A method, device and equipment for classifying the operating state of a surge arrester and a storage medium

Abstract

The application discloses a lightning arrester operation state classification method, device and equipment and a storage medium. The method comprises the following steps: obtaining online monitoring data of a lightning arrester; inputting the online monitoring data into a trained classification vector machine to obtain an original classification result of the online monitoring data; taking the original classification result as an observation sequence, obtaining an optimal labeling sequence of the observation sequence by using a conditional random field, and determining a target classification result of the online monitoring data according to the optimal labeling sequence. The embodiment of the application can realize accurate classification of the operation state of the lightning arrester.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of lightning arrester operation state classification, and particularly relates to a lightning arrester operation state classification method, device, equipment and storage medium. BACKGROUND

[0002] The zinc oxide lightning arrester (MOA) is a device used to protect power equipment. Due to the non-linear characteristic of the zinc oxide lightning arrester resistance, the current flowing through the lightning arrester is extremely small under normal voltage. However, under overvoltage, the zinc oxide lightning arrester resistance rapidly decreases, and energy is quickly released, thereby playing a role in protecting power equipment. The zinc oxide lightning arrester has simple structure, small size and strong current-carrying capacity, and is widely used in power systems. The zinc oxide lightning arrester may appear aging, moisture and other conditions during operation, resulting in increased leakage current, lightning arrester heating, and even explosion in severe cases. Therefore, the operation state of the zinc oxide lightning arrester must be monitored.

[0003] With the popularization of digital intelligent substations, lightning arrester online monitoring technology has gradually been put into use. The principle is consistent with the charged test capacitive compensation method of the lightning arrester, which can be summarized as installing the charged test equipment on the production site, and then transmitting the test data to the monitoring terminal in real time through the communication line, so as to realize uninterrupted tracking of the operating resistive current and total current of the lightning arrester, and provide a basis for the formulation of test and maintenance plans.

[0004] In coastal or air quality poor areas, salt and other contaminants are easily attached to the surface of the MOA porcelain sleeve, reducing the insulation of the porcelain sleeve. In actual production work, the resistive current index of the MOA online monitoring in the coastal area is relatively high. However, after wiping the MOA after power failure, the direct current leakage test result is normal. The reason is that the salt attached to the surface of the MOA porcelain sleeve reduces the insulation resistance of the porcelain sleeve and increases the surface leakage current, but the MOA valve piece itself has not deteriorated. The false alarm of MOA online monitoring caused by surface contamination has become a problem to be solved. SUMMARY

[0005] The present application provides a lightning arrester operation state classification method to realize accurate classification of the operation state of the lightning arrester.

[0006] According to an aspect of the present application, a lightning arrester operation state classification method is provided, which comprises:

[0007] obtaining online monitoring data of the lightning arrester;

[0008] inputting the online monitoring data into a trained classification vector machine to obtain the original classification result of the online monitoring data;

[0009] The original classification result is taken as an observation sequence, an optimal annotation sequence of the observation sequence is obtained by using a conditional random field, and a target classification result of the online monitoring data is determined according to the optimal annotation sequence.

[0010] According to another aspect of the present application, there is provided a lightning arrester operation state classification device, comprising:

[0011] a detection data obtaining module configured to obtain online monitoring data of a lightning arrester;

[0012] an original result obtaining module configured to input the online monitoring data into a trained classification vector machine to obtain an original classification result of the online monitoring data;

[0013] a target result obtaining module configured to take the original classification result as an observation sequence, obtain an optimal annotation sequence of the observation sequence by using a conditional random field, and determine a target classification result of the online monitoring data according to the optimal annotation sequence.

[0014] According to another aspect of the present application, there is provided an electronic device, comprising:

[0015] at least one processor; and

[0016] a memory in communication with the at least one processor; wherein

[0017] the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the lightning arrester operation state classification method according to any one of the embodiments of the present application.

[0018] According to another aspect of the present application, there is provided a computer readable storage medium storing computer instructions for enabling a processor to perform the lightning arrester operation state classification method according to any one of the embodiments of the present application.

[0019] The embodiments of the present application effectively improve the generalization ability of the SVM-CRF model by taking the original classification result generated by the vector machine according to the monitoring data as an observation sequence to perform classification correction by the conditional random field algorithm, and can accurately identify whether the MOA valve piece is deteriorated or surface contaminated, thereby solving the problem that the MOA online monitoring system is prone to false alarm in highly contaminated areas.

[0020] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments description. Obviously, the drawings in the following description only show some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without any creative effort.

[0022] Figure 1A is a flow chart of a method for classifying the operating state of a surge arrester according to an embodiment of the present application;

[0023] Figure 1B is a schematic diagram of a vector machine classification result according to an embodiment of the present application;

[0024] Figure 1C is a schematic diagram of a combined vector machine and conditional random field classification result according to an embodiment of the present application;

[0025] Figure 2A is a flow chart of a method for classifying the operating state of a surge arrester according to another embodiment of the present application;

[0026] Figure 2B is a schematic diagram of a MOA simulation circuit according to another embodiment of the present application;

[0027] Figure 2C is a schematic diagram of a leakage current when a MOA is in normal operation according to another embodiment of the present application;

[0028] Figure 2D is a schematic diagram of a state equation relationship according to another embodiment of the present application;

[0029] Figure 3 is a structural schematic diagram of a device for classifying the operating state of a surge arrester according to another embodiment of the present application;

[0030] Figure 4 is a structural schematic diagram of an electronic device for implementing an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to make the technical personnel in the art better understand the present application, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without any creative effort should be within the scope of protection of the present application.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and in the above drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to only those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0033] Figure 1A A flowchart of a method for classifying the operating state of a surge arrester is provided for an embodiment of the present application. The embodiment can be applicable to the case where, after obtaining the original classification result of the online monitoring data of the surge arrester according to the vector machine, the original classification result is corrected by the conditional random field algorithm to obtain a more accurate target classification result. The method can be executed by a surge arrester operating state classification device, which can be realized in the form of hardware and / or software, and can be configured in an electronic device with corresponding data processing capability. As shown in the figure, the method comprises: Figure 1A

[0034] S110, obtaining online monitoring data of the surge arrester.

[0035] S120, inputting the online monitoring data into the trained classification vector machine to obtain the original classification result of the online monitoring data.

[0036] The online monitoring data at least includes an online data sequence of the total harmonic distortion rate of the surge arrester voltage, an online data sequence of the total current of the surge arrester leakage current, and an online data sequence of the resistive current of the surge arrester leakage current. The vector machine (SVM) is a data classification algorithm with strong generalization ability, which is particularly suitable for classification tasks of small sample data in combination with the kernel trick.

[0037] Specifically, the total current I x , the resistive current I r , and the total harmonic distortion rate I THD of the surge arrester can be obtained through the equivalent circuit simulation model of the surge arrester. x When the valve piece of the surge arrester deteriorates and the surface of the surge arrester is contaminated, different numerical changes will occur in different rules. By analyzing the specific change rules of the three, it can be determined whether the valve piece of the surge arrester deteriorates and the surface of the surge arrester is contaminated, and the specific degree of the deterioration and contamination. The total current I r , the resistive current I THD ​input the trained support vector machine (SVM), and the SVM can determine a state classification result corresponding to the moment according to the input online monitoring data of the moment, and the state classification result output by the SVM is determined as the original classification result.

[0038] Optionally, before the online monitoring data of the lightning arrester is input into the trained support vector machine, the method further includes:

[0039] obtaining at least three state classification labels corresponding to the training set; constructing a binary classification support vector machine for any two state classification labels, and determining a boundary vector between the state classification labels corresponding to the support vector machine according to the input features of the samples in the training set.

[0040] The state classification labels include at least one of the following: normal, uniform degradation, non-uniform degradation, uniform contamination and non-uniform contamination.

[0041] Specifically, training samples D={L, R} are selected from simulation data of an equivalent circuit simulation model, input features L={I r ,I x ,I THD}, and classification labels R={1=normal, 2=uniform degradation, 3=non-uniform degradation, 4=uniform contamination, 5=non-uniform contamination}. Therefore, the task of the SVM is to train a binary classification model between the i-th state classification and the j-th state classification, that is, to solve a quadratic programming problem. There is a boundary (support) vector between any two state classifications. For the SVM with five state classification labels of normal, degradation, non-uniform degradation, uniform contamination and non-uniform contamination, there should be (5*4) / 2=20 boundary vectors. Specifically:

[0042]

[0043] where w ij is the boundary vector between the i-th data and the j-th data, C is a penalty factor, is a relaxation factor, and φ() is a kernel function for mapping x t to a high-dimensional space. The SVM model is trained using the matlab machine learning toolbox and the data set D, and the model accuracy under various kernel functions is shown in Table 1:

[0044] Table 1 SVM kernel function accuracy

[0045]

[0046] S130, the original classification result is used as an observation sequence, and a conditional random field is used to obtain an optimal annotation sequence of the observation sequence, and a target classification result of the online monitoring data is determined according to the optimal annotation sequence.

[0047] Specifically, according to simulation data of the equivalent circuit simulation model, when the MOA is deteriorated or the surface contamination degree is low, the total harmonic distortion rate of the leakage current I x , the resistive current I r , and the leakage current I THD have little difference in the change rule in the two cases, so the input features L of the SVM are similar, and the SVM cannot effectively identify whether it is mild deterioration or mild contamination. Therefore, after the SVM outputs the original classification result, the conditional random field (CRF) algorithm is further introduced. In the conditional random field, the original classification result is taken as an observation sequence X, and the optimal labeling sequence Y corresponding to the observation sequence is obtained through the CRF algorithm. The optimal labeling sequence Y labels the optimal probability of each classification label corresponding to each sequence element in the observation sequence. The optimal classification probability of each sequence element is converted into a specific classification result through an optimal probability selection algorithm such as the Viterbi algorithm, and the target classification result of the online monitoring data is obtained.

[0048] Exemplarily, Figure 1B is a schematic diagram of a vector machine classification result provided by an embodiment of the present application, Figure 1C is a schematic diagram of a combined classification result of a vector machine and a conditional random field provided by an embodiment of the present application. Wherein, Figure 1B It is known that the output error of the SVM is large, and the separation boundary of the sample sequence before and after the two states is difficult to determine, so the specific error value cannot be counted. From the results, the classification result of the SVM in each stage is scattered at a high level, and the consistency is poor, and the result is obviously inconsistent with the common sense. From Figure 1C It is known that after using the CRF algorithm, the highest value of the probability output of the CRF algorithm corresponds to the normal, uniform contamination, non-uniform contamination, and non-uniform deterioration stages, and the classification result after the CRF algorithm is processed is basically consistent with the sample label, and the classification result presents a high continuity and consistency. The difference is that in the process of transition from the third stage to the fourth stage, there is an incorrect classification result. According to the simulation results of the simulation model, the feature vectors of mild uniform deterioration and non-uniform deterioration are very similar, so there is an incorrect classification result. But with the increase of the sequence length, the characteristics of non-uniform deterioration are enhanced, and the final output result is also correct and stable. It can be seen that the introduction and use of the CRF algorithm effectively improve the generalization ability of the SVM-CRF model, and can accurately identify whether the MOA valve piece is deteriorated or surface contaminated, thereby solving the problem that the MOA online monitoring system is prone to false alarm in a highly contaminated area.

[0049] The embodiment of the present application effectively improves the generalization ability of the SVM-CRF model by taking the original classification result generated by the vector machine according to the monitoring data as an observation sequence to correct it through the conditional random field algorithm, and can accurately identify whether the MOA valve piece is deteriorated or surface contaminated, thereby solving the problem that the MOA online monitoring system in a highly contaminated area is prone to false alarm.

[0050] Figure 2A A flowchart of a method for classifying the operating state of a surge arrester is provided in another embodiment of the present application, which is optimized and improved on the basis of the above-mentioned embodiment. As shown in the figure, Figure 2A the method comprises:

[0051] S210, an equivalent circuit simulation model of the surge arrester under operating voltage is established; and a training set for a support vector machine is generated according to the equivalent circuit simulation model of the surge arrester.

[0052] Specifically, taking the Southern Power Grid as an example, the MOA online monitoring system currently applied in the production field basically adopts the capacitive current compensation method, and the principle thereof is to obtain the phase difference between the system voltage and the MOA leakage current, and then separate the resistive component and the capacitive component of the leakage current. At this time, the resistive current I r is an important indicator for judging the operating state of the MOA, and the full current I x can be used as an auxiliary judgment indicator. Due to the nonlinear characteristics of the valve piece in the surge arrester, the current flowing through the valve piece will be distorted, that is, the leakage current has harmonic components, and the leakage current harmonic distortion rate I THD is also an indicator reflecting the operating state of the MOA. Figure 2B is a schematic diagram of a MOA simulation circuit provided in another embodiment of the present application. For simplicity of analysis, the MOA valve piece is divided into upper, middle and lower parts, and the lumped equivalent parameters of each part are shown in Table 2.

[0053] Table 2 Simulation circuit parameter settings

[0054]

[0055] Through the above simulation circuit and parameter settings, the leakage current of the MOA under normal operation is obtained as shown in the figure. Figure 2C From the simulation results, it can be seen that the leakage current effective value of the MOA under normal operation is 0.3328 mA. The leakage current leads the terminal voltage by 87.4°, and is basically a capacitive current component. Due to the existence of the nonlinear resistor, the leakage current waveform is slightly distorted, and the leakage current harmonic total distortion rate is 4.58% through FFT analysis. The above simulation results are consistent with the theoretical analysis results, verifying the correctness of the MOA simulation model in this paper.

[0056] The degradation of MOA valve discs is a slow and irreversible process, and the degree of degradation directly determines their health status. The optimization direction of MOA online monitoring algorithms lies in identifying the degree of valve disc degradation earlier and more accurately. The nonlinear coefficient 'a' of the valve disc during MOA valve disc degradation... i Protection voltage U ref Equivalent capacitance C of the valve plate i All will decrease in size. Simulink simulations will be used to study the impact of MOA degradation on online monitoring data by setting different degradation parameters.

[0057] Assume the surface of the MOA ceramic sleeve is dry and free of dirt. Figure 2B The nonlinear coefficient α of the upper, middle, and lower MOA valve plates in the MOA circuit shown is... i The values ​​are 19, 17, and 15 respectively; the protection voltage U i ref The values ​​are 50kV, 49.5kV, and 49kV respectively; the equivalent capacitance of the valve plate C i The values ​​were 120 pF, 110 pF, and 100 pF, respectively; other parameters were set according to Table 2. Simulations were performed for three states: normal, mild uniform degradation, and severe uniform degradation. The simulation results are shown in Table 3.

[0058] Table 3 Online monitoring values ​​under normal and uniform degradation conditions

[0059]

[0060] Simulation results show that uniform degradation has a relatively low impact on the total current, but a more severe impact on the resistive current and waveform distortion rate. This is why resistive current is a key indicator reflecting the operating status of the MOA.

[0061] Keeping other parameters unchanged, we set the parameters for mild and severe degradation for one part of the upper, middle, and lower MOA respectively, while keeping the other two parts in normal condition, to simulate the case of uneven degradation of MOA. The simulation results are shown in Table 4.

[0062] Table 4. Online monitoring values ​​under non-uniform degradation conditions

[0063]

[0064] Simulation results of non-uniform degradation of the MOA show that degradation in the upper, middle, and lower parts of the MOA has different effects, and the simulation results are quite close to those of mild uniform degradation. Overall, degradation in the upper part has the greatest impact on online monitoring indicators.

[0065] By varying the surface insulation resistance of the MOA (Metal Oxide Aerator) to simulate the degree of contamination of the ceramic bushing, the impact of MOA surface leakage on online monitoring results was studied.

[0066] The porcelain sleeve surface insulation resistance R f1 , R f2 , R f3 are the same value, and other parameters are set according to Table 3, the simulation of the porcelain sleeve surface is uniformly contaminated, and the simulation results are shown in Table 5:

[0067] Table 5 On-line monitoring values under uniform contamination

[0068]

[0069] From the simulation results, it can be seen that the uniform contamination of the MOA porcelain sleeve surface has little effect on the total current, but it can cause the resistive current to increase significantly. This is the reason why the MOA on-line monitoring often alarms in coastal areas. Due to the non-linear characteristics of the valve piece being unaffected, combined with the effect of surface leakage, the leakage current waveform distortion rate shows a slight decrease.

[0070] The porcelain sleeve surface insulation resistance R f1 , R f2 , R f3 are different values, and other parameters are set according to Table 2, the simulation of the porcelain sleeve surface is non-uniformly contaminated, and the simulation results are shown in Table 6:

[0071] Table 6 On-line monitoring values under non-uniform contamination

[0072]

[0073] From the simulation results of non-uniform contamination, it can be seen that the light non-uniform contamination has little effect on each index of on-line monitoring. From the single index, there is no significant distinguishing feature. Only when the non-uniform contamination is serious can the influence be shown. It is worth noting that non-uniform contamination may cause the resistive current to decrease in some cases, even negative, and thus affect the MOA operating state recognition. In addition, non-uniform contamination will cause uneven voltage distribution of the MOA valve piece, accelerating the deterioration of the valve piece.

[0074] S220, obtaining on-line monitoring data of the lightning arrester;

[0075] S230, Kalman filtering on the on-line data sequence of the lightning arrester voltage harmonic total distortion rate to obtain filtered on-line monitoring data.

[0076] Specifically, according to the simulation data of the simulation model, the leakage current waveform distortion rate is an important feature to distinguish the deterioration and contamination of the MOA. However, the factors affecting the leakage current waveform distortion rate of the MOA under operating state not only have its own non-linear characteristics, but also have the interference of power grid voltage harmonics. Kalman filtering (KF) is a system state estimation algorithm for reducing random interference. For the change process of the leakage current waveform distortion rate of the MOA, the Kalman filtering can be used to filter out the random interference of the power grid voltage harmonics, and the filtered on-line monitoring data can be obtained. Figure 2DThe state equation relationship diagram shown represents. According to the state equation relationship diagram, the state equation and the observation equation of the Kalman time-discrete dynamic system can be represented as:

[0077] I THD (t)=I THD (t-1)+w(t-1)

[0078]

[0079] Wherein, I THD (t) is the system estimation value at time t, which represents the leakage current waveform distortion rate caused by the nonlinear characteristics of the MOA in this paper; I MTHD (t) is the observation value at time t, which represents the actual measured leakage current waveform distortion rate; w(t), v(t) are dynamic noise and observation noise at time t, which represent the state change of the MOA and the harmonic interference of the power grid respectively. For this system, the measurement equation can be represented as follows:

[0080]

[0081]

[0082] P(t)=(1-K(t))P - (t)

[0083] P - (t)=P(t-1)+Q(t)

[0084] Wherein, K is the Kalman gain, R is the measurement noise; P is the error covariance; P- is the estimated error covariance; Q is the observation noise, which represents the influence degree of the power grid harmonic on the leakage current, and is defined as:

[0085]

[0086] In the formula, A is a constant, U mTHD is the bus voltage waveform distortion rate. By adjusting the hyperparameters A and R, the influence of the measurement noise v(t) can be reduced, and the optimal estimated value I THD (t) after filtering can be obtained.

[0087] S240, input the online monitoring data into the trained classification vector machine to obtain the original classification result of the online monitoring data;

[0088] S250, respectively acquire the confidence data of the elements in the observation sequence on each state classification label, and the relative change trend between adjacent elements and the boundary vector; according to the confidence data and the relative change trend, determine the optimal labeling sequence of the observation sequence.

[0089] Specifically, the state change of the MOA is very slow relative to the online monitoring data acquisition rate, so it can be assumed that the data before and after monitoring are correlated. According to the CRF theory, the probability distribution of the label sequence Y under the condition of a given observation sequence X can be expressed as:

[0090]

[0091] where λ and μ are corresponding weights, Z(x) is a normalization factor, and is expressed as:

[0092]

[0093] f k is the state feature function of the observation sequence X and the label sequence Y, and the physical meaning is the confidence of each sequence element (the input feature L of the SVM) to each classification label. Let the distance between the input feature L and the support vector hyperplane (w ij ,b ij ) of the SVM model in this paper be r k , then f k can be defined as:

[0094]

[0095] h k is the transition feature function of the two states before and after, and the physical meaning is the relative change trend of the two input features and the support vector, and is defined as:

[0096]

[0097] S260, determining the target classification result of the online monitoring data according to the optimal label sequence.

[0098] The embodiment of the application reduces the influence of power grid harmonics on the leakage current by Kalman filtering of the online monitoring data, and improves the authenticity and effectiveness of the total harmonic distortion rate of the leakage current.

[0099] Figure 3 is a structural schematic diagram of a lightning arrester operating state classification device provided by another embodiment of the application. As shown in Figure 3 , the device comprises:

[0100] a monitoring data acquisition module 310 for acquiring online monitoring data of a lightning arrester;

[0101] an original result acquisition module 320 for inputting the online monitoring data into a trained classification vector machine to obtain an original classification result of the online monitoring data;

[0102] The target result acquisition module 330 is configured to take the original classification result as an observation sequence, acquire an optimal labeling sequence of the observation sequence by using a conditional random field, and determine a target classification result of the online monitoring data according to the optimal labeling sequence.

[0103] The device for classifying the operating state of the lightning arrester provided in the embodiments of the present application can execute the method for classifying the operating state of the lightning arrester provided in any of the embodiments of the present application, and has the corresponding function modules and beneficial effects of the execution method.

[0104] Optionally, the target result acquisition module 330 comprises:

[0105] The labeling data acquisition unit is configured to acquire confidence data of elements in the observation sequence on each state classification label and a relative change trend between adjacent elements and a boundary vector.

[0106] The observation sequence labeling unit is configured to determine an optimal labeling sequence of the observation sequence according to the confidence data and the relative change trend.

[0107] Optionally, the optimal labeling sequence is calculated according to the following formula:

[0108]

[0109] wherein f k is a state feature function of the observation sequence X and the labeling sequence Y, and has a physical meaning of confidence of elements on each state classification label; h k is a transition feature function of two adjacent state classifications, and has a physical meaning of a relative change trend of two adjacent elements and a corresponding boundary vector.

[0110] Optionally, the optimal labeling sequence is calculated according to the following formula:

[0111]

[0112] wherein f k is a state feature function of the observation sequence X and the labeling sequence Y, and has a physical meaning of confidence of elements on each state classification label; h k is a transition feature function of two adjacent state classifications, and has a physical meaning of a relative change trend of two adjacent elements and a corresponding boundary vector.

[0113] Optionally, the state classification label comprises at least one of the following: normal, uniform degradation, non-uniform degradation, uniform contamination, and non-uniform contamination.

[0114] Optionally, the online monitoring data comprises at least a lightning arrester voltage harmonic total distortion rate online data sequence, a lightning arrester leakage current full current online data sequence, and a lightning arrester leakage current resistive current online data sequence, and the device further comprises:

[0115] The monitoring data filtering module is configured to perform Kalman filtering on the arrester voltage harmonic total distortion rate online data sequence to obtain filtered arrester line monitoring data.

[0116] Optionally, the device further comprises:

[0117] The simulation model establishing module is configured to establish an equivalent circuit simulation model of the arrester under operating voltage.

[0118] The training set generating module is configured to generate a training set for the support vector machine according to the equivalent circuit simulation model of the arrester.

[0119] The arrester operating state classification device further described herein can also perform the arrester operating state classification method provided by any embodiment of the present application, and has the corresponding function modules and beneficial effects of performing the method.

[0120] Figure 4 A structural schematic diagram of an electronic device 40 that can be used to implement embodiments of the present application is shown. The electronic device is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices (e.g., headsets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit the implementations of the present application described and / or claimed in this document.

[0121] As shown in Figure 4 The electronic device 40 includes at least one processor 41 and a memory, such as a read-only memory (ROM) 42, a random access memory (RAM) 43, etc., which is communicatively connected to the at least one processor 41, wherein the memory stores a computer program that can be executed by the at least one processor. The processor 41 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 42 or the computer program loaded from the storage unit 48 into the random access memory (RAM) 43. In the RAM 43, various programs and data required for the operation of the electronic device 40 can also be stored. The processor 41, the ROM 42, and the RAM 43 are connected to each other through a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0122] A number of components in the electronic device 40 are connected to the I / O interface 45, including: an input unit 46, such as a keyboard, a mouse, etc.; an output unit 47, such as various types of displays, speakers, etc.; a storage unit 48, such as a magnetic disk, an optical disk, etc.; and a communication unit 49, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 49 allows the electronic device 40 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0123] The processor 41 can be various general and / or special purpose processing components with processing and computing capabilities. Some examples of the processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 41 performs various methods and processes described above, such as the arrester operating condition classification method.

[0124] In some embodiments, the arrester operating condition classification method can be implemented as a computer program tangibly embodied in a computer readable storage medium, such as the storage unit 48. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 40 via the ROM 42 and / or the communication unit 49. When the computer program is loaded onto the RAM 43 and executed by the processor 41, one or more steps of the arrester operating condition classification method described above can be performed. Alternatively, in other embodiments, the processor 41 can be configured to perform the arrester operating condition classification method by any other suitable means, such as by means of firmware.

[0125] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a programmable logic device (PLD), a computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0126] Computer programs for implementing the methods of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the computer program, when executed, can cause instructions defined in the flow charts and / or block diagrams to be implemented. The computer program can be executed entirely on a machine, partially on a machine, partially on a machine as a stand-alone software package, and partially on a remote machine or entirely on a remote machine or server.

[0127] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer-readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium will include one or more lines of a program of instructions in a transitory signal, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0128] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0129] The systems and techniques described herein can be implemented in a computing system that includes a back end component, e.g., as a data server, or that includes a middleware component, e.g., an application server, or that includes a front end component, e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described herein, or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication, e.g., a communication network. Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0130] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. A server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0131] It should be understood that the various forms of flow shown above can be re-ordered, added to, or deleted from without departing from the scope of the present disclosure. For example, the steps recited in the present disclosure can be executed in parallel, executed in series, or executed in different orders, as long as the desired results of the technical solutions of the present disclosure can be achieved, and the present disclosure is not limited herein.

[0132] The above detailed description does not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for classifying the operating status of surge arresters, characterized in that, The method includes: Obtain online monitoring data of surge arresters; The online monitoring data is input into the trained classification vector machine to obtain the original classification result of the online monitoring data; The original classification results are used as the observation sequence. The optimal labeling sequence of the observation sequence is obtained using a conditional random field. The target classification result of the online monitoring data is determined based on the optimal labeling sequence. The step of using the original classification result as an observation sequence and obtaining the optimal labeling sequence of the observation sequence using a conditional random field includes: The confidence scores of elements in the observation sequence for each state classification label and the relative change trends between adjacent elements and the boundary vector are obtained respectively. Based on the confidence data and the relative change trend, the optimal labeled sequence of the observed sequence is determined; wherein, the optimal labeled sequence labels the optimal probability of each sequence element in the observed sequence corresponding to each classification label, and the optimal classification probability of each sequence element is converted into a specific classification result through the optimal probability selection algorithm.

2. The method according to claim 1, characterized in that, The formula for calculating the optimal labeled sequence is as follows: Among them, f k To observe the state feature functions of sequence X and labeled sequence Y, the physical meaning is the confidence level of each element with respect to the classification label of each state; h k The transition feature function is used to classify the two states before and after, and its physical meaning is the relative change trend of the two elements and the corresponding boundary vector.

3. The method according to claim 1, characterized in that, Before inputting the online monitoring data of the surge arrester into the trained support vector machine, the method further includes: Obtain at least three state classification labels corresponding to the training set; Construct a binary support vector machine for any two state classification labels, and determine the boundary vector between the corresponding state classification labels of the support vector machine based on the input features of the samples in the training set.

4. The method according to any one of claims 2-3, characterized in that, The status classification label includes at least one of the following: normal, uniform degradation, non-uniform degradation, uniform dirtiness, and non-uniform dirtiness.

5. The method according to claim 1, characterized in that, The online monitoring data includes at least the online data sequence of total harmonic distortion of the surge arrester voltage, the online data sequence of the total current of the surge arrester leakage current, and the online data sequence of the resistive current of the surge arrester leakage current. After acquiring the online monitoring data of the surge arrester, the process further includes: Kalman filtering was applied to the online data sequence of total harmonic distortion of surge arrester voltage to obtain filtered online monitoring data.

6. The method according to claim 1, characterized in that, Before acquiring the online monitoring data of the surge arrester, the process also includes: Establish an equivalent circuit simulation model of the surge arrester under operating voltage; A training set for a support classification vector machine is generated based on the equivalent circuit simulation model of the surge arrester.

7. A surge arrester operating status classification device, characterized in that, The device includes: The monitoring data acquisition module is used to acquire online monitoring data of the surge arrester; The raw result acquisition module is used to input the online monitoring data into the trained classification vector machine to obtain the raw classification result of the online monitoring data; The target result acquisition module is used to take the original classification result as an observation sequence, use a conditional random field to obtain the optimal labeling sequence of the observation sequence, and determine the target classification result of the online monitoring data based on the optimal labeling sequence. The target result acquisition module includes: The labeled data acquisition unit is used to acquire the confidence data of elements in the observation sequence with respect to each state classification label, and the relative change trend between adjacent elements and the boundary vector. The observation sequence labeling unit is used to determine the optimal labeling sequence of the observation sequence based on the confidence data and the relative change trend; wherein, the optimal labeling sequence labels the optimal probability of each sequence element in the observation sequence corresponding to each classification label, and the optimal classification probability of each sequence element is converted into a specific classification result through the optimal probability selection algorithm.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the surge arrester operating status classification method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are used to cause a processor to execute the surge arrester operation state classification method according to any one of claims 1-6.

Images