CN108921812B - Intelligent evaluation method of circuit breaker spring fatigue state based on image recognition - Google Patents

Abstract

The invention discloses an intelligent evaluation method for a fatigue state of a spring of a circuit breaker based on image recognition, and relates to the technical field of circuit breaker detection. According to the intelligent evaluation method for the fatigue state of the spring of the circuit breaker based on image recognition, a basic database for evaluating the spring state of the circuit breaker is established through multiple off-line test sample statistical analysis and classification, a key motion target position in the spring deformation process is detected and analyzed by adopting a high-speed image sequence and an NCC algorithm, the spring deformation detection in the normal opening and closing process of a high-voltage circuit breaker is realized, a curve representing the fatigue state of the spring is obtained, and therefore a characteristic parameter vector of the fatigue state of the spring is obtained; and analyzing the obtained spring fatigue characteristic parameter vector by adopting a GA-SALBP model to obtain the fatigue state value and the stress relaxation condition of the spring of the circuit breaker, thereby realizing the purpose of monitoring the state of the spring of the high-voltage circuit breaker.

Description

Intelligent evaluation method of spring fatigue state of circuit breaker based on image recognition

technical field

The invention belongs to the technical field of circuit breaker detection, and in particular relates to an intelligent evaluation method for a circuit breaker spring fatigue state based on image recognition.

Background technique

The reliability of the high-voltage circuit breaker operation is very important to the protection and control of the power grid. According to statistics, most of the high-voltage circuit breaker failures are operating mechanism failures, and the circuit breaker spring is an important component of the circuit breaker operating mechanism. By affecting the circuit breaker operating mechanism, it determines whether the circuit breaker can be opened and closed normally. High-voltage circuit breaker springs mainly bear varying loads during normal operation, and most of their failure forms are fatigue damage. Due to the long-term fatigue work of the spring and the sudden breakage, the accident of circuit breaker failure often occurs, so it is of great significance to study the fatigue state monitoring of the operating mechanism spring.

At present, the monitoring of the operating mechanism of the high-voltage circuit breaker is more concentrated on the overall mechanical performance of the operating mechanism, and the monitoring of the spring is measured directly or indirectly by measuring the opening and closing stroke curve of the moving contact, the current of the opening and closing coil, etc. There are very few studies on the method of state assessment. In practice, only regular inspections can be used to find the springs that may have problems, but it is generally difficult to judge the problems of the springs with the naked eye, and it is very labor-intensive, time-consuming and labor-intensive, and the efficiency is not high. The function of the spring fatigue testing machine is to calculate the fatigue curve of the spring, which needs to be tested after the circuit breaker spring is disassembled, and cannot realize the rapid analysis of the force and deformation state of the operating mechanism spring during the operation of the circuit breaker.

Based on this, the present invention proposes an intelligent evaluation method for the fatigue state of a circuit breaker spring based on image recognition, and applies methods such as computer vision, pattern recognition and neural network to the monitoring of the fatigue state of the spring. Using computer vision technology to detect the spring deformation of the circuit breaker operating mechanism, and dynamically evaluate the spring performance during the opening and closing process, is a new method of testing the mechanical state of non-contact circuit breakers. This method is used in the detection of the motion parameters of the circuit breaker spring, and the high-speed camera is used instead of the traditional displacement and deformation sensor to obtain the motion image sequence of the circuit breaker spring. The relevant working parameters of the spring motion curve process are obtained by fitting, and the characteristic vector is established based on the deformation characteristics, and the fatigue state and stress relaxation degree of the spring are judged by intelligent comparison with the spring deformation in the normal state.

SUMMARY OF THE INVENTION

In view of the deficiencies of the prior art, the present invention provides an intelligent evaluation method for the fatigue state of a circuit breaker spring based on image recognition.

The present invention solves the above-mentioned technical problems through the following technical solutions: a method for intelligently evaluating the fatigue state of circuit breaker springs based on image recognition, comprising the following steps:

(1) Establish a basic database for circuit breaker spring state evaluation;

(2) Setting the parameters of the test device for measuring the spring deformation process of the circuit breaker, capturing the images in the spring deformation process, and obtaining the high-speed image sequence of the spring after processing;

(3) Take the pixel coordinates of the fixed end of the spring as the initial point, and the spring between the initial point and the top of the movable end of the spring is the moving target that needs to be identified; select a certain area on the moving target as the matching template, and use the image matching algorithm to identify the target. In the described step (2) high-speed image sequence, each frame of image obtains the corresponding matching template center pixel position coordinates, and this pixel position represents the actual working position of the operating mechanism spring;

(4) calculate the distance from matching template center pixel coordinates to initial point pixel coordinates in each frame of image, draw the displacement-time curve of the moving target in the spring deformation process of the operating mechanism, and match the speed-time curve of the template center pixel coordinates;

(5) Obtain spring working parameters through displacement-time curve and velocity-time curve;

(6) Extracting test samples with obvious representative characteristics from the basic database in the step (1), obtaining the working parameters and fatigue state evaluation values of the test samples, and establishing a fatigue characteristic parameter vector representing the fatigue state of the spring of the operating mechanism of the high-voltage circuit breaker ; Simultaneously obtain the measured spring fatigue characteristic parameter vector;

(7) Use genetic algorithms-self-adaption learning rate Backpropation (GA-SALBP) to analyze the fatigue state of the spring: extract the typical samples in the basic database, and obtain the fatigue characteristic parameters of the typical samples The GA-SALBP model is trained by taking the fatigue characteristic parameter vector of the typical sample as the training sample; then the spring fatigue characteristic parameter vector of the unknown fatigue state is substituted into the trained GA-SALBP model, and the fatigue state value of the circuit breaker spring is obtained. And stress relaxation, and finally set the fatigue accident early warning threshold to achieve the purpose of monitoring the spring state of the high-voltage circuit breaker.

Further, in the basic database of the step (1), the inherent size of the spring is recorded according to the model of the circuit breaker spring, and the basic data of the spring fatigue test is recorded, and the expansion and deformation process of the circuit breaker spring is related to the pressure and fatigue degree.

Further, the test device of the step (2) is a high-speed camera, and the parameters that the high-speed camera needs to set include shooting focal length, trigger speed, camera frame rate, resolution and exposure time, and according to the circuit breaker model and opening and closing. Time sets the shooting duration of the high-speed camera, and determines the actual size of each pixel of the image captured by the high-speed camera in the actual shooting plane.

Further, the method for determining the actual size of each pixel of the photographed image in the actual photographing plane is as follows: placing a scale with a known specific size in the same plane as the photographed circuit breaker spring, and obtaining the pixel length of the scale in the image to obtain the scale. The quotient of the actual size of the ruler and the pixel length is taken as the actual size of each pixel of the image in the actual shooting plane.

Further, the working parameters of the spring in the step (5) include the initial working height of the spring, the final working height of the spring, the expansion and contraction time, the maximum movement stroke of the spring, the number of times of radial reciprocation of the spring intermediate ring and the maximum acceleration of the spring;

Take the distance from the center pixel coordinate of the initial frame matching template to the initial point pixel coordinate as the initial working height of the spring, and take the distance from the center pixel coordinate of the end frame matching template to the initial point pixel coordinate as the final working height of the spring, and take the upper bound of the displacement-time curve and The difference of the lower bound is taken as the maximum travel of the spring h _m .

Further, the initial frame and the end frame are determined by using the grayscale difference method of the corresponding positions of adjacent frames of the image.

Further, the calculation expression of the initial working height H _t0 of the spring is H _t0 =k|s ₀ -s|, the calculation expression of the final working height H _t of the spring is H _t =k|s _t -s|, the spring The telescopic time t is the maximum value of the intersection of y=s _t ±ε and the displacement-time curve;

Among them, s ₀ is the center pixel coordinate of the initial frame matching template, s is the pixel coordinate of the initial point, s _t is the center pixel coordinate of the end frame matching template, k is the actual size of each pixel of the image in the actual shooting plane, and ε is the identified The maximum value of spring expansion and contraction.

Further, described step (7) adopts GA-SALBP model to carry out the concrete process of spring fatigue state analysis as follows:

(7.1) Suppose the input is the N-dimensional vector x of the characteristic parameters of spring fatigue state, and suppose there are M samples in the basic database, calculate the mean value of each characteristic parameter, and compare it with the characteristic value of each sample, excluding the influence of environmental factors Or identify the samples with obvious abnormality in some characteristic parameters caused by recording errors, and leave appropriate M' samples to form an N×M' characteristic parameter matrix of the circuit breaker operating mechanism spring and M' fatigue state degree evaluation values and stress relaxation;

(7.2) Establish a BP neural network, and determine the number of neurons in the middle layer according to the number of characteristic parameters of the spring fatigue state. The product of the number of neurons; the GA genetic algorithm is used to obtain a complete set of weights with a small error in the BP neural network as the initial weights of the BP neural network, and then the sample value, that is, the characteristic parameter matrix of the spring, is substituted into the BP neural network. learning and training;

(7.3) In the training process, the error energy and the learning rate are adjusted according to the spring performance, and the learning rate is adjusted by the change of the error energy after adjusting the weight twice before and after; It can judge the increase or decrease of the error, and there is no need to set the increase/decrease factor. When the error increases, the learning rate is reduced, and the learning rate is appropriately increased in the direction of error reduction;

(7.4) When the error energy is less than the set threshold, that is, the result obtained by inputting the fatigue state characteristic parameters of any spring sample is very close to the actual fatigue state evaluation value of the sample, the BP neural network is trained; Track the spring movement process of the operating mechanism to be tested, calculate the characteristic parameter vector of the fatigue state, and substitute it into the trained BP neural network, that is, analyze its fatigue state, and achieve the purpose of intelligent evaluation of the spring fatigue state of the circuit breaker operating mechanism.

k的取值为大于1的常数。Further, the adjustment expression of the learning rate in the step (7.3) is:

The value of k is a constant greater than 1.

Compared with the prior art, the intelligent evaluation method of circuit breaker spring fatigue state based on image recognition provided by the present invention establishes a basic database for circuit breaker spring state evaluation through statistical analysis and classification of multiple offline test samples, and adopts high-speed image sequence and NCC. The algorithm detects and analyzes the key moving target positions in the spring deformation process, realizes the spring deformation detection in the normal opening and closing process of the high-voltage circuit breaker, obtains the curve representing the spring fatigue state, and obtains the spring fatigue characteristic parameter vector; the GA-SALBP model is used to compare the obtained The spring fatigue characteristic parameter vector is analyzed, that is, the fatigue state value and stress relaxation of the circuit breaker spring are obtained, and the purpose of monitoring the spring state of the high-voltage circuit breaker is realized. When the GA-SALBP model is used to analyze the obtained spring fatigue characteristic parameter vector, the combination of The error energy index of circuit breaker spring performance evaluation realizes the evaluation of test spring performance, which makes the whole evaluation method more effective and faster in calculation speed.

Description of drawings

In order to illustrate the technical solutions of the present invention more clearly, the following briefly introduces the accompanying drawings used in the description of the embodiments. Obviously, the accompanying drawings in the following description are only an embodiment of the present invention, which is very important in the art. For those of ordinary skill, other drawings can also be obtained from these drawings without any creative effort.

Fig. 1 is a schematic flowchart of an image recognition-based intelligent evaluation method for spring fatigue state of circuit breaker according to the present invention;

Fig. 2 is the schematic diagram describing the spring parameter of the present invention;

FIG. 3 is a schematic flow chart of the GA-SALBP neural network of the present invention for judging the spring fatigue state and stress relaxation degree of the operating mechanism.

Detailed ways

The technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present invention.

As shown in FIG. 1 , an image recognition-based intelligent evaluation method for circuit breaker spring fatigue state provided by the present invention includes the following steps:

(1) Establish the basic database of circuit breaker spring state evaluation.

The intrinsic size of the spring is recorded in the basic database according to the type of the circuit breaker spring. The intrinsic size includes the diameter d of the spring, the middle diameter D of the spring, the inner diameter D ₁ , the outer diameter D ₂ , the free height H ₀ , the helix angle α, and the original pitch b, the number of spring coils n (part of the parameters can be derived from each other); at the same time, the basic data of the spring fatigue test are recorded, and the relationship between the expansion and contraction process of the circuit breaker spring and the pressure and fatigue degree, the relationship includes the working height of the spring, the movement speed, the force The key parameters that describe the spring performance, such as the expansion and contraction curve and the oscillation frequency. The data samples in the basic database are on-site simulating the opening and closing state of the circuit breaker through a spring fatigue testing machine. The characteristics of the spring expansion and contraction process are identified by shooting and analysis by a high-speed camera. Combined with the test results of the mechanical sensor, through repeated tests of similar springs in the fatigue testing machine The obtained data samples are stored in the basic database after statistical classification, which is the basic database for spring state evaluation of circuit breakers.

(2) Set the parameters of the test device for measuring the spring deformation process of the circuit breaker, capture the images in the spring deformation process, and obtain the high-speed image sequence of the spring after processing.

The test device is a high-speed camera. Find a suitable position to install the high-speed camera near the spring of the circuit breaker to be tested, and adjust the parameters of the high-speed camera to make the spring image of the operating mechanism clear; the parameters to be set for the high-speed camera include shooting focal length, trigger speed, camera frame rate, resolution rate and exposure time, and set the shooting time of the high-speed camera according to the type of circuit breaker and the opening and closing time, and determine the actual size k (mm) of each pixel of the image captured by the calibrated high-speed camera in the actual shooting plane.

The method of determining the actual size k is: place a ruler with a known specific size in the same plane as the photographed circuit breaker spring, obtain the pixel length of the ruler in the image, and take the quotient of the actual size of the ruler and the pixel length as each pixel of the image The actual size k in the actual shooting plane is easy to calculate the final spring working parameters, and compared with the spring free state parameters, making the results more intuitive.

When the current trigger of the circuit breaker opening and closing coil exceeds the threshold, it is judged that the opening and closing of the circuit breaker starts, and the high-speed camera is controlled to start capturing images, and then the high-speed image sequence of the high-voltage circuit breaker spring is obtained through image acquisition, preprocessing, and storage.

(3) Search the initial frame of the spring motion in the high-speed image sequence, take the pixel coordinates of the fixed end of the spring as the initial point A, and use it to calculate the actual working length and movement speed changes of the spring during the deformation process. (ie the matching template) is the target point B, and the spring between the initial point and the target point is the moving target to be identified (as shown in Figure 2); That is the characteristic internode of the spring of the obvious circuit breaker operating mechanism.

The image matching algorithm (NCC algorithm) is used to identify step (2) the central pixel position of the spring top adjacent area (ie matching template) of each frame of image in the high-speed image sequence (ie the pixel position of the spring feature node in each frame of image) , the pixel position represents the actual working position of the operating mechanism spring.

(4) Calculate the distance from the pixel coordinates of the target point B to the pixel coordinates of the initial point A in each frame of the image (that is, the working height of the spring), draw the displacement-time curve of the moving target during the spring deformation of the operating mechanism, and the target point B Velocity-time curve, the velocity value is the point-by-point slope value of the displacement-time curve of the spring target point B in the movement direction.

(5) Obtain the working parameters of the spring through the displacement-time curve and the speed-time curve. The working parameters of the spring include the initial working height of the spring, the final working height of the spring, the expansion and contraction time, the maximum movement stroke of the spring, the number of times of radial reciprocation of the spring intermediate ring and the spring. The maximum acceleration of work; the distance from the pixel coordinate s ₀ of the initial frame target point to the pixel coordinate s of the initial point is used as the initial working height H _t0 of the spring, and the distance from the pixel coordinate s _t of the end frame target point to the pixel coordinate s of the initial point is used as the final spring of the spring. Working height H _t , the expansion and contraction time t of the spring is obtained from the displacement-time curve, and the difference between the upper and lower bounds of the displacement-time curve is taken as the maximum movement stroke h _m of the spring, and the number of times of radial reciprocation of the middle ring of the spring in the expansion and contraction time is calculated. f _c (that is, the number of changes in the direction of the target point B on the spring during the spring expansion and contraction time), and obtain the maximum acceleration a _m of the spring.

The calculation expression of the initial working height H _t0 of the spring is H _t0 =k|s ₀ -s|, the calculation expression of the final working height H _t of the spring is H _t =k|s _t -s|, and the spring expansion and contraction time t is y = s _t ±ε and the maximum value of the intersection of the displacement-time curve; ε is the maximum value of the identified spring expansion and contraction amplitude, and ε takes 0-2 pixels by default, eliminating the fact that the spring expansion and contraction amplitude is too small, which may be caused by identification errors The action of , k is a coefficient, which reflects the calculated relationship between the actual size of the spring and the pixel length.

The initial frame and the end frame are determined by the gray difference method of the corresponding positions of adjacent frames of the image, and a large number of irrelevant still frame images are filtered out, so as to avoid a large number of redundant images before and after the circuit breaker operation affecting the processing speed.

(6) Extract the test samples with obvious representative characteristics from the basic database in step (1), and obtain the working parameters and fatigue state evaluation values of the test samples; through the extraction of principal components, analyze the correlation of the data between the test samples to obtain the spring The spring working parameters that may change due to the change of the fatigue state, establish the fatigue characteristic parameter vector representing the fatigue state of the spring of the operating mechanism of the high-voltage circuit breaker to reduce the complexity of the parameters; at the same time, obtain the measured spring fatigue characteristic parameter vector [h _m , a _m ,f _c ];

(7) Use genetic algorithms-self-adaption learning rate Backpropation (GA-SALBP) to analyze the fatigue state of the spring: extract the typical samples in the basic database, and obtain the fatigue characteristic parameters of the typical samples The GA-SALBP model is trained by taking the fatigue characteristic parameter vector of the typical sample as the training sample; then the spring fatigue characteristic parameter vector of the unknown fatigue state is substituted into the trained GA-SALBP model, and the fatigue state value of the circuit breaker spring is obtained. (value between 0 and 1) and stress relaxation conditions, and finally set the fatigue accident early warning threshold (which can be adjusted according to actual work requirements) as the early warning value for the occurrence of spring fatigue fracture accidents, to achieve the purpose of high-voltage circuit breaker spring state monitoring.

As shown in Figure 3, the specific process of using the GA-SALBP model to analyze the fatigue state of the spring is as follows:

(7.1) Suppose the input is the N-dimensional vector x of the characteristic parameters of spring fatigue state, and suppose there are M samples in the basic database, calculate the mean value of each characteristic parameter, and compare it with the characteristic value of each sample, excluding the influence of environmental factors Or identify the samples with obvious abnormality in some characteristic parameters caused by recording errors, and leave appropriate M' samples to form an N×M' characteristic parameter matrix of the circuit breaker operating mechanism spring and M' fatigue state degree evaluation values and Stress relaxation.

(7.2) Establish a three-layer BP neural network, and determine the number of neurons in the middle layer according to the number of spring fatigue state characteristic parameters. The number of weights of the BP neural network is the number of spring characteristic parameters N plus the number of evaluation indicators and the The product of the number of neurons in the middle layer; the GA genetic algorithm is used to obtain a complete set of weights with small errors in the BP neural network as the initial weights of the BP neural network, and then the sample value, that is, the characteristic parameter matrix of the spring, is substituted into the BP neural network. Learning and training of neural networks.

(即实际弹簧疲劳状态程度评估值及应力松弛度的误差值平方和的

e_j(n)为第j次弹簧疲劳状态程度评估值与应力松弛度误差)及学习率η调整权值w，用前后两次调整权值后误差能量的变化调整学习率η，学习率的调整表示式为

p的取值为大于1的常数，η(n)为第n次训练过程中的学习率，为调整后第n+1次训练过程学习率，e(n)和e(n+1)为递归的两次训练弹簧性能评估误差能量；根据需要调整学习率的变化快慢，优点在于不仅免去了误差的增减判断，而且不用设定增/减因子，在误差增大的时候降低学习率，误差减小方向适当提高学习率。(7.3) Evaluate error energy from spring properties during training

(that is, the sum of the squares of the actual spring fatigue state evaluation value and the error value of the stress relaxation degree

e _j (n) is the j-th spring fatigue state evaluation value and stress relaxation error) and the learning rate η to adjust the weight w, and the learning rate η is adjusted by the change of the error energy after adjusting the weight twice before and after. The adjusted expression is

The value of p is a constant greater than 1, η(n) is the learning rate in the nth training process, and is the learning rate in the n+1th training process after adjustment, e(n) and e(n+1) are Recursive two training spring performance evaluation error energy; adjust the speed of change of the learning rate according to the need, the advantage is that not only does the error increase or decrease judgment, but also does not need to set the increase/decrease factor, and reduce the learning rate when the error increases , the error reduction direction appropriately increases the learning rate.

(7.4) When the error energy is less than the set threshold, that is, the result obtained by inputting the fatigue state characteristic parameters of any spring sample is very close to the actual fatigue state evaluation value of the sample, the BP neural network is trained; Track the spring movement process of the operating mechanism to be tested, calculate the characteristic parameter vector of the fatigue state, and substitute it into the trained BP neural network, that is, analyze its fatigue state, and achieve the purpose of intelligent evaluation of the spring fatigue state of the circuit breaker operating mechanism.

The use of genetic algorithm to obtain the initial weights and thresholds of the BP neural network accelerates the convergence speed of the BP neural network while ensuring the calculation accuracy of the BP neural network. The convergence rate is also relatively fast.

The above disclosure is only the specific embodiment of the present invention, but the protection scope of the present invention is not limited to this. should be included within the protection scope of the present invention.

Claims (8)

1. an intelligent evaluation method for circuit breaker spring fatigue state based on image recognition, is characterized in that, comprises the following steps: (1) Establish a basic database for circuit breaker spring state evaluation; (2) Setting the parameters of the test device for measuring the spring deformation process of the circuit breaker, capturing the images in the spring deformation process, and obtaining the high-speed image sequence of the spring after processing; (3) Take the pixel coordinates of the fixed end of the spring as the initial point, and the spring between the initial point and the top of the movable end of the spring is the moving target that needs to be identified; select a certain area on the moving target as the matching template, and use the image matching algorithm to identify the target. In the step (2) high-speed image sequence, each frame of image obtains the corresponding matching template center pixel position coordinates; (4) calculate the distance from matching template center pixel coordinates to initial point pixel coordinates in each frame of image, draw the displacement-time curve of the moving target in the spring deformation process of the operating mechanism, and match the speed-time curve of the template center pixel coordinates; (5) Obtain spring working parameters through displacement-time curve and velocity-time curve; (6) Extracting test samples with obvious representative characteristics from the basic database in the step (1), obtaining the working parameters and fatigue state evaluation values of the test samples, and establishing a fatigue characteristic parameter vector representing the fatigue state of the spring of the operating mechanism of the high-voltage circuit breaker ; Simultaneously obtain the measured spring fatigue characteristic parameter vector; (7) Using the GA-SALBP model to analyze the fatigue state of the spring: extract the typical samples in the basic database, obtain the fatigue characteristic parameter vector of the typical sample, and use the fatigue characteristic parameter vector of the typical sample as the training sample to train the GA-SALBP model; Then, the spring fatigue characteristic parameter vector of the unknown fatigue state is substituted into the trained GA-SALBP model to obtain the fatigue state value and stress relaxation of the circuit breaker spring. Finally, the fatigue accident warning threshold is set to realize the monitoring of the spring state of the high voltage circuit breaker. the goal of; The specific process of using the GA-SALBP model to analyze the fatigue state of the spring in the step (7) is as follows: (7.1) Suppose the input is the N-dimensional vector x of the characteristic parameters of spring fatigue state, and suppose there are M samples in the basic database, calculate the mean value of each characteristic parameter, and compare it with the characteristic value of each sample, excluding the influence of environmental factors Or identify the samples with obvious abnormality in some characteristic parameters caused by recording errors, and leave appropriate M' samples to form an N×M' characteristic parameter matrix of the circuit breaker operating mechanism spring and M' fatigue state degree evaluation values and stress relaxation; (7.2) Establish a BP neural network, and determine the number of neurons in the middle layer according to the number of characteristic parameters of the spring fatigue state. The product of the number of neurons; the GA genetic algorithm is used to obtain a complete set of weights with a small error in the BP neural network as the initial weights of the BP neural network, and then the sample value, that is, the characteristic parameter matrix of the spring, is substituted into the BP neural network. learning and training; (7.3) In the training process, the error energy and the learning rate are adjusted according to the spring performance evaluation, and the learning rate is adjusted by the change of the error energy after adjusting the weights twice before and after; (7.4) When the error energy is less than the set threshold, that is, the result obtained by inputting the fatigue state characteristic parameters of any spring sample is very close to the actual fatigue state evaluation value of the sample, the BP neural network is trained; Track the spring movement process of the operating mechanism to be tested, calculate the characteristic parameter vector of the fatigue state, and substitute it into the trained BP neural network, that is, analyze its fatigue state, and achieve the purpose of intelligent evaluation of the spring fatigue state of the circuit breaker operating mechanism. 2. The method for intelligently evaluating the fatigue state of a circuit breaker spring based on image recognition as claimed in claim 1, wherein the basic database of the step (1) records the inherent size of the spring according to the circuit breaker spring model, and simultaneously records There are basic data of spring fatigue test, and the relationship between the expansion and contraction process of circuit breaker spring is related to pressure and fatigue degree. 3. the circuit breaker spring fatigue state intelligent evaluation method based on image recognition as claimed in claim 1, is characterized in that, the test device of described step (2) is high-speed camera, and the parameter that described high-speed camera needs to set includes shooting focal length , trigger speed, camera frame rate, resolution and exposure time, and set the shooting time of the high-speed camera according to the circuit breaker model and opening and closing time, and determine the actual size of each pixel of the image captured by the calibrated high-speed camera in the actual shooting plane. 4. The method for intelligently evaluating the fatigue state of circuit breaker springs based on image recognition as claimed in claim 3, wherein the method for determining the actual size of each pixel of the photographed image in the actual photographing plane is: A ruler with a known specific size is placed in the same plane, the pixel length of the ruler in the image is obtained, and the quotient of the actual size of the ruler and the pixel length is taken as the actual size of each pixel of the image in the actual shooting plane. 5. The method for intelligently evaluating the fatigue state of circuit breaker springs based on image recognition according to claim 1, wherein the step (5) spring working parameters include the initial working height of the spring, the final working height of the spring, the expansion and contraction time, the spring The maximum movement stroke, the number of times of radial reciprocation of the intermediate ring of the spring and the maximum acceleration of the spring; Take the distance from the center pixel coordinate of the initial frame matching template to the initial point pixel coordinate as the initial working height of the spring, and take the distance from the center pixel coordinate of the end frame matching template to the initial point pixel coordinate as the final working height of the spring, and take the upper bound of the displacement-time curve and The difference of the lower bound is taken as the maximum travel of the spring h _m . 6 . The method for intelligently evaluating the fatigue state of circuit breaker springs based on image recognition according to claim 5 , wherein the initial frame and the end frame are determined by using a grayscale difference method of corresponding positions of adjacent frames of the image. 7 . 7 . The method for intelligently evaluating the fatigue state of a circuit breaker spring based on image recognition according to claim 5 , wherein the calculation expression of the initial working height H _t0 of the spring is H _t0 =k|s ₀ −s|, 8 . The calculation expression of the final working height H _t of the spring is H _t =k|s _t -s|, and the spring expansion and contraction time t is the maximum value of the intersection of y=s _t ±ε and the displacement-time curve; Among them, s ₀ is the center pixel coordinate of the initial frame matching template, s is the pixel coordinate of the initial point, s _t is the center pixel coordinate of the end frame matching template, k is the actual size of each pixel of the image in the actual shooting plane, and ε is the identified The maximum value of spring expansion and contraction. 8. The method for intelligently evaluating the fatigue state of circuit breaker springs based on image recognition according to claim 1, wherein the adjustment expression of the learning rate in the step (7.3) is:

The value of p is a constant greater than 1, and e(n) and e(n+1) are the error energies of the recursive two training spring performance evaluations.

Images