Vacuum breaking type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction method, system and device and medium

Abstract

The invention discloses a vacuum breaking type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction method and system, equipment and a medium, and relates to the technical field of intelligent state monitoring and operation and maintenance management of power equipment, and the method comprises the specific steps: collecting and preprocessing a signal, and constructing a multi-dimensional feature vector; establishing a multi-scale fatigue damage identification model to obtain a damage grade classification result; and collecting an accumulated damage historical sequence for preprocessing, calculating through a calculation module, constructing a life prediction model, and predicting the life of the GIS operating mechanism. Through multi-source data collaborative analysis, multi-scale fatigue damage identification and life prediction model construction, the problems that a traditional method cannot realize continuous tracking and prediction of the damage evolution process of the operating mechanism and cannot fully capture multi-frequency response and local damage signals in the operating load are solved; the problems of diagnosis lag, low accuracy and weak generalization in the prior art are solved, and the operation safety and the maintenance scientificity of the high-voltage electrical equipment are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent state monitoring and operation and maintenance management of power equipment, and particularly relates to a vacuum breaking type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction method, system, device and medium. BACKGROUND

[0002] With the development of ultra-high voltage power transmission technology, GIS equipment is widely used in power systems due to its compactness, high reliability and environmental adaptability. The core execution unit of GIS equipment, the operating mechanism, undertakes the key task of controlling breaking, closing and other operation actions. However, the vacuum breaking type operating mechanism is subjected to complex coupling effects such as electromagnetic drive impact, mechanical friction and temperature difference stress during long-term operation, which makes the internal moving parts prone to micro fatigue cracks, plastic deformation and connection loosening and other damage problems. The existing operation and maintenance system relies on periodic maintenance and manual experience judgment. Another method uses a single-scale fatigue evaluation method to capture multi-frequency responses and local damage signals in the operating load, which can also achieve the purpose of saving operation and maintenance resources and solving unnecessary safety hazards, forming a system of inspection and maintenance procedures.

[0003] However, the disadvantages of such processing methods are:

[0004] (1) unable to continuously track and predict the damage evolution process of the operating mechanism;

[0005] (2) the traditional single-scale fatigue evaluation method cannot fully capture multi-frequency responses and local damage signals in the operating load, and has limitations such as diagnostic lag, low accuracy and weak generalization. SUMMARY

[0006] In view of the above existing problems, the present application is proposed.

[0007] To solve the above technical problems, the present application provides the following technical solutions:

[0008] In a first aspect, the present application provides a vacuum breaking type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction method, comprising:

[0009] Laying sensors to collect signals and constructing a signal set;

[0010] Preprocessing the signal set and constructing a feature vector for the preprocessed signal set;

[0011] According to the feature vector, a multi-scale fatigue damage identification model is established to obtain damage grade classification results;

[0012] Collecting and preprocessing the cumulative damage history sequence, and generating a new sequence by accumulating the cumulative damage history sequence.

[0013] The new sequence is input into the uncertainty system analysis method, a cumulative sequence prediction function is obtained, and a life prediction model is constructed;

[0014] According to the life prediction model, a critical threshold is set, and the life of the GIS operating mechanism is predicted in combination with the damage grade classification result.

[0015] As a preferred scheme of the multi-scale fatigue damage real-time monitoring and residual life prediction method of the vacuum breaking type GIS operating mechanism, wherein: a multi-scale fatigue damage identification model is established, and a damage grade classification result is obtained, including:

[0016] According to the signal set, a time-frequency domain transformation method is used to enhance the local features of the signal, and frequency domain features are extracted;

[0017] According to the frequency domain features, a damage sensitive index is constructed, and a multi-dimensional feature vector is constructed.

[0018] The beneficial effects of the preferred technical scheme are:

[0019] By using the time-frequency domain transformation method to enhance the local features of the signal, the frequency domain features are extracted, the damage sensitive index is constructed, and the multi-dimensional feature vector is constructed, which solves the problems that the traditional single-scale fatigue evaluation method cannot fully capture the multi-frequency response and local damage signal in the running load, and has the problems of diagnostic lag, low accuracy and weak generalization.

[0020] As a preferred scheme of the multi-scale fatigue damage real-time monitoring and residual life prediction method of the vacuum breaking type GIS operating mechanism, wherein, according to the life prediction model, a critical threshold is set, and the life of the GIS operating mechanism is predicted in combination with the damage grade classification result, including:

[0021] According to the life prediction model, the cumulative damage value after each future operation is sequentially predicted;

[0022] A critical threshold is set, and the damage grade classification result is combined to determine when the critical threshold is first reached, and the corresponding period is the residual life.

[0023] The beneficial effects of the preferred technical scheme are:

[0024] By constructing the life prediction model, the relationship between the cumulative generated sequence and the prediction value is dynamically reflected, and the problem of being unable to realize continuous tracking and prediction of the damage evolution process of the operating mechanism in the above technical problems is solved.

[0025] In a second aspect, the present application provides a multi-scale fatigue damage real-time monitoring and residual life prediction system for a vacuum breaking type GIS operating mechanism, comprising:

[0026] An acquisition module acquires signals collected by the sensors to form a signal set;

[0027] A preprocessing and vector construction module pre-processes the signal set and constructs a feature vector based on the pre-processed signal set;

[0028] A damage identification module establishes a multi-scale fatigue damage identification model based on the feature vector to obtain a damage level classification result;

[0029] A calculation module collects and pre-processes a cumulative damage history sequence, inputs an uncertainty system analysis method, and obtains a cumulative sequence prediction function;

[0030] A life prediction module combines the cumulative sequence prediction function, constructs a life prediction model, sets a critical threshold, and combines the damage level classification result to predict the life of the GIS operating mechanism.

[0031] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of the method for real-time monitoring and residual life prediction of multi-scale fatigue damage of a vacuum interrupter type GIS operating mechanism when executing the computer program.

[0032] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the method for real-time monitoring and residual life prediction of multi-scale fatigue damage of a vacuum interrupter type GIS operating mechanism when executed by a processor.

[0033] Compared with the prior art, the present application has the following advantages: the present application obtains a damage level classification result by constructing a multi-scale fatigue damage identification model, and predicts the life of the GIS operating mechanism by constructing a life prediction model. The system integrates multi-source sensors, signal multi-scale processing algorithms, damage identification models, and life prediction mechanisms, and can dynamically and finely depict the health status and life decay law of the operating mechanism, thereby effectively supporting state maintenance, early warning, and precise operation and maintenance. BRIEF DESCRIPTION OF DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0035] Figure 1 The overall flowchart of the method for real-time monitoring and residual life prediction of multi-scale fatigue damage of a vacuum interrupter type GIS operating mechanism according to an embodiment of the present application is shown.

[0036] Figure 2 A schematic diagram of a classification result of a multi-scale fatigue damage real-time monitoring and residual life prediction method of a vacuum interrupter type GIS operating mechanism according to an embodiment of the present application is shown in the figure.

[0037] Figure 3 A prediction model result and life trend curve diagram of a multi-scale fatigue damage real-time monitoring and residual life prediction method of a vacuum interrupter type GIS operating mechanism according to an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0038] In order to make the above objectives, characteristics and advantages of the present application more apparent, obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor should fall within the scope of protection of the present application.

[0039] Embodiment 1, Reference Figure 1 According to an embodiment of the present application, a multi-scale fatigue damage real-time monitoring and residual life prediction method of a vacuum interrupter type GIS operating mechanism is provided, comprising:

[0040] S100: arranging sensors to collect signals and constructing a signal set;

[0041] S200: pre-processing the signal set and constructing a feature vector for the pre-processed signal set;

[0042] S300: establishing a multi-scale fatigue damage identification model according to the feature vector to obtain damage grade classification results;

[0043] S400: collecting and pre-processing a cumulative damage history sequence and inputting an uncertainty system analysis method to obtain a cumulative sequence prediction function;

[0044] S500: performing an inversion operation according to the cumulative sequence prediction function to construct a life prediction model;

[0045] S600: combining the life prediction model, setting a critical threshold, combining the damage grade classification results, and predicting the life of the GIS operating mechanism.

[0046] As the core execution unit of GIS equipment, GIS operating mechanism undertakes the key task of controlling the operation of opening and closing. In its operation process, multi-frequency response and local damage signal are the key features, and if they cannot be fully captured, it will lead to diagnostic lag, low accuracy and weak generalization. With the increase of operation time, the operating mechanism will continue to be damaged, so it is also important to continuously track and predict the damage evolution process of the operating mechanism.

[0047] Therefore, in view of the above-mentioned dynamic monitoring and health prediction problems, through the steps of S100-S600, the sensor is arranged to collect signals, the signal set is constructed, the feature vector is constructed, the multi-scale fatigue damage identification model is established, the multi-frequency response and local damage signal and other key features are monitored in real time, and the problems of diagnostic lag, low accuracy and weak generalization caused by missing key features are avoided. Collecting the cumulative damage history sequence, processing it, inputting the uncertainty system analysis method, constructing the life prediction model, combining the damage level classification result, and realizing the continuous tracking and health prediction of the damage evolution process of the operating mechanism.

[0048] Embodiment 2, refer to Figure 2 、 Figure 3 For an embodiment of the present application, based on the above embodiment, a multi-scale fatigue damage real-time monitoring and residual life prediction method for vacuum opening type GIS operating mechanism is provided.

[0049] In the embodiment of the application, in step S100, the sensor is arranged to collect signals, and the signal set is constructed, including the following steps A1-A2:

[0050] A1: installing a sensor in the GIS operating mechanism to monitor the structural stress state signal and the mechanical state signal, and obtaining the dynamic signal according to the feedback of the equipment;

[0051] A2: unifying all sensor signals to construct a signal set.

[0052] Specifically, the specific steps of collecting signals and constructing signal sets in A1-A2 are as follows:

[0053] High-sensitivity strain gauges are arranged at strain-sensitive components in the operating mechanism to ensure accurate monitoring of the fatigue-sensitive area; acceleration sensors are installed at the ends of the transmission shaft and the connection points of the transmission arm to monitor the operation impact and vibration characteristics; current and voltage sensors are arranged at the electromagnetic coil or capacitive energy storage element to capture the operation trigger timing and excitation strength.

[0054] A high sampling rate (≥10 kHz) data acquisition card (DAQ) is used to manage and synchronously upload all kinds of sensor signals with unified time stamp.

[0055] It should be noted that the data acquisition card (DAQ) is preferred because it can integrate multi-source data signals, manage and synchronize uploading of various sensor signals with unified time stamps, realize coverage of comprehensive data signal dimensions, and avoid the problem of single data signal dimension affecting the detection result.

[0056] In the embodiment of the present application, the signal set is preprocessed in step S200, and a multi-dimensional feature vector is constructed based on the preprocessed signal set, including the following steps B1-B2:

[0057] B1: using a time-frequency domain transformation method to enhance the local features of the signal and extract the frequency domain features;

[0058] B2: constructing a damage sensitive index based on the frequency domain features and constructing a multi-dimensional feature vector.

[0059] In the embodiment of the present application, the frequency domain transformation method in B1 is a wavelet packet decomposition method, which can be represented by the following formula:

[0060]

[0061] where p i represents the energy proportion of the i-th sub-band, E i is the original energy of the i-th sub-band, ∑ j E j is the total energy of all sub-bands, j is the sub-band index, and i is the wavelet packet decomposition layer number.

[0062] It should be noted that the wavelet packet decomposition method uses db4 mother wavelet, decomposes 3 to 5 layers, and obtains multi-band sub-signals. On each sub-band, the energy proportion is calculated to capture the evolution characteristics of the frequency energy distribution with structural damage.

[0063] In the embodiment of the present application, the damage sensitive index in B2 is constructed using the information entropy idea, and the energy entropy is defined as:

[0064]

[0065] where H is the energy entropy, p i is the energy proportion of the i-th sub-band, n is the total number of sub-bands, and the minus sign - represents the positive normalization of the entropy value.

[0066] It should be noted that the energy entropy can effectively measure the complexity and non-stationary changes of the signal, and is an important index for damage identification. Finally, a multi-dimensional feature vector is formed by different sub-band energy proportions, energy entropies, peak amplitudes, etc.

[0067] In an alternative embodiment, the time-frequency domain transformation method is used in step B1 to enhance the local features of the signal, and the frequency domain features are extracted. The combination of time domain statistical features can also be used to achieve this, but it is susceptible to random noise interference, and is only effective for macroscopic damage, with a sensitivity that is weaker than the technical solution adopted by the present application.

[0068] In another alternative embodiment, the time-frequency domain transformation method is used in step B1 to enhance the local features of the signal, and the frequency domain features are extracted. The 1D-CNN automatic feature extraction can also be used to achieve this, but since it uses a CNN model, it needs to be trained with ten-thousand-level samples, and the feature interpretability is weak.

[0069] In the embodiments of the present application, the multi-scale fatigue damage identification model is established in step S300 to obtain the damage level classification result, including the following steps C1-C3:

[0070] C1: training known state samples using a multi-scale fatigue damage identification model;

[0071] C2: inputting the feature vector into the trained multi-scale fatigue damage identification model;

[0072] C3: outputting the health category to which the current operating mechanism belongs, and outputting the damage level classification result.

[0073] In the embodiments of the present application, the multi-scale fatigue damage identification model in C1 uses a support vector machine (SVM) model, which specifically includes the following steps:

[0074] The multi-scale fatigue damage identification uses an SVM classifier to train known state samples, trains a support vector machine (SVM) model, and uses a radial basis function (RBF) as the kernel function to improve the recognition accuracy of non-linearly separable data.

[0075] An exemplary result after classification using the SVM in the embodiments of the present application is shown in the following figure: Figure 2

[0076] It should be noted that this model is preferred because its core logic is divided into two parts: small sample high generalization, based on the principle of structural risk minimization, maximizing the classification interval under limited samples to avoid overfitting; high-dimensional feature space processing, converting non-linear features such as energy entropy and frequency band energy proportion into a separable space through the RBF kernel function to solve the linearly inseparable problem.

[0077] Specifically, in C2-C3, the output damage level classification result specifically includes the following steps:

[0078] ​The optimal kernel width and penalty factor C are selected by grid search and cross-validation method, the real-time collected feature vector is input into the trained SVM model, the model outputs the health category to which the current operating mechanism running state belongs, and online damage level identification is realized.

[0079] For example, Table 1 is a damage level classification table:

[0080] Table 1 is a damage level classification table.

[0081]

[0082] It should be noted that this method is preferred because it combines the grid search method and the cross-validation method, integrates the comprehensiveness of network search and the robustness of cross-validation, maximizes the classification accuracy, and controls the model complexity.

[0083] In an alternative embodiment, in step S300, the multi-scale fatigue damage identification model can also be established by using a random forest algorithm to construct a multi-decision tree ensemble learning, sample and train each tree, and finally vote to determine the damage classification result. However, this scheme has a fuzzy classification boundary and cannot output a clear decision like SVM.

[0084] In an alternative embodiment, in step S300, the multi-scale fatigue damage identification model can also be established by using a one-dimensional convolutional neural network method to directly learn end-to-end from the original time-frequency signal and automatically extract damage-sensitive features. However, this method requires a large amount of data, ≥500 labeled samples, and actual fault samples are scarce.

[0085] In the embodiments of the present application, in step S400, the cumulative damage history sequence is collected and preprocessed, and input into the uncertainty system analysis method to obtain the analysis method expression, including the following steps D1-D2:

[0086] D1: Collect the cumulative damage history sequence, perform accumulation to generate a new sequence;

[0087] D2: Input the new sequence into the uncertainty system analysis method to obtain the analysis method expression.

[0088] Specifically, in D1, the new sequence can be represented by the following formula:

[0089]

[0090] where D (1) (k) is the cumulative damage value after the kth operation, k is the number of operations, D (0) (i) is the damage value of the original cumulative damage sequence after the ith operation.

[0091] It should be noted that in the cumulative damage history sequence, the mechanical damage accumulation itself conforms to the exponential growth law, but this characteristic will be covered by noise, so it is preprocessed by the cumulative generation method to suppress high-frequency noise, highlight the exponential growth law of damage, and meet the requirements of the following input.

[0092] In the embodiment of the present application, the uncertainty system analysis method in D2 adopts a grey prediction model IGM(1, 1), and the analysis method expression can be embodied by the following formula:

[0093]

[0094] D (1) is the cumulative sum of the original damage sequence, t is the operation number, a is the development coefficient, b is the grey action amount, and c is the linear correction term coefficient.

[0095] Exemplarily, D Figure 3 is a grey prediction model result and a life trend curve diagram in the embodiment of the present application.

[0096] It should be noted that, is the instantaneous change rate of the cumulative sequence, which is used to quantify the dynamic characteristics of damage evolution.

[0097] In an alternative embodiment, the new sequence is input into the uncertainty system analysis method in step D2 to obtain the analysis method expression, which can also be realized by adopting a long short-term memory network (LSTM) time series prediction, but it needs to be trained with ≥1000 groups of samples, which is not suitable for small sample scenarios.

[0098] In another alternative embodiment, the new sequence is input into the uncertainty system analysis method in step D2 to obtain the analysis method expression, which can also be realized by adopting a Bayesian filtering framework to represent the probability distribution of the damage state with a random particle swarm, but since it adopts a random particle swarm, resampling will cause the loss of particle diversity, and the number of particles needs to be manually adjusted.

[0099] In the embodiment of the present application, the analysis method expression is solved in step S500 to obtain the prediction function of the cumulative sequence, and a life prediction model is constructed, including the following steps E1-E2:

[0100] E1: the analysis method expression is solved by using the least square method to obtain the prediction function of the cumulative sequence;

[0101] E2: the predicted value of the cumulative sequence is inverted to the original cumulative damage sequence to construct a life prediction model.

[0102] In the embodiment of the present application, in E1, the prediction function of the cumulative sequence can be embodied by the following formula:

[0103]

[0104] wherein, is the predicted value of the 1-AGO sequence at time t, D (1) (1) is an accumulated value, a is a development coefficient, b is a grey action amount, c is a coefficient of a linear correction term, t is the number of operations, e -a(t-1) is an exponential function term.

[0105] It should be noted that the prediction function needs to use the least squares method to solve the above analysis method expression to obtain, represents the initial accumulated data, reflecting the basic situation of the cumulative damage accumulation after the first few operations; a represents the development coefficient of the model, which reflects the change trend rate of the cumulative damage accumulation sequence; b represents the grey action amount, embodying the comprehensive influence of other factors on the cumulative damage accumulation process in addition to the basic development trend; c represents the coefficient of the linear correction term, used to make up for the deficiency of the traditional grey model in fitting the nonlinear trend; e -a(t-1) represents the exponential function term, based on the development coefficient a and the number of operations t, describes the trend characteristics of the accumulated damage sequence with the number of operations, and embodies the exponential growth or decay law that may exist in the damage accumulation process by means of the exponential form.

[0106] In the embodiments of the present application, the life prediction model in E2 can be embodied by the following formula:

[0107]

[0108] wherein is the predicted value of the k+1th data of the 1-AGO sequence, is the first data of the original cumulative damage sequence, and k is the operation number index.

[0109] It should be noted that the traditional grey model GM(1,1) is only applicable to monotonically changing sequences, while mechanical damage accumulation often presents nonlinear acceleration characteristics. By inversely restoring the predicted accumulated value to the original damage value, a life prediction model is established, which retains the nonlinear evolution characteristics and achieves the technical effect of more accurately capturing the damage acceleration inflection point.

[0110] In an optional implementation, the life prediction model constructed in step S500 can also be implemented by adopting a degradation modeling based on a Wiener process, regarding the cumulative damage evolution as a random process with drift. This model requires that the damage increment obeys a Gaussian distribution, but the actual damage may be nonlinearly mutated.

[0111] In another alternative embodiment, the construction of the life prediction model in step S500 can also be achieved by adopting a proportional hazard model to associate damage accumulation with external covariates to construct a failure risk function, but this scheme requires sufficient failure cases and has high computational complexity.

[0112] In the embodiments of the present application, the life of the GIS operating mechanism is predicted in step S600 in combination with the life prediction model, the setting of the critical threshold, and the damage grade classification result, including the following steps F1-F2:

[0113] F1: sequentially predicting the cumulative damage value after each future operation according to the life prediction model;

[0114] F2: setting a critical threshold and combining the damage grade classification result to determine when it first reaches the critical threshold, and the corresponding period is the remaining life.

[0115] Specifically, the setting rule of the critical threshold in F1-F2 can be embodied by the following formula:

[0116]

[0117] wherein, D lim is the critical threshold, D base is the basic threshold, K t is the stress concentration coefficient, and m is a material constant.

[0118] It should be noted that the basic threshold D base is determined based on the material fatigue characteristics, and standard fatigue tests are performed on the key materials of the operating mechanism. The corresponding values of cumulative damage and crack initiation and macroscopic fracture under different load levels are recorded, and the damage-failure probability curve is fitted through the test data. The damage value at 95% survival rate is taken as the basic threshold D base . The basic threshold D base is corrected, and the stress concentration coefficient Kt is introduced for correction at the positions with sharp corners, bolt holes, and welds (such as the shaft hole transition of the operating mechanism connecting rod), to obtain the critical threshold D lim .

[0119] In summary, the present application solves the problems of traditional methods that cannot continuously track and predict the damage evolution process of the operating mechanism, cannot fully capture the multi-frequency response and local damage signals in the operating load, and have diagnostic lag, low accuracy, and weak generalization, significantly improving the safety of high-voltage electrical equipment operation and the scientificity of maintenance.

[0120] Embodiment 3, a schematic scheme of a vacuum interrupter type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction method. It should be noted that the technical scheme of the vacuum interrupter type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction system belongs to the same concept as the above-mentioned technical scheme of the vacuum interrupter type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction method. The technical scheme of the vacuum interrupter type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction system in this embodiment is not described in detail. The details can be referred to the description of the technical scheme of the vacuum interrupter type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction method.

[0121] The embodiment also provides a vacuum interrupter type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction system, comprising:

[0122] The acquisition module acquires signals collected by the sensor and constructs a signal set.

[0123] The preprocessing and vector construction module pre-processes the signal set and constructs a feature vector for the pre-processed signal set.

[0124] The damage identification module establishes a multi-scale fatigue damage identification model according to the feature vector and obtains a damage grade classification result.

[0125] The calculation module collects and pre-processes a cumulative damage history sequence, inputs an uncertainty system analysis method, and obtains a cumulative sequence prediction function.

[0126] The life prediction module combines the cumulative sequence prediction function, constructs a life prediction model, sets a critical threshold, combines the damage grade classification result, and predicts the life of the GIS operating mechanism.

[0127] The embodiment also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of the vacuum interrupter type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction method when executing the computer program.

[0128] The embodiment provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the steps of the vacuum interrupter type GIS operating mechanism multi-scale fatigue damage real-time monitoring and residual life prediction method.

[0129] Those skilled in the art can clearly understand the present application by the above description of the embodiments, and the present application can be realized by software and necessary general hardware, and of course, can also be realized by hardware. Based on such understanding, the technical solutions of the present application or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a floppy disk, a read-only memory (ROM), a random access memory (RAM), a FLASH, a hard disk, or an optical disc, and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods of various embodiments of the present application.

[0130] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application, and although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and all should be covered in the scope of the claims of the present application.

Claims

1. A method for real-time monitoring of multi-scale fatigue damage and residual life prediction of vacuum interrupter type GIS operating mechanism, characterized in that, The method comprises the following steps: The sensor is laid out to collect signals and construct a signal set; The signal set is preprocessed, and a multi-dimensional feature vector is constructed based on the preprocessed signal set; A multi-scale fatigue damage identification model is established based on the feature vector, and a damage level classification result is obtained; The accumulated damage history sequence is preprocessed and input into an uncertainty system analysis method to obtain an accumulated sequence prediction function; An inversion operation is performed based on the accumulated sequence prediction function to construct a life prediction model; A critical threshold is set based on the life prediction model, and the life of the GIS operating mechanism is predicted in combination with the damage level classification result.

2. The method according to claim 1, characterized in that, The sensor is laid out to collect signals and construct a signal set, which comprises: A sensor is installed in the GIS operating mechanism to monitor structural stress state signals and mechanical state signals, and dynamic signals are obtained based on device feedback; All sensor signals are unified to construct a signal set.

3. The method of claim 2, wherein the method is characterized by: The signal set is preprocessed, and a multi-dimensional feature vector is constructed based on the preprocessed signal set, which comprises: The signal local features are enhanced using a time-frequency domain transformation method based on the signal set, and frequency domain features are extracted; Based on the frequency domain features, a damage sensitive index is constructed, and a multi-dimensional feature vector is constructed.

4. The method of claim 3, wherein the method is characterized by: The multi-scale fatigue damage identification model is established, which comprises: Known state samples are trained for the multi-scale fatigue damage identification model; The feature vector is input into the trained multi-scale fatigue damage identification model; The model outputs the health category to which the current operating mechanism operating state belongs, and outputs the damage level classification result.

5. The method of claim 4, wherein the method is characterized by: The accumulated damage history sequence is preprocessed and input into an uncertainty system analysis method to obtain an analysis method expression, which comprises: The accumulated damage history sequence is preprocessed to obtain a new sequence: where D (1) (k) is the cumulative damage value after the kth operation, k is the number of operations, D (0) (i) is the damage value of the original cumulative damage sequence after the ith operation, n is the number of recorded operations; The new sequence is input into the uncertainty system analysis method to obtain an analysis method expression: where D (1) is the cumulative sum of the original damage sequence, t is the number of operations, a is the development coefficient, b is the grey action, and c is the linear correction term coefficient.

6. The method of claim 5, wherein the method is characterized by: The analysis method expression is solved to obtain a prediction function of the accumulated sequence, and a life prediction model is constructed, which comprises: The analysis method expression is solved to obtain a prediction function of the accumulated sequence: wherein, is the predicted value of the 1-AGO sequence at time t, D (1) (1) is the cumulative value, a is the development coefficient, b is the ash amount, c is the coefficient of the linear correction term, t is the number of operations, e -a(t-1) is the exponential function term; The life prediction model is constructed: wherein is the prediction of the (k+1)th data of the sequence generated by 1st order accumulation, is the first data of the original cumulative damage sequence, and k is the operation number index.

7. The method of claim 6, wherein the method is characterized by: Based on the life prediction model, a critical threshold is set, and the life of the GIS operating mechanism is predicted in combination with the damage level classification result, which comprises: The accumulated damage value after each future operation is sequentially predicted based on the life prediction model; A critical threshold is set, and the first time when the critical threshold is reached is determined based on the damage level classification result, and the corresponding period is the remaining life.

8. A multi-scale fatigue damage real-time monitoring and residual life prediction system for vacuum interrupter type GIS operating mechanism, applying the method of any one of claims 1-7, characterized in that, It comprises: An acquisition module is arranged to lay out a sensor to collect signals and construct a signal set; A preprocessing and vector construction module is arranged to preprocess the signal set and construct a feature vector based on the preprocessed signal set; A damage identification module is arranged to establish a multi-scale fatigue damage identification model based on the feature vector, and obtain a damage level classification result; A calculation module is arranged to preprocess the accumulated damage history sequence and input it into an uncertainty system analysis method to obtain an accumulated sequence prediction function; A life prediction module is arranged to construct a life prediction model based on the accumulated sequence prediction function, set a critical threshold, and predict the life of the GIS operating mechanism in combination with the damage level classification result. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The processor implements the steps of the method for real-time monitoring and residual life prediction of multi-scale fatigue damage of a vacuum interrupter type GIS operating mechanism according to any one of claims 1 to 7 when the computer program is executed.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program, when executed by the processor, implements the steps of the method for real-time monitoring and residual life prediction of multi-scale fatigue damage of a vacuum interrupter type GIS operating mechanism according to any one of claims 1 to 7.

Images