CN112613219A - Floating plate track steel spring damage positioning and identifying method - Google Patents

Floating plate track steel spring damage positioning and identifying method Download PDF

Info

Publication number
CN112613219A
CN112613219A CN202011582783.0A CN202011582783A CN112613219A CN 112613219 A CN112613219 A CN 112613219A CN 202011582783 A CN202011582783 A CN 202011582783A CN 112613219 A CN112613219 A CN 112613219A
Authority
CN
China
Prior art keywords
output
damage
steel spring
evidence
model
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
Application number
CN202011582783.0A
Other languages
Chinese (zh)
Other versions
CN112613219B (en
Inventor
任娟娟
郑健龙
欧成章
杜威
邓世杰
刘伟
章恺尧
荀宇星
叶文龙
曾学勤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Southwest Jiaotong University
Changsha University of Science and Technology
China Railway Liuyuan Group Co Ltd
Original Assignee
Southwest Jiaotong University
Changsha University of Science and Technology
China Railway Liuyuan Group Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Southwest Jiaotong University, Changsha University of Science and Technology, China Railway Liuyuan Group Co Ltd filed Critical Southwest Jiaotong University
Priority to CN202011582783.0A priority Critical patent/CN112613219B/en
Publication of CN112613219A publication Critical patent/CN112613219A/en
Application granted granted Critical
Publication of CN112613219B publication Critical patent/CN112613219B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)

Abstract

The invention discloses a floating plate track steel spring damage positioning and identifying method, which adopts 4 BP neural networks to respectively identify the damage position of a steel spring on 4 damage indexes of displacement, curvature, strain and strain energy, obtains independent evidence bodies, and carries out fusion identification on the evidence through a D-S evidence fusion theory, thereby realizing the organic combination of the neural networks and the evidence theory. The method can effectively improve the identification accuracy of the damaged position of the steel spring, reduces the uncertainty of the identification process, has a certain application value in the field of ballastless track damage identification, and is beneficial to related personnel to quickly maintain the track.

Description

Floating plate track steel spring damage positioning and identifying method
Technical Field
The invention relates to the field of floating plate track steel spring damage positioning and identification, in particular to a floating plate track steel spring damage positioning and identification method
Background
By 9 months in 2020, 41 cities in China have opened and operated urban rail transit lines of more than 7000 kilometers, wherein the occupation ratio of the subways is about 77%. In order to relieve the vibration and noise generated in the operation process of the subway, the floating slab track is widely applied to domestic and foreign subways as an effective vibration and noise reduction structure. The floating plate track is mainly divided into a rubber support floating plate track and a steel spring floating plate track according to different supporting conditions, wherein the steel spring floating plate track has the best vibration isolation effect, can effectively reduce the vibration of the track structure and eliminate solid-borne sound, and is suitable for surrounding buildings such as research institutions, hospitals, museums, music halls and other line sections with higher vibration isolation requirements.
Along with the increase of the operation time of the subway line, the steel spring vibration isolator can lose effectiveness under the action of repeated train load, once the steel spring is damaged, the rigidity uniformity and integrity of the rail system are reduced, and the vibration reduction and noise reduction capability is reduced. Studies have shown that steel spring damage exacerbates the dynamic response of the rail system and affects more significantly as the number of isolators increases. However, for the cast-in-place floating slab track, because the number of the steel springs arranged on one floating slab is large, the damage condition of individual steel springs is difficult to detect, and the problem that the damage positioning and identifying reliability of the steel springs of the floating slab track is low in the existing method also exists.
Disclosure of Invention
Aiming at the defects in the prior art, the method for positioning and identifying the damage of the steel spring of the floating plate track solves the problem that the reliability of positioning and identifying the damage of the steel spring of the floating plate track is low in the prior art.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that:
the method for positioning and identifying the damage of the steel spring of the floating plate track comprises the following steps:
s1, establishing a finite element model according to the steel spring floating plate track in the known damage state, and acquiring a displacement mode, a curvature mode, a strain mode and a strain energy mode in the finite element model;
s2, respectively inputting a displacement mode, a curvature mode, a strain mode and a strain energy mode in the finite element model into 4 BP neural networks, and training the 4 BP neural networks;
s3, performing back propagation calculation on the 4 BP neural networks according to the errors of the damage states output by the 4 BP neural networks and the actual damage states in the training process, updating the parameters of the 4 BP neural networks until the error of the identification result of each BP neural network is less than or equal to a set value, and obtaining the trained 4 BP neural networks;
s4, constructing a BP-D-S model: constructing the output of the trained 4 BP neural networks into 4 evidence bodies, fusing the 4 evidence bodies by adopting a D-S evidence theory fusion model, wherein the fusion result is the output result of the BP-D-S model;
s5, acquiring and taking a displacement mode, a curvature mode, a strain mode and a strain energy mode on the target steel spring floating plate track as the input of the BP-D-S model to obtain the output of the BP-D-S model;
and S6, acquiring an identification position according to the acquisition position of the data input into the BP-D-S model, and finishing the damage identification of the steel spring of the floating plate track according to the output of the BP-D-S model.
Further, each BP neural network in step S2 includes an input layer, a hidden layer, and an output layer, which are connected in sequence; wherein:
the output of the hidden layer is:
Figure BDA0002865540410000021
the output of the output layer is:
Figure BDA0002865540410000031
the error is:
Figure BDA0002865540410000032
wherein f is1() is the transfer function of the input layer to the hidden layer; w is aijIs the weight; b1jIs a threshold value; x is the number ofiIs an input sample; n is the total number of samples; s is the total number of hidden layer neurons; f. of2() is the transfer function from the hidden layer to the output layer; t isjkIs the weight; b2kIs a threshold value; y isjIs the output of the hidden layer; m is the total output number of the hidden layer; t is tkThe expected output is determined by the real damage state; z is a radical ofkIs the output of the output layer; and E is an error.
Further, the set value in step S3 is 0.001.
Further, the specific method for fusing the 4 evidence bodies by using the D-S evidence theory fusion model in step S4 includes the following sub-steps:
s4-1, according to the formula:
Figure BDA0002865540410000033
obtaining focal element for forming the first evidence body
Figure BDA0002865540410000034
Basic degree of confidence of
Figure BDA0002865540410000035
Wherein n is the total number of focal elements; elOutputting the error sum for the BP neural network corresponding to n focal elements in the l evidence body; 1,2,3 and 4, which correspond to 4 evidences respectively;
Figure BDA0002865540410000036
is the focal element in the first evidence
Figure BDA0002865540410000037
The output of the corresponding BP neural network;
s4-2, according to the formula:
Figure BDA0002865540410000038
Figure BDA0002865540410000039
acquiring the ith focal element B in the new evidence body B after the fusion of the ith focal element in the 1 st evidence body and the 2 nd evidence bodyiBasic confidence m (B) ofi) Further obtaining the basic credibility of each focal element in the new evidence body B; wherein
Figure BDA0002865540410000041
Representing an empty set;
Figure BDA0002865540410000042
to representThe ith focal element in the 1 st evidence body;
Figure BDA0002865540410000043
represents the ith focal element in the 2 nd evidence body; p is the coefficient of conflict between evidence bodies;
s4-3, fusing the new evidence body B and the 3 rd evidence body by adopting the same method as the step S4-2 to obtain a new evidence body C and the basic credibility of each focal element in the new evidence body C;
s4-4, fusing the new evidence body C and the 4 th evidence body by adopting the same method as the step S4-2 to obtain a new evidence body D and the basic credibility of each focal element in the new evidence body D;
and S4-5, taking the basic reliability of each focal element in the new evidence body D as a fusion result at the corresponding position on the target steel spring floating plate track, namely the output result of the BP-D-S model.
Further, the specific method for completing the floating plate rail steel spring damage identification according to the output of the BP-D-S model in step S6 is as follows:
if the output value of the BP-D-S model is greater than 0.6, judging that the floating plate track steel spring at the position is damaged; and if the output value of the BP-D-S model is less than or equal to 0.6, judging that the floating plate track steel spring at the position is not damaged.
The invention has the beneficial effects that: the method adopts 4 BP neural networks to respectively identify the damage position of the steel spring on 4 damage indexes of displacement, curvature, strain and strain energy, obtains each independent evidence body, and carries out fusion identification on the evidence through a D-S evidence fusion theory, thereby realizing the organic combination of the neural networks and the evidence theory. The method can effectively improve the identification accuracy of the damaged position of the steel spring, reduces the uncertainty of the identification process, has a certain application value in the field of ballastless track damage identification, and is beneficial to related personnel to quickly maintain the track.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a plan view of a floating plate track solid model containing steel spring damage in the example;
FIG. 3 is a cross-sectional view taken along line A-A of FIG. 2;
FIG. 4 is a schematic diagram of a displacement mode shape obtained by simulation in the embodiment;
FIG. 5 is a schematic diagram of a curvature mode shape obtained by simulation in the embodiment;
FIG. 6 is a schematic diagram of a strain mode shape obtained by simulation in the example;
FIG. 7 is a schematic diagram of a strain energy mode obtained by simulation in an embodiment;
FIG. 8 is a schematic diagram of displacement modes at different damage levels at position 1 in the example;
FIG. 9 is a schematic view showing the mode of curvature at different damage levels at position 1 in the example;
FIG. 10 is a schematic diagram of the strain mode at different damage levels at position 1 in the example;
FIG. 11 is a schematic diagram of strain energy modes at different damage levels at position 1 in the example.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
As shown in fig. 1, the method for locating and identifying the damage of the steel spring of the floating plate track comprises the following steps:
s1, establishing a finite element model according to the steel spring floating plate track in the known damage state, and acquiring a displacement mode, a curvature mode, a strain mode and a strain energy mode in the finite element model;
s2, respectively inputting a displacement mode, a curvature mode, a strain mode and a strain energy mode in the finite element model into 4 BP neural networks, and training the 4 BP neural networks;
s3, performing back propagation calculation on the 4 BP neural networks according to the errors of the damage states output by the 4 BP neural networks and the actual damage states in the training process, updating the parameters of the 4 BP neural networks until the error of the identification result of each BP neural network is less than or equal to a set value, and obtaining the trained 4 BP neural networks; the set value is 0.001;
s4, constructing a BP-D-S model: constructing the output of the trained 4 BP neural networks into 4 evidence bodies, fusing the 4 evidence bodies by adopting a D-S evidence theory fusion model, wherein the fusion result is the output result of the BP-D-S model;
s5, acquiring and taking a displacement mode, a curvature mode, a strain mode and a strain energy mode on the target steel spring floating plate track as the input of the BP-D-S model to obtain the output of the BP-D-S model;
and S6, acquiring an identification position according to the acquisition position of the data input into the BP-D-S model, and finishing the damage identification of the steel spring of the floating plate track according to the output of the BP-D-S model.
Each BP neural network in the step S2 comprises an input layer, a hidden layer and an output layer which are connected in sequence; wherein:
the output of the hidden layer is:
Figure BDA0002865540410000061
the output of the output layer is:
Figure BDA0002865540410000062
the error is:
Figure BDA0002865540410000063
wherein f is1() is the transfer function of the input layer to the hidden layer; w is aijIs the weight; b1jIs a threshold value; x is the number ofiIs an input sample; n is the total number of samples; s is the total number of hidden layer neurons; f. of2() is the transfer function from the hidden layer to the output layer; t isjkIs the weight; b2kIs a threshold value; y isjIs the output of the hidden layer; m is the total output number of the hidden layer; t is tkThe expected output is determined by the real damage state; z is a radical ofkIs the output of the output layer; and E is an error.
The specific method for fusing the 4 evidence bodies by adopting the D-S evidence theory fusion model in the step S4 comprises the following substeps:
s4-1, according to the formula:
Figure BDA0002865540410000071
obtaining focal element for forming the first evidence body
Figure BDA0002865540410000072
Basic degree of confidence of
Figure BDA0002865540410000073
Wherein n is the total number of focal elements; elOutputting the error sum for the BP neural network corresponding to n focal elements in the l evidence body; 1,2,3 and 4, which correspond to 4 evidences respectively;
Figure BDA0002865540410000074
is the focal element in the first evidence
Figure BDA0002865540410000075
The output of the corresponding BP neural network;
s4-2, according to the formula:
Figure BDA0002865540410000076
Figure BDA0002865540410000077
obtaining the No. 1 evidence body and the No. 2 evidence body in the new evidence body B after the fusion of the No. i focal elementi focal elements BiBasic confidence m (B) ofi) Further obtaining the basic credibility of each focal element in the new evidence body B; wherein
Figure BDA0002865540410000078
Representing an empty set;
Figure BDA0002865540410000079
represents the ith focal element in the 1 st evidence body;
Figure BDA00028655404100000710
represents the ith focal element in the 2 nd evidence body; p is the coefficient of conflict between evidence bodies;
s4-3, fusing the new evidence body B and the 3 rd evidence body by adopting the same method as the step S4-2 to obtain a new evidence body C and the basic credibility of each focal element in the new evidence body C;
s4-4, fusing the new evidence body C and the 4 th evidence body by adopting the same method as the step S4-2 to obtain a new evidence body D and the basic credibility of each focal element in the new evidence body D;
and S4-5, taking the basic reliability of each focal element in the new evidence body D as a fusion result at the corresponding position on the target steel spring floating plate track, namely the output result of the BP-D-S model.
The specific method for completing the floating plate track steel spring damage identification according to the output of the BP-D-S model in the step S6 is as follows: if the output value of the BP-D-S model is greater than 0.6, judging that the floating plate track steel spring at the position is damaged; and if the output value of the BP-D-S model is less than or equal to 0.6, judging that the floating plate track steel spring at the position is not damaged.
In a specific implementation process, when the neural network training is carried out, the LM algorithm can be selected as a training function, the expected error is 0.001, the maximum training iteration number is 100, the learning rate is 0.1, and the number of the neurons in the hidden layer is selected through trial calculation according to the principle of minimizing the training error.
In one embodiment of the invention, a floating plate track solid model containing steel spring damage is established by ABAQUS, the floating plate model with different damage degrees at different positions of a steel spring is analyzed, and the fastener of the floating plate track and the steel spring are specifically arranged as shown in figures 2 and 3. In fig. 2 and 3, the transverse spacing of the steel springs is 2.0m, the longitudinal spacing thereof close to the floating slab end is 0.625m, and the spacing of the middle part is 1.25 m; the distance between the fasteners and the end close to the board is 0.3125m, and the middle part is 0.625 m.
In the model, the steel rail and the floating plate are simulated by solid units, the fasteners and the steel springs are simulated by spring units, and the values of relevant track structure parameters are shown in a table 1. The two ends of the steel rail and the bottom of the steel spring are fully constrained, and the two end surfaces of the floating plate, the two side surfaces in the length direction and the bottom surface of the steel rail are fully constrained by the other degrees of freedom except the degrees of freedom of vertical displacement and rotation around the z axis.
Table 1: steel spring floating plate orbit model parameter
Figure BDA0002865540410000081
Figure BDA0002865540410000091
The embodiment aims at different damage positions of the steel springs, and considers the symmetry of the model, starting from the first pair of steel springs at the end of the floating plate and sequentially going backwards until the ninth pair of steel springs. Meanwhile, the rigidity of the damaged steel spring, namely K, is calculated by introducing a damage coefficient alpha for representing different damage degrees of the steel spring at the same positiond＝(1-α)K0，K0The steel spring rate is in a damage-free state. In the embodiment, the damage coefficients of 0.1-1.0 and the increments of 0.1 are set at the same damage position of the steel spring, the damage degrees are respectively 10% -100%, the increments are 10%, and the modes of the floating plates of the steel spring after the steel spring is damaged in different degrees at 9 different positions are calculated.
In the embodiment, the intersection point of the horizontal connecting line of the central axis of the floating slab and the fastener is taken as 36 data nodes, the damage and damage of the floating slab track on the steel springs are respectively generated in the first pair of steel springs (position 1) to the ninth pair of steel springs (position 9) through simulation calculation, and the distribution rule of 4 types of damage indexes with the damage degrees of 100% along the longitudinal direction of the model is shown in fig. 4, 5, 6 and 7. As can be seen from fig. 4, 5, 6 and 7, the 4-type damage indexes of the floating slab have a large difference between the positions 1 to 9 where the steel spring is not damaged and the damage occurs respectively, and it can be seen that the modal parameter selected by the invention is sensitive to the damage position and can be used for identifying the damage location of the steel spring.
Taking the damage position 1 as an example, the distribution rule of the damage indexes at the same damage position and different damage degrees (0-100%) along the longitudinal direction of the model is calculated and obtained as shown in fig. 8, 9, 10 and 11. As can be seen from fig. 8, 9, 10 and 11, when the steel spring at position 1 is damaged, the 4-class damage index of the floating plate is very insensitive to the damage degree, and the modal parameters obtained under different damage degrees are very different. Therefore, the method can only identify the damage position of the steel spring, and simultaneously neglect the influence of the damage degree of the steel spring on the damage position, so that the identification effect precision of the method can be effectively guaranteed.
In the embodiment, the number of the damage positions of the steel spring is 9, each damage position corresponds to 10-100% of 10 damage degrees, and each damage index obtained through calculation has 90 samples. For each injury position, samples with injury degrees of 30%, 60% and 90% were selected as tests, 27 in total, and the remaining samples were used for training of the BP neural network. Taking the output value of the neural network at the position 1-4 with the damage degree of 30% in the test sample as an example, the damage judgment threshold value is set to be 0.6, namely when the output value of the BP-D-S model exceeds 0.6, the steel spring at the position is damaged. The BP neural network local output values are shown in table 2.
Table 2: local output value of BP neural network
Figure BDA0002865540410000101
Figure BDA0002865540410000111
As can be seen from the above table, the 4 sub-neural networks constructed by using the 4 types of damage indexes can preliminarily identify the damage positions of the steel springs, but it can be seen from the identification result that the damage positions of the steel springs in a plurality of test samples cannot be judged only according to the output value of the single BP neural network, and the identification accuracy is not high.
Basic belief distributions of the test samples were constructed for the 4 sub-BP neural networks described above, with the results shown in table 3. And then fusing the 4 types of damage indexes to obtain sample fusion identification results of 4 different damage positions as shown in table 4.
Table 3: basic belief allocation for BP neural networks
Figure BDA0002865540410000112
Figure BDA0002865540410000121
Table 4: D-S evidence theory fusion results
Figure BDA0002865540410000122
As can be seen from table 4 above, the damage positions of some steel springs cannot be given by using a single BP neural network, and can be accurately identified after evidence fusion processing. In order to comprehensively evaluate the effect of the steel spring floating plate damage positioning and identifying method based on multi-modal index fusion, 4 types of single indexes of displacement, curvature, strain and strain energy and the identification accuracy of the fusion index on 27 test samples are contrastively analyzed, and the result is shown in table 5.
Table 5: comparison of single index and fused index recognition results
Figure BDA0002865540410000131
As can be seen from table 5, for the same test sample, a higher recognition accuracy can be achieved by using the fusion index than by selecting a single modal index for recognition, which indicates that complementary information between different modalities can be fully used after the method utilizes the evidence fusion theory to fuse the single index, thereby improving the reliability of location recognition.
In conclusion, the invention adopts 4 BP neural networks to respectively identify the damage position of the steel spring on 4 damage indexes of displacement, curvature, strain and strain energy, obtains each independent evidence body, and carries out fusion identification on the evidence through a D-S evidence fusion theory, thereby realizing the organic combination of the neural networks and the evidence theory. The method can effectively improve the identification accuracy of the damaged position of the steel spring, reduces the uncertainty of the identification process, has a certain application value in the field of ballastless track damage identification, and is beneficial to related personnel to quickly maintain the track.

Claims (5)

1. A floating plate rail steel spring damage positioning and identifying method is characterized by comprising the following steps:
s1, establishing a finite element model according to the steel spring floating plate track in the known damage state, and acquiring a displacement mode, a curvature mode, a strain mode and a strain energy mode in the finite element model;
s2, respectively inputting a displacement mode, a curvature mode, a strain mode and a strain energy mode in the finite element model into 4 BP neural networks, and training the 4 BP neural networks;
s3, performing back propagation calculation on the 4 BP neural networks according to the errors of the damage states output by the 4 BP neural networks and the actual damage states in the training process, updating the parameters of the 4 BP neural networks until the error of the identification result of each BP neural network is less than or equal to a set value, and obtaining the trained 4 BP neural networks;
s4, constructing a BP-D-S model: constructing the output of the trained 4 BP neural networks into 4 evidence bodies, fusing the 4 evidence bodies by adopting a D-S evidence theory fusion model, wherein the fusion result is the output result of the BP-D-S model;
s5, acquiring and taking a displacement mode, a curvature mode, a strain mode and a strain energy mode on the target steel spring floating plate track as the input of the BP-D-S model to obtain the output of the BP-D-S model;
and S6, acquiring an identification position according to the acquisition position of the data input into the BP-D-S model, and finishing the damage identification of the steel spring of the floating plate track according to the output of the BP-D-S model.
2. The floating plate rail steel spring damage positioning and identifying method according to claim 1, wherein each BP neural network in the step S2 includes an input layer, a hidden layer and an output layer which are connected in sequence; wherein:
the output of the hidden layer is:
Figure FDA0002865540400000011
the output of the output layer is:
Figure FDA0002865540400000021
the error is:
Figure FDA0002865540400000022
wherein f is1() is the transfer function of the input layer to the hidden layer; w is aijIs the weight; b1jIs a threshold value; x is the number ofiIs an input sample; n is the total number of samples; s is the total number of hidden layer neurons; f. of2() is the transfer function from the hidden layer to the output layer; t isjkIs the weight; b2kIs a threshold value; y isjIs the output of the hidden layer; m is the total output number of the hidden layer; t is tkThe expected output is determined by the real damage state; z is a radical ofkIs the output of the output layer; and E is an error.
3. The floating plate rail steel spring damage location identification method of claim 1, wherein the set value in step S3 is 0.001.
4. The floating plate rail steel spring damage positioning and identifying method as claimed in claim 1, wherein the specific method for fusing 4 evidence bodies by using the D-S evidence theory fusion model in the step S4 comprises the following substeps:
s4-1, according to the formula:
Figure FDA0002865540400000023
obtaining focal element for forming the first evidence body
Figure FDA0002865540400000024
Basic degree of confidence of
Figure FDA0002865540400000025
Wherein n is the total number of focal elements; elOutputting the error sum for the BP neural network corresponding to n focal elements in the l evidence body; 1,2,3 and 4, which correspond to 4 evidences respectively;
Figure FDA0002865540400000026
is the focal element in the first evidence
Figure FDA0002865540400000027
The output of the corresponding BP neural network;
s4-2, according to the formula:
Figure FDA0002865540400000028
Figure FDA0002865540400000031
acquiring the ith focal element B in the new evidence body B after the fusion of the ith focal element in the 1 st evidence body and the 2 nd evidence bodyiBasic confidence m (B) ofi) Further obtaining the basic credibility of each focal element in the new evidence body B; wherein
Figure FDA0002865540400000032
Representing an empty set;
Figure FDA0002865540400000033
represents the ith focal element in the 1 st evidence body;
Figure FDA0002865540400000034
represents the ith focal element in the 2 nd evidence body; p is the coefficient of conflict between evidence bodies;
s4-3, fusing the new evidence body B and the 3 rd evidence body by adopting the same method as the step S4-2 to obtain a new evidence body C and the basic credibility of each focal element in the new evidence body C;
s4-4, fusing the new evidence body C and the 4 th evidence body by adopting the same method as the step S4-2 to obtain a new evidence body D and the basic credibility of each focal element in the new evidence body D;
and S4-5, taking the basic reliability of each focal element in the new evidence body D as a fusion result at the corresponding position on the target steel spring floating plate track, namely the output result of the BP-D-S model.
5. The method for locating and identifying damage to a steel spring of a floating plate rail according to claim 1, wherein the specific method for completing identification of damage to a steel spring of a floating plate rail according to the output of the BP-D-S model in step S7 comprises:
if the output value of the BP-D-S model is greater than 0.6, judging that the floating plate track steel spring at the position is damaged; and if the output value of the BP-D-S model is less than or equal to 0.6, judging that the floating plate track steel spring at the position is not damaged.
CN202011582783.0A 2020-12-28 2020-12-28 Floating plate track steel spring damage positioning and identifying method Active CN112613219B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011582783.0A CN112613219B (en) 2020-12-28 2020-12-28 Floating plate track steel spring damage positioning and identifying method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011582783.0A CN112613219B (en) 2020-12-28 2020-12-28 Floating plate track steel spring damage positioning and identifying method

Publications (2)

Publication Number Publication Date
CN112613219A true CN112613219A (en) 2021-04-06
CN112613219B CN112613219B (en) 2022-03-15

Family

ID=75248450

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011582783.0A Active CN112613219B (en) 2020-12-28 2020-12-28 Floating plate track steel spring damage positioning and identifying method

Country Status (1)

Country Link
CN (1) CN112613219B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5327358A (en) * 1991-08-07 1994-07-05 The Texas A&M University System Apparatus and method for damage detection
CN104316341A (en) * 2014-11-17 2015-01-28 金陵科技学院 Underground structure damage identification method based on BP neural network

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5327358A (en) * 1991-08-07 1994-07-05 The Texas A&M University System Apparatus and method for damage detection
CN104316341A (en) * 2014-11-17 2015-01-28 金陵科技学院 Underground structure damage identification method based on BP neural network

Also Published As

Publication number Publication date
CN112613219B (en) 2022-03-15

Similar Documents

Publication Publication Date Title
Chen et al. Damage detection in long suspension bridges using stress influence lines
Cha et al. Structural damage detection using modal strain energy and hybrid multiobjective optimization
Au et al. Parameter identification of vehicles moving on continuous bridges
Mangalathu et al. Ground motion-dependent rapid damage assessment of structures based on wavelet transform and image analysis techniques
Xin et al. Uncertainty and sensitivity analysis for train-ballasted track–bridge system
CN101145214A (en) Cable-stayed bridge cable damage positioning method based on modified reverse transmittance nerve network
Li et al. Probabilistic safety analysis of coupled train-bridge system using deep learning based surrogate model
Guo et al. Optimal sensor placement for damage detection of bridges subject to ship collision
CN105096614A (en) Newly established crossing traffic flow prediction method based on generating type deep belief network
Zhang et al. Data‐driven system parameter change detection for a chain‐like uncertainties embedded structure
Sediek et al. Seismic debris field for collapsed RC moment resisting frame buildings
Tabrizian et al. Charged system search algorithm utilized for structural damage detection
Raich et al. Improving the performance of structural damage detection methods using advanced genetic algorithms
Li et al. Bearing damage detection of a bridge under the uncertain conditions based on the Bayesian framework and matrix perturbation method
CN112613219B (en) Floating plate track steel spring damage positioning and identifying method
CN117669389B (en) Random vibration analysis method for earthquake-vehicle-bridge system based on deep learning
Lin et al. Dynamic finite element model updating of prestressed concrete continuous box-girder bridge
Siemaszko et al. The idea of using Bayesian networks in forecasting impact of traffic-induced vibrations transmitted through the ground on residential buildings
Xu et al. Development of BP-based seismic behavior optimization of RC and steel frame structures
Xie et al. Research on Vibration Fatigue Damage Locations of Offshore Oil and Gas Pipelines Based on the GA‐Improved BP Neural Network
Jeng et al. Quick seismic response estimation of prestressed concrete bridges using artificial neural networks
Kerh et al. Neural network approach for analyzing seismic data to identify potentially hazardous bridges
Gopalakrishnan et al. Global optimization of pavement structural parameters during back-calculation using hybrid shuffled complex evolution algorithm
Shi et al. Estimation of cement asphalt mortar disengagement degree using vehicle dynamic response
Liu et al. Damage identification of urban overpass based on hybrid neurogenetic algorithm using static and dynamic properties

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