load parameter matrix on the glass plate can be obtained, that is, the load position and size P = [x1 y1 w1 x2 y2 w2…x k y k w k ] T .
