CN117744465A - Structural damage identification method, system, equipment and medium - Google Patents
Structural damage identification method, system, equipment and medium Download PDFInfo
- Publication number
- CN117744465A CN117744465A CN202311374315.8A CN202311374315A CN117744465A CN 117744465 A CN117744465 A CN 117744465A CN 202311374315 A CN202311374315 A CN 202311374315A CN 117744465 A CN117744465 A CN 117744465A
- Authority
- CN
- China
- Prior art keywords
- locust
- individual
- updating
- population
- structural damage
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000006378 damage Effects 0.000 title claims abstract description 146
- 238000000034 method Methods 0.000 title claims abstract description 84
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 73
- 238000005457 optimization Methods 0.000 claims abstract description 56
- 230000002441 reversible effect Effects 0.000 claims abstract description 56
- 241000238814 Orthoptera Species 0.000 claims abstract description 37
- 230000003247 decreasing effect Effects 0.000 claims abstract description 17
- 230000003044 adaptive effect Effects 0.000 claims abstract description 15
- 230000006870 function Effects 0.000 claims description 38
- 238000004364 calculation method Methods 0.000 claims description 25
- 230000008569 process Effects 0.000 claims description 17
- 238000003860 storage Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 12
- 230000008859 change Effects 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 11
- 208000027418 Wounds and injury Diseases 0.000 claims description 6
- 208000014674 injury Diseases 0.000 claims description 6
- 101000870363 Oryctolagus cuniculus Glutathione S-transferase Yc Proteins 0.000 claims description 3
- 101150036841 minJ gene Proteins 0.000 claims description 3
- 230000017105 transposition Effects 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 14
- 238000005259 measurement Methods 0.000 description 10
- 238000009826 distribution Methods 0.000 description 9
- 230000001133 acceleration Effects 0.000 description 7
- 238000012937 correction Methods 0.000 description 7
- 239000000463 material Substances 0.000 description 7
- 230000002829 reductive effect Effects 0.000 description 6
- 230000004044 response Effects 0.000 description 6
- 230000008901 benefit Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000005315 distribution function Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000011002 quantification Methods 0.000 description 3
- 241001206881 Myrmeleon inconspicuus Species 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 238000005452 bending Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012821 model calculation Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 241000238633 Odonata Species 0.000 description 1
- 229910000831 Steel Inorganic materials 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000005291 chaos (dynamical) Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000010921 in-depth analysis Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- MYVIATVLJGTBFV-UHFFFAOYSA-M thiamine(1+) chloride Chemical compound [Cl-].CC1=C(CCO)SC=[N+]1CC1=CN=C(C)N=C1N MYVIATVLJGTBFV-UHFFFAOYSA-M 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a method, a system, equipment and a medium for identifying structural damage, wherein the method comprises the following steps: s1, acquiring structural working condition data, and establishing a structural damage identification model; s2, optimizing the structural damage recognition model by using a locust optimization algorithm improved by a self-adaptive Levy flight strategy and an elite reverse learning strategy, and obtaining an optimal solution: s21, initializing a locust optimization algorithm; s22, calculating the fitness value of each locust individual, and determining an initial optimal solution; s23, updating a linear decreasing coefficient; s24, standardizing the distance between the locusts to be 1,4, and updating the positions of the locusts; s25, updating the position of the locust individual by using an adaptive Levy flight strategy; s26, updating the locust population by using an elite reverse learning strategy, calculating the fitness value of the locust individual, and determining the optimal solution of the current iteration; and S27, judging whether the current iteration times L is greater than or equal to the maximum iteration times L, if so, ending the iteration, outputting a global optimal solution, otherwise, enabling l=l+1, and returning to the step S23.
Description
Technical Field
The invention belongs to the technical field of structural monitoring, and particularly relates to a structural damage identification method, a system, equipment and a medium.
Background
At present, bridge construction in China also faces new challenges such as 'aging' and service condition deterioration of service bridges. Therefore, structural damage identification needs to be carried out on the operation bridge, and further safety performance of the operation bridge is controlled.
Structural damage identification is a typical inverse problem, and structural damage determination, positioning and quantification are achieved by performing in-depth analysis on response data acquired by a structural health monitoring system. In the actual research process, the structural damage identification can be divided into two major categories, namely model-based driving and data-based driving according to whether the finite element model is relied on or not. Data-driven methods are generally only capable of judging whether a structure is damaged, while model-driven methods can be used for positioning and even quantifying the damage, so that the method is more commonly used in structural damage identification research. Model-driven structural damage identification methods are typically implemented in a model-corrected manner. The basic idea of the method is to modify model parameters so that errors between the model output response or characteristics calculated through multiple iterations and the measured values of the actual structure are minimized, and therefore parameterized evaluation of the actual structure state is achieved. The model modification process involves a large number of iterative computations. The swarm intelligence algorithm is a coastal computing method for simulating the intelligence of the biological swarm, is good at excavating the optimal solution in a complex high-dimensional space, and has the characteristics of convenient and effective application. Therefore, the method is widely studied in the field of structural damage identification. At present, particle swarm algorithm, ant lion algorithm, butterfly algorithm and the like are successfully applied to the field of structural damage identification. However, most algorithms still have problems of insufficient recognition accuracy, unstable multiple recognition, and noise robustness to be improved.
Disclosure of Invention
In order to solve the problems in the background art, the invention aims to provide a method, a system, equipment and a medium for identifying structural damage.
The invention is realized by the following technical scheme:
a method of identifying structural damage comprising the steps of:
s1, acquiring structural working condition data, and establishing a structural damage identification model;
s2, optimizing the structural damage recognition model by using a locust optimization algorithm improved by a self-adaptive Levy flight strategy and an elite reverse learning strategy to obtain an optimal solution;
the method for optimizing the structural damage recognition model by using the locust optimization algorithm improved by the self-adaptive Levy flight strategy and the elite reverse learning strategy comprises the following steps of:
s21, initializing the number N of locust populations and the maximum iteration number L of a locust optimization algorithm, and setting the current iteration number L to be 1;
s22, calculating the fitness value of each locust individual, and determining the locust individual with the optimal fitness value as an initial optimal solution;
s23, updating a linear decreasing coefficient;
s24, standardizing the distance between the locusts to be 1,4, and updating the positions of the locusts;
s25, for the locusts after position updating, updating the positions of the locusts by using a self-adaptive Levy flight strategy, if the fitness value of the locusts after position updating is better than that of the locusts before position updating, reserving the locusts after position updating, otherwise, reserving the locusts before position updating;
S26, updating the locust population by using an elite reverse learning strategy, calculating the fitness value of each locust individual in the updated locust population, and determining the locust individual with the optimal fitness value as the optimal solution of the current iteration;
s27, judging whether the current iteration times L is greater than or equal to the maximum iteration times L, if so, ending the iteration, outputting a global optimal solution based on the initial optimal solution and the optimal solution of each iteration, otherwise, enabling l=l+1, and returning to the step S23.
Further, the step of updating the position of the locust individual using the adaptive Levy flight strategy comprises:
the self-adaptive nonlinear change factor beta of the curve is calculated, and the calculation formula is as follows:
wherein L is the current iteration number, L is the maximum iteration number, and a and b are formula adjustment coefficients;
updating the individual locusts based on the curve-adaptive nonlinear variation factor beta:
in the method, in the process of the invention,to update the position of the ith locust individual, X i The position of the ith locust individual before updating.
Further, the step of updating the locust population using elite reverse learning strategy comprises:
the locust individuals in the locust population are ranked according to the fitness value, the first preset number percent of the locust individuals before ranking are regarded as elite locust population, the first preset number percent of the locust individuals after ranking are regarded as obsolete locust population, and the rest middle locust individuals are regarded as common locust population;
For elite locust population, generating elite reverse population through reverse points, wherein the calculation formula of the reverse points is as follows:
wherein r is el Representing pseudo-random numbers subject to U (0, 1), X i,m ∈[Eu m ,El m ],Eu m =max(X m ),El m =min(X m ),X i,m For the position of the ith locust individual in the m-th dimension, X m Eu is the position vector of all locust individuals in the m-th dimension m And El m Representing the maximum value and the minimum value of the position of the locust individual in the m-th dimension;
and combining the elite locust population, the elite reverse population and the common locust population to obtain an updated locust population.
Further, the step of updating the position of the locust individual includes:
updating the position of the locust individual by the following formula:
in the method, in the process of the invention,mth dimension information, ub, representing the ith locust individual at the 1+1th iteration m And lb m Respectively represent the upper limit and the lower limit of the m-th dimension of the locust individual, and the +.>Representing the optimal position of the mth dimension in the first iteration, parameter c l For the linear decreasing coefficient of the first iteration s is a function and +.>Wherein f represents the attractive force intensity, R represents the attractive force range, c max And c min The maximum value and the minimum value of the linear decreasing coefficient are respectively, L is the current iteration number, and L is the maximum iteration number.
Further, the step of establishing the structural damage identification model includes:
The function of the structural damage recognition model is:
α opt =arg α minJ(α) (6);
wherein alpha is opt For the optimal damage factor vector, J (alpha) is the target functionA number.
Further, the function of the structural damage recognition model is:
wherein w is 1 、w 2 Is two weight factors, the sum of the two weight factors is 1, N m For the order of the mode shape,i=1, 2, …, N for the i-th order modality m ,K j A unit rigidity contribution matrix of the j-th unit, T is a vector transposition, f i Frequency f of the ith target variable i And (alpha) is the calculated frequency of the ith target variable.
Further, in the step of establishing the structural damage recognition model, the method further comprises the step of adding a regularization penalty term to the objective function to carry out constraint, wherein the calculation formula of the L1 norm is as follows:
L 1 =‖α‖ 1 /N m (11);
in the formula II alpha II 1 Is the 1 norm of the injury factor vector alpha, |alpha|| 1 =|α 1 |+|α 2 |+…+|α Nele |。
The invention also provides a structural damage identification system, which comprises:
the acquisition module is used for acquiring the structural working condition data and establishing a structural damage identification model;
the optimization module is used for optimizing the structural damage recognition model by using a locust optimization algorithm improved by a self-adaptive Levy flight strategy and an elite reverse learning strategy and obtaining an optimal solution;
wherein, the optimization module includes:
the initializing sub-module is used for initializing the number N of locust populations and the maximum iteration number L of the locust optimization algorithm, and setting the current iteration number L to be 1;
The calculation sub-module is used for calculating the fitness value of each locust individual and determining the locust individual with the optimal fitness value as an initial optimal solution;
a first updating sub-module for updating the linear decreasing coefficient;
the standardized submodule is used for standardizing the distance between the locusts to be 1,4 and updating the positions of the locusts;
the Levy flight module is used for updating the position of the locust individual by using a self-adaptive Levy flight strategy for the locust individual after the position update, if the fitness value of the locust individual after the position update is better than that of the locust individual before the position update, the locust individual after the position update is reserved, and otherwise, the locust individual before the position update is reserved;
the elite reverse learning module is used for updating the locust population by using an elite reverse learning strategy, calculating the fitness value of each locust individual in the updated locust population, and determining the locust individual with the optimal fitness value as the optimal solution of the current iteration;
the judging module is used for judging whether the current iteration times L is greater than or equal to the maximum iteration times L, if yes, ending the iteration, outputting a global optimal solution based on the initial optimal solution and the optimal solution of each iteration, otherwise, enabling l=l+1, and returning to the first updating sub-module.
The invention also provides an electronic device, which comprises:
a processor;
a memory for storing an executable computer program;
wherein the processor, when executing the computer program, implements the steps of the method for identifying structural damage.
The invention also provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of a method of identifying structural damage.
Compared with the prior art, the invention has the beneficial effects that: by fusing the self-adaptive Levy flight strategy and the elite reverse learning strategy in the locust optimization algorithm, the self-adaptive Levy flight strategy enables the locust population search range to be wider, the global search capability of the standard locust optimization algorithm is enhanced, the elite reverse learning strategy utilizes the advantage that elite individuals carry more effective information to form dynamic boundaries and construct reverse populations, and therefore accuracy and stability of structural damage identification are improved, and accuracy and robustness of structural damage identification of the algorithm under incomplete measurement and noise interference are improved.
Drawings
FIG. 1 is a flow chart illustrating steps of a method for identifying structural damage according to the present invention;
FIG. 2 is a flowchart showing the steps of optimizing and solving a structural damage recognition model by using an improved locust optimization algorithm in the structural damage recognition method of the present invention;
FIG. 3 is a diagram of the influence of the beta value on the Levy flight step length in the Levy flight strategy, FIG. 3 (a) is a diagram of the lower limit of the beta value, and FIG. 3 (b) is a diagram of the upper limit of the beta value;
FIG. 4 is a graph of the dynamic change of beta values in a Levy flight strategy;
FIG. 5 is a box plot of the results of 10 independent runs;
FIG. 6 is a schematic diagram of a vertical simulation model of a simply supported beam;
FIG. 7 is a schematic diagram of the structure damage recognition results for different conditions;
FIG. 8 is a schematic diagram of the result damage recognition result after L1 regularization is introduced;
FIG. 9 is a graph showing the result of IGOA-ALOL-based structural damage identification at a noise level of 1.5%;
FIG. 10 is a schematic diagram of a structural damage recognition structure of an experimental structure;
FIG. 11 is a diagram illustrating a method of identifying structural damage according to the present invention;
fig. 12 is a hardware configuration diagram of the electronic device of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further definition or explanation thereof is necessary in the following figures. Meanwhile, in the description of the present invention, the terms "first", "second", and the like are used only to distinguish the description, and are not to be construed as indicating or implying relative importance.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
In the description of the present invention, it should be noted that, directions or positional relationships indicated by terms such as "upper", "lower", "inner", "outer", etc., are directions or positional relationships based on those shown in the drawings, or those that are conventionally put in use, are merely for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the apparatus or elements to be referred to must have a specific direction, be constructed and operated in a specific direction, and thus should not be construed as limiting the present invention.
Referring to fig. 1 and 2, fig. 1 is a flowchart illustrating steps of a structural damage recognition method according to the present invention, and fig. 2 is a flowchart illustrating steps of optimizing and solving a structural damage recognition model by using an improved locust optimization algorithm in the structural damage recognition method according to the present invention. A method of identifying structural damage comprising the steps of:
s1, acquiring structural working condition data, and establishing a structural damage identification model;
s2, optimizing the structural damage recognition model by using a locust optimization algorithm improved by a self-adaptive Levy flight strategy and an elite reverse learning strategy to obtain an optimal solution;
the method for optimizing the structural damage recognition model by using the locust optimization algorithm improved by the self-adaptive Levy flight strategy and the elite reverse learning strategy comprises the following steps of:
S21, initializing the number N of locust populations and the maximum iteration number L of a locust optimization algorithm, and setting the current iteration number L to be 1;
s22, calculating the fitness value of each locust individual, and determining the locust individual with the optimal fitness value as an initial optimal solution;
s23, updating a linear decreasing coefficient;
s24, standardizing the distance between the locusts to be 1,4, and updating the positions of the locusts;
s25, for the locusts after position updating, updating the positions of the locusts by using a self-adaptive Levy flight strategy, if the fitness value of the locusts after position updating is better than that of the locusts before position updating, reserving the locusts after position updating, otherwise, reserving the locusts before position updating;
s26, updating the locust population by using an elite reverse learning strategy, calculating the fitness value of each locust individual in the updated locust population, and determining the locust individual with the optimal fitness value as the optimal solution of the current iteration;
s27, judging whether the current iteration times L is greater than or equal to the maximum iteration times L, if so, ending the iteration, outputting a global optimal solution based on the initial optimal solution and the optimal solution of each iteration, otherwise, enabling l=l+1, and returning to the step S23.
In the step S1, a plurality of acceleration sensors are installed at different positions on the structure to be monitored, then acceleration responses of the different positions of the structure to be monitored are obtained through the plurality of acceleration sensors, then material properties and structure dimensions of the structure to be monitored are obtained, the obtained acceleration responses, the material properties and the structure dimensions of the different positions of the structure to be monitored form structure working condition data, frequencies are obtained through fourier transformation based on the acceleration responses, vibration modes are obtained through a polymax method, and therefore the frequencies and the vibration modes of the structure can be obtained, then an objective function for identifying structural damage is built by combining the material properties and the structure dimensions, a structural damage identification model is built, specifically, the frequencies are used as input, actual measurement frequencies and calculation frequencies are fitted through an algorithm, the material properties and the structure dimensions are used as output after the materials and the structure dimensions are matched, and the structural damage identification model is built.
Further, the step of establishing the structural damage identification model includes:
s11, a function of the structural damage identification model is as follows:
α opt =arg α minJ(α) (6);
wherein alpha is opt For the optimal injury factor vector, J (α) is the objective function.
In the above step S11, mathematically, the structural damage recognition problem may be converted into an optimization problem as shown in the formula (6) to solve, and iterate continuously through the model correction technique to narrow the model Error between the calculated result and the actually measured result to finally obtain the optimal damage factor vector alpha of the reaction structure state opt . In the problem of identifying structural damage, the objective function is usually calculated by the dynamic characteristics of the structure, such as the calculation frequency and the vibration mode, and is required to be calculated according to the rigidity matrix of the structure, and based on the calculation, a unit rigidity reduction model is usually adoptedPerforming damage description, wherein K d As a matrix of overall stiffness, α i For the damage factor of the ith unit, the damage factors of each unit are combined to form a damage factor vector alpha= { alpha 1 ,α 2 ,…,α i },N ele For the number of units, K ui Is a matrix of cell stiffness; the model ignores the influence of damage on mass distribution, reduces the rigidity matrix according to the damage setting condition, and leads the whole rigidity matrix K after damage d Represented as a cell stiffness matrix K ui The linear superposition of the two-dimensional structural damage has the advantages of simple expression and clear physical meaning, thereby obtaining the dynamic characteristics after the structural damage.
Further, the modal strain energy index is constructed by the stiffness contribution matrix and the modal shape vector of the structure, and not only the dynamic characteristics of the structure but also the physical characteristics of the structure are considered. Compared with the traditional index which only considers frequency or vibration mode, the method has higher precision and stronger damage quantification capability when identifying damage. The modal strain energy of a structure is expressed as:
Where MSE ij The ith order modal strain energy for the jth cell,for the ith order mode, K j A matrix is contributed to the cell stiffness of the j-th cell.
Thus, the modal strain energy vector of each unit of the structure can be obtained after calculation:
wherein T is the vector transposition, N m Is the modal order.
In order to measure the difference between the target structure and the calculated structure. Based on the target MSE j And calculate MSE j (a) The cosine distance between them suggests the modal strain energy assurance standard Measurement (MSEAC), defined as:
if the target MSE j With MSE j (a) Identical, then MSEAC j (a) Is equal to 1 and the damage factor vector a reflects well the condition of the target structure. In addition, the influence of frequency is also considered by introducing a Frequency Change Rate (FCR), and the calculation formula is:
wherein f i Is the firstiThe frequency of the individual target variables, f i And (alpha) is the calculated frequency of the ith target variable.
The modal strain energy assurance criteria are linearly combined with the rate of change of frequency to form a new objective function as follows:
wherein alpha is opt Is the optimal result, also called SDD (Structural damage detection, structural damage identification) result, w 1 And w 2 Is two weight factors, the sum of the two weight factors is 1,i=1, 2, …, N for the i-th order modality m ,K j A matrix is contributed to the cell stiffness of the j-th cell.
In one embodiment, w 1 And w 2 Values of 0.1 and 0.9 were empirically chosen.
Further, when the structure is damaged, only few local areas are obviously damaged, and most areas still remain intact, so that the structure has the characteristic of sparse distribution. The L0 norm represents the number of non-zero elements in the vector, so the sparseness of the vector can be adjusted by minimizing the L0 norm, thereby obtaining a sparse solution. However, the L0 norm optimization problem is an NP-hard problem that is difficult to solve. Therefore, in practical applications, the L1 norm is generally used instead of the L0 norm. The L1 norm is the optimal convex relaxation approximation of the L0 norm, which is the norm of the vector space in mathematics, representing the sum of the absolute values of all elements in the vector.
Therefore, the step of building the structural damage recognition model further comprises the step of adding a regularization penalty term to the objective function to carry out constraint, wherein the regularization penalty term is L1 norm, and the calculation formula of the L1 norm is as follows:
L 1 =‖α‖ 1 /N m (11);
in the formula II alpha II 1 Is the 1 norm of the injury factor vector alpha, |alpha|| 1 =|α 1 |+|α 2 |+…+|α Nele |。
The objective function is expressed as follows after adding the regularization penalty term:
J(α)=J(α)+λL 1 (12);
i.e. the
In the formula, λ is a regularization coefficient.
In the step S2, the locust optimization algorithm has the advantages of simple structure, stable effect and strong local search performance, but has the problems of weak global optimization capability, single individual behavior mode and the like. Therefore, the locust optimization algorithm is improved in the aspects of individual behavior, global optimization, convergence accuracy and the like of the locust, the self-adaptive Levy flight strategy and elite reverse strategy are combined, an improved locust optimization algorithm is provided, and the structural damage recognition model is optimized and solved through the improved locust optimization algorithm, so that the accuracy and stability of structural damage recognition are improved.
In the step S21, parameters of the optimization algorithm of the locust are initialized, including the number N of locust populations, the maximum iteration number L, and the initial position X of the individual locust i I=1, 2, …, N, current iteration number i. And the initial value of the current iteration number l is set to 1. Each individual locust carries a vector of injury factors.
In the step S22, the fitness value of each locust individual is calculated, the locust individuals are ranked according to the fitness value, the position of the locust individual T with the optimal fitness value (i.e., the fitness value is the smallest) is recorded, and the locust individual T with the optimal fitness value is determined as the initial optimal solution.
In the step S23, the improved locust optimization algorithm is iterated, and when the current iteration number L is smaller than the maximum iteration number L, the following formula is adopted to adaptively update the linear decreasing coefficient c:in c max And c min The maximum value and the minimum value of the linear decreasing coefficient are respectively, L is the current iteration number, and L is the maximum iteration number.
In the above step S24, for each locust individual, by a function of the distance correlation between the locust individuals:distance between standardized locusts is 1,4]Thereby normalizing the distance between locust individuals to the range [1,4 ] ]. Wherein f represents the attractive force intensity, the value is 0.5, R represents the attractive force range, l represents 1.5, exp represents an exponential function based on e. And updating the position of the locust individual by the following formula:
in the method, in the process of the invention,mth dimension information, ub, representing the ith locust individual at the 1+1th iteration m And lb m Respectively represent the lb of the locust individual m Upper and lower dimension limit, +.>Representing the optimal position of the mth dimension in the first iteration, parameter c l For the linear decreasing coefficient of the first iteration s is a function and +.>Wherein f represents the attractive force intensity, R represents the attractive force range, c max And c min The maximum value and the minimum value of the linear decreasing coefficient are respectively, L is the current iteration number, and L is the maximum iteration number.
In the step S25, the Levy flight is a non-gaussian random process related to the chaos theory, and the step size of the Levy flight obeys the Levy probability distribution function, and the distribution function can be expressed by a power law formula: l (gamma): |gamma| -1-β . Similarly, the Levy distribution function has the characteristic of a power law distribution function, the parameter beta plays a main control role in determining the Levy distribution, the probability distribution of Levy flying is controlled, and different values of beta can obtain probability distributions with different shapes. Smaller values of beta will result in a more uniform probability distribution of the Levy distribution with a greater probability of large step sizes, a typical step size distribution pattern being shown in fig. 3 (a). While a larger value of beta gives a larger probability of a small step, as shown in fig. 3 (b). By utilizing the characteristics, the Levy flight strategy is introduced in the early iteration stage of the locust optimization algorithm, so that the global optimizing capability of the locust optimization algorithm can be improved. In the local search phase, the algorithm needs to search around the solution space more carefully, so a larger β value as shown in fig. 3 (b) can be selected to meet the search requirement of the later iteration phase. In fig. 3, fig. 3 (a) is a schematic diagram of the lower limit of the β value, and fig. 3 (b) is a schematic diagram of the upper limit of the β value.
Thus, based on the above considerations, further, in step S25, the step of updating the position of the individual locust using the adaptive Levy flight strategy comprises:
s251, calculating a curve self-adaptive nonlinear change factor beta, wherein the calculation formula is as follows:
wherein L is the current iteration number, L is the maximum iteration number, and a and b are formula adjustment coefficients; beta max And beta min Respectively an upper limit and a lower limit of the beta value;
s252, updating the individual locusts:
in the method, in the process of the invention,to update the position of the ith locust individual, X i The position of the ith locust individual before updating.
In the above step S251 and step S252, the values of a and b may be 14.00 and 2.25, respectively. After the curve self-adaptive nonlinear change factor beta is introduced, as shown in fig. 4, in the iterative process of the locust optimization algorithm, the beta value is self-adaptively adjusted relative to the iterative times, so that the flexible change of the beta parameter from 1.5 to 2 nonlinear changes is realized in different stages. After the introduction of the adaptive Levy flight strategy, for each locust individual, the position of the offspring individual was calculated according to equation (2).
In the step S25, the position of the locust individual obtained by introducing the update of the adaptive Levy flight strategy is not necessarily better than the original position because of the randomness of the adaptive Levy flight strategy. Therefore, the locust individual obtained after the adaptive Levy flight strategy is executed does not directly enter iteration, but participates in survival competition together with the locust individual before the adaptive Levy flight strategy is executed, a greedy algorithm is used for comparing the fitness values of the locust individual before and after the adaptive Levy flight strategy, and only the optimal locust individual can enter the next iteration, so that if the fitness value of the locust individual after the adaptive Levy flight strategy is superior (the fitness is smaller than) the fitness value of the locust individual before the adaptive Levy flight strategy is executed, the locust individual after the adaptive Levy flight strategy is reserved, otherwise, the locust individual before the adaptive Levy flight strategy is reserved.
In the step S26, the elite inverse learning strategy utilizes the characteristic that elite individuals contain more effective information, and forms a dynamic boundary according to the maximum value and the minimum value of elite locust groups, so that the elite inverse solution can be further converged into a narrower dynamic search space when the algorithm converges, thereby better retaining the search experience. The population diversity can be effectively improved by introducing a reverse learning strategy, and the risk of the algorithm falling into local optimum is reduced.
Further, in step S26, the step of updating the locust population using the elite reverse learning strategy includes:
s261, sorting locust individuals in the locust population from small to large according to fitness values, regarding the first preset number percent of locust individuals before ranking as elite locust population, regarding the first preset number percent of locust individuals after ranking as obsolete locust population, and regarding the rest middle locust individuals as common locust population;
s262, generating elite reverse population through reverse points for elite locust population, wherein the calculation formula of the reverse points is as follows:
wherein r is el Representing pseudo-random numbers subject to U (0, 1), X i,m ∈[Eu m ,El m ],Eu m =max(X m ),El m =min(X m ),X i,m For the position of the ith locust individual in the m-th dimension, X m Eu is the position vector of all locust individuals in the m-th dimension m And El m Representing the position of the locust individual in the m-th dimensionSetting a maximum value and a minimum value;
and S263, combining the elite locust population, the elite reverse population and the ordinary locust population to obtain updated locust population.
In the above steps S261 to S263, when the elite reverse learning strategy is performed, the elite locust population consisting of the first 10% of the locust individuals, the ordinary locust population consisting of the middle 80% of the locust individuals, and the obsolete population requiring the elite reverse learning strategy to be performed by the last 10% of the locust individuals are ordered according to the fitness value from small to large. And then finding out the position boundary (namely the maximum position and the minimum position) of the locust individual as a range for determining the reversal point, and calculating the reversal point. Based on the first 10% elite locust population, 10% elite reverse population is generated through the reverse points, and the elite locust population, the elite reverse population and the ordinary locust population are combined to form a new population, so that the updated locust population is obtained. And then calculating the fitness value of each locust individual in the updated locust population, sorting the locust individuals according to the fitness value, recording the position of the locust individual with the optimal fitness value (namely, the fitness value with the smallest fitness value), and determining the locust individual with the optimal fitness value as the optimal solution of the current iteration.
In the above step S27, it is determined whether the improved locust optimization algorithm satisfies the iteration termination condition: judging whether the current iteration number L is greater than or equal to the maximum iteration number L, if L is greater than or equal to L, ending the iteration, comparing the obtained optimal solution based on the initial optimal solution obtained for the first time and the optimal solution of the current iteration obtained by each iteration, recording the position of the locust individual with the optimal fitness value (namely, the fitness value is minimum), outputting the locust individual with the optimal fitness value as a global optimal solution, and directly reflecting the damage degree of the structure by the global optimal solution, namely, an optimal damage factor vector; otherwise, returning to the step S23 for the next iteration until the iteration termination condition is met.
Four CEC reference test functions are selected below to evaluate the performances of six optimization algorithms, such as a structural damage recognition method (IGOA-ALOL), a standard locust optimization algorithm (GOA), an ant lion optimization Algorithm (ALO), a dragonfly optimization algorithm (DA), a particle swarm optimization algorithm (PSO) and a moth fire suppression optimization algorithm (MFO), on different optimization problems. By the method of the benchmark function test, the performance of GOA before and after improvement in optimizing different problems can be evaluated. Typically, to increase the diversity of populations in improved algorithms, many algorithms introduce pseudo-random numbers that are subject to a uniform distribution of [0,1 ]. However, this approach may also lead to improved algorithms incorporating such strategies that produce a weak tendency toward the origin when optimizing test functions with a function value of 0, resulting in exceptionally good iterative results. Thus, in benchmark, the method of translating a function is adopted herein to avoid the positive impact of the function origin on the function performance. In the following, for the convenience of discussion and study, the influence of translation is removed, and the function is restored to the original state.
The benchmark function in table 1 is used as a performance benchmark for different improvement strategies.
TABLE 1
Wherein f 1 The optimization capacity and convergence accuracy of an optimization algorithm can be tested as a unimodal function; f (f) 2 、f 3 And f 4 The ability of the optimization algorithm to search globally and jump out of local optima can be tested for multimodal functions. The minimum of the four functions is 0. The population number is set to 30, the dimension is set to 30, the maximum iteration number is set to 100, and other parameters of the algorithm are set according to the original paper of the locust optimization algorithm. The box plot shown in fig. 5 is plotted from the 10 calculations run independently in succession. The box diagram is a simple way of summarizing data, the maximum value, the minimum value, the median and the upper and lower quartiles of the sequenced data are drawn into lines, two quartiles are connected to draw a box body, and then the upper edge and the lower edge are connected with the box body. This way, critical information about the location and dispersion of the data can be presented intuitively. The box diagram can intuitively understand that 6 algorithms run for a plurality of timesConvergence accuracy and stability of (c). As can be seen from fig. 5, in the optimization function f 1 、f 2 And f 3 In this case, the median, upper and lower edges, and upper and lower quartiles of the IGOA-ALOL are almost coincident, which indicates that the IGOA-ALOL has better stability than the other 5 algorithms. At f 1 And f 2 In the above, the convergence accuracy of IGOA-ALOL and GOA is higher, and the running result of GOA (10 running averages: 8.29E+02, 7.97E+00) is inferior to IGOA-ALOL (10 running averages: 2.03E-02, 3.27E-02). This shows that standard GOA also has a certain potential in function optimization, and the performance of GOA can be further improved by improving GOA. Further, at f 3 And f 4 In the comparison of the results of (a), IGOA-ALOL is significantly superior to other algorithms in terms of both stability and convergence accuracy.
The finite element model shown in fig. 6 is used for bridge structure damage identification study. . The model is a simply supported beam, the length is 25.0m, the equal length is divided into 20 units, and the unit type is a two-node four-degree-of-freedom plane bending beam unit. The elastic modulus E=210 GPa, the material density rho=7850 kg/m of the model 3 The cross section is a rectangular cross section with an area a=0.48 m 2 Moment of inertia i=0.0256 m 4 。
Basic parameters for identifying structural damage by different algorithms are kept consistent, the number of population individuals is 100, and the maximum iteration number is 100. Taking N m =5, wherein the mode shape extracts only the vertical degrees of freedom of each unit node. The definition of the objective function is as follows:
to more specifically compare structural damage recognition results, damage recognition result accuracy is quantified using a damage vector consistency (Damage Vector Consistency, DVC) index, DVC100 representing that the recognition result is exactly consistent with the hypothetical damage. In order to avoid the problem that the solution precision is insufficient due to the fact that the swarm intelligence algorithm possibly falls into local optimum in single calculation, the method is used for obtaining the structural damage identification result by independently running 100 times in a mode of repeatedly calculating the average value for many times, and the DVC100 index is used for quantitatively counting the identification precision.
Example 1
The combination of GOA with Levy flight strategy and elite reverse learning strategy can result in 3 different versions of improved GOA. Wherein IGOA-AL represents GOA combined with the adaptive Levy flight strategy only, IGOA-OL represents GOA combined with the elite reverse learning strategy only, and IGOA-ALOL represents GOA combined with the Levy flight strategy and the elite reverse learning strategy. Meanwhile, for convenience of comparison, the coefficient λ was set to 0 without considering the introduction of L1 regularization in example 1, and the damage conditions were as shown in table 2, considering single damage, two damage, and multiple damage conditions, respectively.
Table 2 damage condition settings
As can be seen from fig. 7, the standard GOA algorithm does not successfully identify structural damage. When the single damage working condition is identified, misjudgment of different degrees occurs in the IGOA-AL and IGOA-OL in the units close to the beam support and the adjacent units at the damage position, and the IGOA-ALOL has almost no misjudgment unit, so that the identification of the damage degree is the most accurate. Misjudgment of different degrees occurs in three versions of IGOA when the damage degree of the multi-damage working condition is identified, wherein the misjudgment of IGOA-ALOL is slight, and the structural damage condition can be reflected accurately.
It can be further found from table 3 that the success rate of recognizing the damage by the GOA-aL, the IGOa-OL and the IGOA-ALOL of the GOA algorithm fused with the improved strategies is greatly improved compared with that of the standard GOA algorithm, which indicates that the success rate of recognizing the damage by the improved algorithm can be better improved by properly combining the two improved strategies.
TABLE 3 DVC100 values for SDD Condition
In summary, under the working conditions set in embodiment 1, the improved algorithm can successfully identify structural damage, while IGOA-ALOL has the highest identification accuracy, can provide the most accurate identification result, and has the least erroneous judgment unit.
Example 2
On the basis of embodiment 1, L1 regularization was introduced, and the regularization coefficient λ was empirically set to 0.03.
As can be seen from fig. 8, the standard GOA algorithm still fails to successfully identify structural damage after L1 regularization is introduced. The misjudgment conditions of the units of the three IGOAs close to the beam support and the adjacent units of the damaged position are reduced, wherein the misjudgment conditions of the IGOA-ALOL are still the slightest, and the damage condition of the structure can be reflected most accurately.
TABLE 4 DVC100 values for SDD Condition
As can be seen from table 4, the success rate of identifying damage by three types of IGOA is further improved after L1 norm regularization is introduced, and the improvement effect by IGOA-aLOL is particularly obvious. This shows that the method provided herein can effectively improve the structural damage recognition accuracy using only IGOA-ALOL, and greatly reduce unit misjudgment.
Example 3
On the basis of example 2, noise is added to both frequency and mode shape, and the noise addition formula is as follows:
r n =r cal (1+E p N oise ) (14);
wherein r is n 、r cal Respectively represent noisy and noiseless data, E p Represents the noise level, N oise Is a random number subject to a gaussian distribution of N (0, 1). Conditions 2, 4, 5 were selected and gaussian white noise was added at a level of 1.5%.
As can be seen from fig. 9, the IGOA-ALOL still better locates and quantifies the major lesions in the structure in noisy environments. After L1 regularization is introduced, the accuracy and the robustness of the algorithm are improved to a certain extent, the unit identification accuracy is improved to some extent, and meanwhile, the misjudgment condition at the non-damaged position is reduced to some extent.
The following experiment verifies the structural damage identification method of the invention:
and building a single-span square tube section simply supported beam model in a laboratory to carry out experimental verification. The experimental model has a length of 3m, a cross section dimension of 0.06m×0.014m, and a square tube cross section thickness of 0.003m. In the initial case, the elastic modulus of the experimental beam is 210GPa, and the density is 7800kg/m 3 。
During the experiment, the acceleration response of the structure was obtained by arranging 21 acceleration sensors (PCB: ICP333B 30) on the structure, and arranging a vibration exciter (HEV-200) at a distance of 1.65m from the left end mount. . The first third-order frequency and the mode shape of the structure are extracted through an algorithm built in an LMS SADAS mode data acquisition system. Structural damage was simulated by cutting cracks along the height of the steel beam.
In the experimental process, due to the influences of factors such as deviation between actual material parameters and ideal conditions, deviation between actual geometric dimensions and ideal conditions, non-ideal hinging of a structural support and the like, errors necessarily exist between experimental modal data measured under the nondestructive condition and an initial finite element model calculation result. Therefore, before structural damage identification is performed, model correction must be performed on the initial finite element structure according to experimental modal measurement under non-destructive conditions, so as to obtain a reference model (Benchmark) that can effectively serve subsequent structural damage identification.
By a modal test on the lossless model, the front third-order mode shape and the front third-order frequency of the structure can be obtained (as shown in table 5). The observation table shows that the ratio of the measured frequency to the fundamental frequency is slightly lower than the theoretical order square, and the larger the order rise deviation is. The error calculation in table 5 uses the method of "(calculated value-measured value)/measured value×100%". The value calculation result reflects that the error between the calculation frequency and the actual measurement frequency of the initial finite element model is large, and the initial finite element model is required to be corrected. The four parameters are considered to be selected for initial finite element model correction, wherein the initial finite element model correction comprises linear density rho A, bending stiffness EI and vertical spring stiffness k at a support v Torsional spring stiffness k r . The values of the four parameters before and after correction are shown in the table6, the table shows that the change rate is within +/-13%, and no large change occurs, which ensures the actual physical meaning of the correction parameters. In addition, the calculated frequency (shown in table 5) and the measured frequency of the modified finite element model are basically consistent, which indicates that the selected modification parameters are effective and the modification result is reasonable.
Table 5 actual measurement frequency and model calculation frequency
Table 6 comparison of parameters before and after correction of simply supported beam model
The laboratory cut the beam across the width to simulate damage. The experimental procedure simulates a total of 5 damage conditions, the detailed information being shown in table 7. The table also shows the structure front third-order measurement frequency under different damage working conditions, and the front third-order frequency can be observed to be reduced to different degrees along with the aggravation of unit damage, wherein the third-order frequency is reduced most obviously. This is consistent with theory, indicating that the frequency measurements of experimental damage conditions are reasonable and efficient. The algorithm and regularization coefficient λ settings are the same as above.
TABLE 7 Experimental damage condition and corresponding measurement frequency
As can be seen from the recognition result shown in fig. 10: under the single damage working condition, the IGOA-ALOL can accurately identify the position and the degree of the structural damage. However, since the L1 regularization is not introduced, a few erroneous judgment with larger amplitude occur on the recognition result. Under the two damage working conditions, the L1 regularization is introduced, so that not only can the success rate of IGOA-ALOL damage identification be improved, but also the occurrence of erroneous judgment can be reduced. This demonstrates that the proposed IGOA-ALOL can also effectively achieve localization and quantification of structural damage under experimental conditions after L1 regularization is introduced.
Referring to fig. 11, fig. 11 is a schematic block diagram of a structural damage identifying system according to the present invention. Corresponding to the embodiment of the method for identifying structural damage of the present invention, the present invention also provides a system for identifying structural damage, comprising:
the acquisition module 1 is used for acquiring structural working condition data and establishing a structural damage identification model;
the optimization module 2 is used for optimizing the structural damage recognition model by using a locust optimization algorithm improved by a Levy flight strategy and an elite reverse learning strategy and obtaining an optimal solution;
wherein the optimization module 2 comprises:
the initializing sub-module is used for initializing the number N of locust populations and the maximum iteration number L of the locust optimization algorithm, and setting the current iteration number L to be 1;
the calculating sub-module is used for calculating the fitness value of each locust individual and determining the locust individual with the optimal fitness value as an optimal solution;
a first updating sub-module for updating the linear decreasing coefficient;
the standardized submodule is used for standardizing the distance between the locusts to be 1,4 and updating the positions of the locusts;
the Levy flight sub-module is used for updating the position of the locust individual by using a Levy flight strategy for the locust individual after the position update, if the fitness value of the locust individual after the position update is better than that of the locust individual before the position update, the locust individual after the position update is reserved, and otherwise, the locust individual before the position update is reserved;
The elite reverse learning sub-module is used for updating the locust population by using an elite reverse learning strategy and calculating the fitness value of each locust individual in the updated locust population;
the second updating sub-module is used for updating the optimal solution based on the fitness value of each locust individual in the updated locust population and the fitness value corresponding to the optimal solution;
and the judging sub-module is used for judging whether the current iteration number L is greater than or equal to the maximum iteration number L, if so, ending the iteration, otherwise, making l=l+1, and returning to the first updating sub-module.
Further, the Levy flight submodule includes:
the first calculation unit is used for calculating a curve self-adaptive nonlinear variation factor beta, and the calculation formula is as follows:
wherein L is the current iteration number, L is the maximum iteration number, and a and b are formula adjustment coefficients;
a first updating unit for updating the individual locusts position:
in the method, in the process of the invention,to update the position of the ith locust individual, X i The position of the ith locust individual before updating.
Further, the elite reverse learning submodule includes:
the sorting unit is used for sorting the locust individuals in the locust population according to the fitness value, taking the locust individuals with the first preset number percentage before ranking as elite locust population, taking the locust individuals with the first preset number percentage after ranking as obsolete locust population, and taking the remaining middle locust individuals as common locust population;
The generation unit is used for generating elite reverse population through reverse points for elite locust population, wherein the calculation formula of the reverse points is as follows:
wherein r is el Representing pseudo-random numbers subject to U (0, 1), X i,m ∈[Eu m ,El m ],Eu m =max(X m ),El m =min(X m ),X i,m For the position of the ith locust individual in the m-th dimension, X m Eu is the position vector of all locust individuals in the m-th dimension m And El m Representing the maximum value and the minimum value of the position of the locust individual in the m-th dimension;
and the merging unit is used for merging the elite locust population, the elite reverse population and the common locust population to obtain updated locust population.
Further, the normalization submodule includes:
a second updating unit for updating the position of the individual locust by the following formula:
in the method, in the process of the invention,mth dimension information, ub, representing the ith locust individual at the 1+1th iteration m And lb m Respectively represent the upper limit and the lower limit of the m-th dimension of the locust individual, and the +.>Representing the optimal position of the mth dimension in the first iteration, parameter c l For the linear decreasing coefficient of the first iteration s is a function and +.>Wherein f represents the attractive force intensity, R represents the attractive force range, c max And c min The maximum value and the minimum value of the linear decreasing coefficient are respectively, L is the current iteration number, and L is the maximum iteration number.
The implementation process of the functions and roles of each module, sub-module and unit in the above system is specifically detailed in the implementation process of the corresponding steps in the above method, and will not be described herein again.
For system embodiments, reference is made to the description of method embodiments for the relevant points, since they essentially correspond to the method embodiments. The system embodiments described above are merely illustrative, wherein elements illustrated as separate elements may or may not be physically separate, and elements shown as elements may or may not be physical elements.
Corresponding to the foregoing embodiment of the structural damage identification method, the present invention further provides an electronic device, where the electronic device may include: a processor; a memory for storing an executable computer program; the processor executes the computer program to implement the method for identifying structural damage in any method embodiment.
The embodiment of the method and the system for identifying the structural damage can be applied to electronic equipment. Taking software implementation as an example, the device in a logic sense is formed by reading corresponding computer program instructions in a nonvolatile memory into a memory by a processor of an electronic device where the device is located for operation. In terms of hardware, as shown in fig. 12, the electronic device may include other hardware, such as a camera module, in addition to the processor, the memory, the network interface, and the nonvolatile memory shown in fig. 12; or may include other hardware, generally according to the actual function of the electronic device, which will not be described in detail.
Corresponding to the foregoing embodiments of the structural damage identification method, embodiments of the present invention further provide a computer readable storage medium having a computer program stored thereon, which when executed by a processor implements the structural damage identification method in any of the foregoing method embodiments.
Embodiments of the invention may take the form of a computer program product embodied on one or more storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing program code. The computer readable storage medium may include: removable or non-removable media, either permanent or non-permanent. The information storage function of the computer readable storage medium may be implemented by any method or technique that may be implemented. The information may be computer readable instructions, data structures, models of a program, or other data.
In addition, computer-readable storage media include, but are not limited to: phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology memory, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices, or other non-transmission media that may be used to store information that may be accessed by a computing device.
Compared with the prior art, the invention has the beneficial effects that: by fusing the self-adaptive Levy flight strategy and the elite reverse learning strategy in the locust optimization algorithm, the self-adaptive Levy flight strategy enables the locust population search range to be wider, the global search capability of the standard locust optimization algorithm is enhanced, the elite reverse learning strategy utilizes the advantage that elite individuals carry more effective information to form dynamic boundaries and construct reverse populations, and therefore accuracy and stability of structural damage identification are improved, and accuracy and robustness of structural damage identification of the algorithm under incomplete measurement and noise interference are improved.
The present invention is not limited to the preferred embodiments, and any simple modification, equivalent variation and modification made to the above embodiments according to the technical substance of the present invention will still fall within the scope of the technical solution of the present invention.
Claims (10)
1. A method for identifying structural damage, comprising the steps of:
s1, acquiring structural working condition data, and establishing a structural damage identification model;
s2, optimizing the structural damage recognition model by using a locust optimization algorithm improved by a self-adaptive Levy flight strategy and an elite reverse learning strategy to obtain an optimal solution;
The step of optimizing the structural damage recognition model and obtaining an optimal solution by using a locust optimization algorithm improved by the self-adaptive Levy flight strategy and the elite reverse learning strategy comprises the following steps of:
s21, initializing the number N of locust populations and the maximum iteration number L of a locust optimization algorithm, and setting the current iteration number L to be 1;
s22, calculating the fitness value of each locust individual, and determining the locust individual with the optimal fitness value as an initial optimal solution;
s23, updating a linear decreasing coefficient;
s24, standardizing the distance between the locusts to be 1,4, and updating the positions of the locusts;
s25, for the locusts after position updating, updating the positions of the locusts by using a self-adaptive Levy flight strategy, if the fitness value of the locusts after position updating is better than that of the locusts before position updating, reserving the locusts after position updating, otherwise, reserving the locusts before position updating;
s26, updating the locust population by using an elite reverse learning strategy, calculating the fitness value of each locust individual in the updated locust population, and determining the locust individual with the optimal fitness value as the optimal solution of the current iteration;
S27, judging whether the current iteration times L is greater than or equal to the maximum iteration times L, if so, ending the iteration, outputting a global optimal solution based on the initial optimal solution and the optimal solution of each iteration, otherwise, enabling l=l+1, and returning to the step S23.
2. The method of claim 1, wherein the step of updating the location of the individual locust using an adaptive Levy flight strategy comprises:
the self-adaptive nonlinear change factor beta of the curve is calculated, and the calculation formula is as follows:
wherein L is the current iteration number, L is the maximum iteration number, and a and b are formula adjustment coefficients;
updating the locusts individual position:
in the method, in the process of the invention,to update the position of the ith locust individual, X i The position of the ith locust individual before updating.
3. The method of claim 1, wherein the step of updating the locust population using elite reverse learning strategy comprises:
the locust individuals in the locust population are ranked according to the fitness value, the first preset number percent of the locust individuals before ranking are regarded as elite locust population, the first preset number percent of the locust individuals after ranking are regarded as obsolete locust population, and the rest middle locust individuals are regarded as common locust population;
For elite locust population, generating elite reverse population through reverse points, wherein the calculation formula of the reverse points is as follows:
wherein r is el Representing pseudo-random numbers subject to U (0, 1), X i,m ∈[Eu m ,l m ],Eu m =max(X m ),El m =min(X m ),X i,m For the position of the ith locust individual in the m-th dimension, X m Eu is the position vector of all locust individuals in the m-th dimension m And El m Representing the locust individual in the m-th dimensionPosition maxima and minima of (2);
and combining the elite locust population, the elite reverse population and the common locust population to obtain an updated locust population.
4. The method of claim 1, wherein the step of updating the location of the individual locust comprises:
updating the position of the locust individual by the following formula:
in the method, in the process of the invention,mth dimension information, ub, representing the ith locust individual at the 1+1th iteration m And lb m Respectively represent the upper limit and the lower limit of the m-th dimension of the locust individual, and the +.>Representing the optimal position of the mth dimension in the first iteration, parameter c l For the linear decreasing coefficient of the first iteration s is a function and +.>Wherein f represents the attractive force intensity, R represents the attractive force range, c max And c min The maximum value and the minimum value of the linear decreasing coefficient are respectively, L is the current iteration number, and L is the maximum iteration number.
5. The method of claim 1, wherein the step of building a structural damage recognition model comprises:
the function of the structural damage recognition model is as follows:
α opt =arg α minJ(α) (6);
wherein alpha is opt For the optimal injury factor vector, J (α) is the objective function.
6. The method of claim 5, wherein the function of the structural damage recognition model is:
wherein w is 1 、w 2 Is two weight factors, the sum of the two weight factors is 1, N m For the order of the mode shape,i=1, 2, …, N for the i-th order modality m ,K j A unit rigidity contribution matrix of the j-th unit, T is a vector transposition, f i Frequency f of the ith target variable i And (alpha) is the calculated frequency of the ith target variable.
7. The method for identifying structural damage according to claim 5, wherein the step of creating the structural damage identification model further comprises adding a regularization penalty term to the objective function to constrain the L1 norm according to the following calculation formula:
L 1 =‖α‖ 1 /N m (11);
in the formula II alpha II 1 Is the 1 norm of the injury factor vector alpha, |alpha|| 1 =|α 1 |+|α 2 |+…+|α Nele |。
8. A structural damage identification system, comprising:
the acquisition module is used for acquiring the structural working condition data and establishing a structural damage identification model;
The optimization module is used for optimizing the structural damage recognition model by using a locust optimization algorithm improved by a self-adaptive Levy flight strategy and an elite reverse learning strategy and obtaining an optimal solution;
wherein, the optimization module includes:
the initializing sub-module is used for initializing the number N of locust populations and the maximum iteration number L of the locust optimization algorithm, and setting the current iteration number L to be 1;
the calculation sub-module is used for calculating the fitness value of each locust individual and determining the locust individual with the optimal fitness value as an initial optimal solution;
a first updating sub-module for updating the linear decreasing coefficient;
the standardized submodule is used for standardizing the distance between the locusts to be 1,4 and updating the positions of the locusts;
the Levy flight module is used for updating the position of the locust individual by using a self-adaptive Levy flight strategy for the locust individual after the position update, if the fitness value of the locust individual after the position update is better than that of the locust individual before the position update, the locust individual after the position update is reserved, and otherwise, the locust individual before the position update is reserved;
the elite reverse learning module is used for updating the locust population by using an elite reverse learning strategy, calculating the fitness value of each locust individual in the updated locust population, and determining the locust individual with the optimal fitness value as the optimal solution of the current iteration;
The judging module is used for judging whether the current iteration times L is greater than or equal to the maximum iteration times L, if yes, ending the iteration, outputting a global optimal solution based on the initial optimal solution and the optimal solution of each iteration, otherwise, enabling l=l+1, and returning to the first updating sub-module.
9. An electronic device, comprising:
a processor;
a memory for storing an executable computer program;
wherein the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 7.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method according to any one of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311374315.8A CN117744465B (en) | 2023-10-20 | 2023-10-20 | Structural damage identification method, system, equipment and medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311374315.8A CN117744465B (en) | 2023-10-20 | 2023-10-20 | Structural damage identification method, system, equipment and medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117744465A true CN117744465A (en) | 2024-03-22 |
CN117744465B CN117744465B (en) | 2024-06-25 |
Family
ID=90258032
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311374315.8A Active CN117744465B (en) | 2023-10-20 | 2023-10-20 | Structural damage identification method, system, equipment and medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117744465B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105183933A (en) * | 2015-07-15 | 2015-12-23 | 中山大学 | Artificial bee colony algorithm based structural damage recognition method |
CN105956294A (en) * | 2016-05-06 | 2016-09-21 | 中山大学 | Fruit fly algorithm-based structure damage identification method |
CN115563490A (en) * | 2022-10-08 | 2023-01-03 | 南京大学 | Super-large-scale low-orbit satellite fault diagnosis method based on improved locust optimization algorithm |
CN115618673A (en) * | 2022-10-10 | 2023-01-17 | 佛山科学技术学院 | Structural damage identification method based on two-stage particle swarm optimization algorithm |
-
2023
- 2023-10-20 CN CN202311374315.8A patent/CN117744465B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105183933A (en) * | 2015-07-15 | 2015-12-23 | 中山大学 | Artificial bee colony algorithm based structural damage recognition method |
CN105956294A (en) * | 2016-05-06 | 2016-09-21 | 中山大学 | Fruit fly algorithm-based structure damage identification method |
CN115563490A (en) * | 2022-10-08 | 2023-01-03 | 南京大学 | Super-large-scale low-orbit satellite fault diagnosis method based on improved locust optimization algorithm |
CN115618673A (en) * | 2022-10-10 | 2023-01-17 | 佛山科学技术学院 | Structural damage identification method based on two-stage particle swarm optimization algorithm |
Non-Patent Citations (3)
Title |
---|
MINSHUI HUANG等: "Structural damage identification based on modal frequency strain energy assurance criterion and flexibility using enhanced moth-flame optimization", 《ELSEVIER》, vol. 28, 31 December 2020 (2020-12-31), pages 1119 - 1136 * |
张潇: "桥梁损伤特征相似性度量对比分析", 《中国优秀硕士学位论文全文数据库》, no. 5, 15 May 2022 (2022-05-15), pages 1 - 110 * |
王博;刘连生;韩绍程;祝世兴;: "基于多策略融合的混合多目标蝗虫优化算法", 计算机应用, vol. 40, no. 09, 21 May 2020 (2020-05-21), pages 2670 - 2676 * |
Also Published As
Publication number | Publication date |
---|---|
CN117744465B (en) | 2024-06-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Liu et al. | Artificial Neural Network (ANN)-Bayesian Probability Framework (BPF) based method of dynamic force reconstruction under multi-source uncertainties | |
JP2005504367A (en) | Combinatorial method for monitoring neural network learning | |
US20220067588A1 (en) | Transforming a trained artificial intelligence model into a trustworthy artificial intelligence model | |
Su et al. | Gaussian process machine-learning method for structural reliability analysis | |
CN111814342B (en) | Complex equipment reliability hybrid model and construction method thereof | |
CN112232495A (en) | Prediction model training method, device, medium and computing equipment | |
CN108182500A (en) | Ammunition Storage Reliability Forecasting Methodology based on accelerated life test | |
CN116560895B (en) | Fault diagnosis method for mechanical equipment | |
Alkayem et al. | Damage diagnosis in 3D structures using a novel hybrid multiobjective optimization and FE model updating framework | |
Khatamsaz et al. | Efficient multi-information source multiobjective bayesian optimization | |
CN111859250A (en) | Method, system and storage medium for evaluating accuracy of rapid modal parameters under environment and forced excitation | |
Xu et al. | An enhanced squared exponential kernel with Manhattan similarity measure for high dimensional Gaussian process models | |
Xi et al. | An ensemble approach for model bias prediction | |
CN117744465B (en) | Structural damage identification method, system, equipment and medium | |
CN117349160A (en) | Software quality assessment method and device and computer readable storage medium | |
Guo et al. | Data mining and application of ship impact spectrum acceleration based on PNN neural network | |
CN116993548A (en) | Incremental learning-based education training institution credit assessment method and system for LightGBM-SVM | |
KR102441442B1 (en) | Method and apparatus for learning graph convolutional network | |
CN113660236B (en) | Abnormal flow detection method based on optimized stacked noise reduction convolution self-coding network, memory and processor | |
US11455531B2 (en) | Trustworthy predictions using deep neural networks based on adversarial calibration | |
CN114398944A (en) | Unmanned aerial vehicle online parallel anomaly detection method and system under resource-limited condition | |
CN113658173A (en) | Compression method, system and computing equipment of detection model based on knowledge distillation | |
CN112329929A (en) | Countermeasure sample generation method and device based on proxy model | |
CN110728292A (en) | Self-adaptive feature selection algorithm under multi-task joint optimization | |
CN113297773B (en) | Uncertainty quantitative analysis method and device for structural damage identification |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |