Partial discharge monitoring operation and maintenance method, system and device based on intelligent agent

Abstract

The invention relates to the technical field of partial discharge monitoring, in particular to a partial discharge monitoring operation and maintenance method, system and device based on an intelligent agent, and the method comprises the steps: periodically collecting target data reported by at least one edge side device, the target data being determined based on at least one power grid terminal covered by the edge side device, the target data comprises monitoring data and pre-diagnosis data; when the target data is obtained in each period, performing data analysis on the target data by using the data analysis model to obtain a real-time analysis result; for each edge side device, converging a real-time analysis result and target data in one analysis interval to obtain a data group; the data set is analyzed through the data analysis model, a partial discharge diagnosis result is obtained, and the partial discharge diagnosis result comprises the abnormal probability of partial discharge abnormity of the power grid terminal and the partial discharge type. The method can realize intelligent evaluation, diagnosis and risk prediction of the partial discharge state of the power grid terminal, and is timely in diagnosis and high in accuracy.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of partial discharge monitoring, and in particular to a partial discharge monitoring operation and maintenance method, system and device based on an intelligent agent. BACKGROUND

[0002] Operation and maintenance of a power system is a key to ensuring stable and efficient operation of power equipment and related facilities in a long-term operation process. Mainly includes periodic operation monitoring, maintenance and fault repair, etc. to ensure the safe, stable and reliable operation of the power system.

[0003] The general operation and maintenance means is artificial periodic detection, mainly relying on operation and maintenance personnel to regularly go to each site to check and record the equipment one by one.

[0004] However, the manual inspection method has low automation degree and low efficiency, especially for the detection period of partial discharge diagnosis of switch cabinets and other equipment in the distribution room is too long and the accuracy is low, which may lead to failure to discover potential partial discharge faults in time. SUMMARY

[0005] The application embodiment provides a partial discharge monitoring operation and maintenance method, system and device based on an intelligent agent to solve the problem of low efficiency and accuracy of the traditional monitoring method.

[0006] In a first aspect, the application embodiment provides a partial discharge monitoring operation and maintenance method based on an intelligent agent, applied to a cloud intelligent agent, the method comprising: periodically collecting target data reported by at least one edge side device, the target data being determined based on at least one power grid terminal covered by the edge side device, the target data comprising monitoring data and pre-diagnosis data, the pre-diagnosis data being generated when a partial discharge signal and / or a discharge interference signal is identified in the monitoring data; when the target data is obtained in each cycle, using a data analysis model to analyze the target data to obtain real-time analysis results; for each edge side device, gathering real-time analysis results and target data in an analysis interval to obtain a data group; wherein the analysis interval refers to an interval corresponding to the Mth collection cycle to the M+Nth collection cycle, M and N are preset values; using the data analysis model to analyze the data group to obtain a partial discharge diagnosis result, the partial discharge diagnosis result comprising an abnormal probability of a partial discharge anomaly existing in the power grid terminal and a partial discharge type.

[0007] In a possible implementation, the monitoring data includes partial discharge monitoring data; the step of performing data analysis on the target data by using the data analysis model to obtain the real-time analysis result includes: performing atlas mapping on the partial discharge monitoring data to obtain an analysis atlas; the analysis atlas includes a partial discharge pulse signal atlas; performing waveform analysis on the partial discharge pulse signal atlas by using an atlas classification and recognition model to determine waveform features of the analysis atlas; the waveform features include one or more of a waveform type, peak information, and valley information, and the waveform type includes a single peak, a double peak, a multi-peak, and a stable peak; and the real-time analysis result includes the waveform features.

[0008] In a possible implementation, the analysis atlas further includes a partial discharge phase analysis atlas, and the step of performing atlas mapping on the partial discharge monitoring data to obtain the analysis atlas further includes: performing feature calculation on the partial discharge pulse signal atlas by using a mechanism model to obtain analysis features, the analysis features including amplitude dispersion features, pulse balance degree features, polarity features, phase aggregation features, 50Hz component significance features, 100Hz component significance features, and 50 / 100Hz component numerical relationship features; performing feature calculation on the partial discharge phase analysis atlas by using the mechanism model to obtain flight map features; and the real-time analysis result further includes the analysis features and the flight map features.

[0009] In a possible implementation, the pre-diagnosis data includes partial discharge alarm information; and the step of performing analysis on the data group by using the data analysis model to obtain the partial discharge diagnosis result includes: determining, for each edge side device, whether there is a discharge interference signal in a power grid terminal covered by the edge side device based on the data group; if there is no discharge interference signal, determining that the partial discharge type is normal, and determining that the abnormal probability is equal to 0; if there is a discharge interference signal, determining whether the number of partial discharge alarm information in the data group is greater than a preset threshold; and if the number of partial discharge alarm information in the data group is not greater than the preset threshold, determining that the partial discharge type is normal, and determining that the abnormal probability is equal to 0.

[0010] In a possible implementation, the pre-diagnosis data includes partial discharge alarm information; the partial discharge types include a target partial discharge type and a particle discharge type; the step of performing data analysis on the data set by using the data analysis model to obtain the partial discharge diagnosis result further includes: determining whether the number of partial discharge alarm information in the data set is greater than a preset threshold; if the number of partial discharge alarm information in the data set is greater than the preset threshold, obtaining a rule table from the database; the rule table includes signal feature information corresponding to different target partial discharge types; the target partial discharge types include but are not limited to air gap discharge type, surface discharge type, suspension discharge type, and corona discharge type; according to the rule table, determining the matching degree of the waveform feature and the analysis feature of the analysis graph corresponding to each data set with the signal feature information of each target partial discharge type in the rule table, and determining the feature probability of each waveform feature and each analysis feature based on the determination result; wherein for each target partial discharge type, if the waveform feature or the analysis feature of the analysis graph corresponding to the data set meets the corresponding signal feature information, the feature probability corresponding to the waveform feature or the analysis feature is 1, and if the waveform feature or the analysis feature does not meet the signal feature information, the feature probability corresponding to the waveform feature or the analysis feature is 0; based on a first formula, calculating the type probability P of each data set corresponding to each target partial discharge type, the first formula is: ; wherein Xi represents the feature probability corresponding to the waveform feature or the analysis feature in the data set, q is equal to the total number of waveform features and analysis features in the data set, and 0 < i ≤ q; if the waveform feature and the analysis feature of the analysis graph corresponding to the data set do not match the signal feature information corresponding to each target partial discharge type, determining whether the flight graph feature presents a significant feature according to the flight graph feature similarity; wherein the flight graph feature presenting a significant feature corresponds to a different flight graph feature similarity from presenting a non-significant feature; if the flight graph feature presents a significant feature, the flight graph feature similarity corresponding to the significant feature is determined as the type probability corresponding to the particle discharge type; based on the type probability of each data set corresponding to each partial discharge type, the partial discharge diagnosis result is obtained.

[0011] In a possible implementation, based on the type probability of each data set corresponding to each type of partial discharge, the step of obtaining the partial discharge diagnosis result comprises: for each data set, determining the maximum probability value in the type probability of the data set corresponding to each type of partial discharge, and counting the number of type probabilities equal to the maximum probability value; if the number of type probabilities equal to the maximum probability value is one, determining that the partial discharge type corresponding to the data set is the target partial discharge type or the particle discharge type corresponding to the maximum probability value, and determining that the abnormal probability is equal to the maximum probability value, to obtain the partial discharge diagnosis result of the edge side device corresponding to the data set; if the number of type probabilities equal to the maximum probability value is multiple, determining that the partial discharge type corresponding to the data set is a random one of the target partial discharge type or the particle discharge type corresponding to the maximum probability value, and determining that the abnormal probability is equal to the maximum probability value, to obtain the partial discharge diagnosis result of the edge side device corresponding to the data set.

[0012] In a possible implementation, the monitoring data further comprises environmental monitoring data; the method further comprises: aggregating the partial discharge type and the abnormal probability corresponding to each edge side device to form a monitoring list; and / or, for each edge side device, generating a treatment suggestion based on the partial discharge type, the abnormal probability, the partial discharge monitoring data, the environmental monitoring data and / or the account data of the power grid terminal, the account data being data recording basic information, operating state and / or maintenance history of the power grid terminal; and / or, obtaining a decision result for the partial discharge diagnosis result to form positive samples and negative samples; the positive samples correspond to the decision result being consistent with the partial discharge diagnosis result, and the negative samples correspond to the decision result being inconsistent with the partial discharge diagnosis result; and training a data analysis model by using the positive samples and the negative samples.

[0013] In a possible implementation, the method further comprises: in response to a text question raised by a user through a web page, calling a text large model to analyze the question and generate an answer; wherein the text large model is trained based on a preset training document, and the preset training document comprises partial discharge monitoring data, environmental monitoring data, account data of the power grid terminal and / or partial discharge knowledge; and displaying the answer on the web page.

[0014] In a second aspect, the embodiments of the present application also provide an agent-based partial discharge monitoring operation and maintenance system, which comprises: an edge-side device and a cloud agent; the edge-side device is configured to: generate target data and report the target data to the cloud agent; the target data is determined based on at least one power grid terminal covered by the edge-side device, and the target data comprises monitoring data and pre-diagnosis data, the pre-diagnosis data being generated when a partial discharge signal and / or a discharge interference signal is identified in the monitoring data; the cloud agent is configured to: periodically collect the target data reported by at least one edge-side device; when the target data is obtained in each period, perform data analysis on the target data by using a data analysis model to obtain real-time analysis results; for each edge-side device, aggregate the real-time analysis results and the target data in one analysis interval to obtain a data group; wherein one analysis interval refers to an interval corresponding to the Mth collection period to the M+Nth collection period, and M and N are preset values; analyze the data group by using the data analysis model to obtain a partial discharge diagnosis result, the partial discharge diagnosis result comprising an abnormal probability of a partial discharge anomaly existing in the power grid terminal and a partial discharge type.

[0015] In a third aspect, the embodiments of the present application also provide an agent-based partial discharge monitoring operation and maintenance device, which is applied to a cloud agent and comprises: a collection module configured to: periodically collect target data reported by at least one edge-side device, the target data being determined based on at least one power grid terminal covered by the edge-side device, and the target data comprising monitoring data and pre-diagnosis data, the pre-diagnosis data being generated when a partial discharge signal and / or a discharge interference signal is identified in the monitoring data; a first analysis module configured to: when the target data is obtained in each period, perform data analysis on the target data by using a data analysis model to obtain real-time analysis results; an aggregation module configured to: for each edge-side device, aggregate the real-time analysis results and the target data in one analysis interval to obtain a data group; wherein one analysis interval refers to an interval corresponding to the Mth collection period to the M+Nth collection period, and M and N are preset values; a second analysis module configured to: analyze the data group by using the data analysis model to obtain a partial discharge diagnosis result, the partial discharge diagnosis result comprising an abnormal probability of a partial discharge anomaly existing in the power grid terminal and a partial discharge type.

[0016] From the above, the embodiment of the application provides a method, system and device for monitoring and operating partial discharge based on an intelligent agent. The method comprises: periodically collecting target data reported by at least one edge-side device, the target data being determined based on at least one power grid terminal covered by the edge-side device, and the target data comprising monitoring data and pre-diagnosis data; when the target data is obtained in each period, performing data analysis on the target data by using a data analysis model to obtain real-time analysis results; for each edge-side device, gathering the real-time analysis results and the target data in an analysis interval to obtain a data group; and performing analysis on the data group by using the data analysis model to obtain a partial discharge diagnosis result, the partial discharge diagnosis result comprising an abnormal probability of a partial discharge anomaly existing in the power grid terminal and a partial discharge type. The method can realize intelligent evaluation, diagnosis and risk prediction of a partial discharge state of a power grid terminal, and the diagnosis is timely and accurate. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The first flowchart of the method for monitoring and operating partial discharge based on an intelligent agent is provided in the embodiment of the application. Figure 2 The second flowchart of the method for monitoring and operating partial discharge based on an intelligent agent is provided in the embodiment of the application. Figure 3 The flowchart of the pre-diagnosis step is provided in the embodiment of the application. Figure 4 The flowchart of the intelligent question-answering is provided in the embodiment of the application. Figure 5 The first structural diagram of the system for monitoring and operating partial discharge based on an intelligent agent is provided in the embodiment of the application. Figure 6 The second structural diagram of the system for monitoring and operating partial discharge based on an intelligent agent is provided in the embodiment of the application. Figure 7 The structural diagram of the device for monitoring and operating partial discharge based on an intelligent agent is provided in the embodiment of the application. DETAILED DESCRIPTION

[0018] In order for those skilled in the art to better understand the technical solutions in the application, the technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the application.

[0019] Before introducing the technical solutions in the embodiments of the application, the terms involved in the embodiments of the application are first described by way of example.

[0020] 1. Agent: An entity that can autonomously perceive the environment, process information, perform tasks, and interact with the outside world.

[0021] 2. Power distribution station / room: A physical building that houses transformers (optional), switchgear, and other power equipment, as well as incoming and outgoing cables. It is equipped with lighting, fire protection, and drainage facilities. The power distribution station / room connects to the substation above and the user-side distribution room below. It primarily performs two functions: The first is medium-voltage to low-voltage power conversion, and the second is the organization of the distribution network through busbars combined with switchgear, supporting flexible distribution of electrical energy. Power distribution stations are primarily divided into two categories and four subcategories: Switching stations and switches (indoor ones are called demarcation rooms, outdoor ones are called ring network cabinets): they realize distribution network organization and load distribution, generally do not perform power transformation and are not directly used by end users.

[0022] Distribution room and box-type transformer: participate in the distribution network organization and realize ring network power supply; realize 10kV / 400V power transformation and provide power for loads.

[0023] 3. Switchgear: A device used in power systems for the distribution, control, protection, and monitoring of electrical energy. A switchgear contains a range of electrical equipment, such as circuit breakers, disconnectors, load switches, grounding switches, transformers, protective relays, and control devices.

[0024] 4. Partial Discharge (PD): PD is a brief electrical discharge that occurs in a localized area within or on the surface of the insulation structure of electrical equipment. It is a sign of insulation degradation. PD does not immediately cause insulation failure, but it can worsen the degradation and, if left untreated, ultimately lead to insulation breakdown and accidents.

[0025] Power system operation and maintenance is key to ensuring the long-term stable and efficient operation of power equipment and related facilities. This includes regular operation monitoring, maintenance, and troubleshooting to ensure the safe, stable, and reliable operation of the power system.

[0026] The general operation and maintenance method is manual periodic inspection, which mainly relies on operation and maintenance personnel to regularly visit each site to check and record the equipment one by one.

[0027] However, manual inspection methods have a low degree of automation and low efficiency. In particular, the detection cycle of partial discharge diagnosis of equipment such as switch cabinets in distribution rooms is too long and the accuracy is low, which easily leads to failure to detect potential partial discharge faults in a timely manner.

[0028] The cause of partial discharge is usually internal defects in the insulation structure or changes in external factors that lead to excessive local electric fields and cause discharge.

[0029] Internal causes: internal or surface insulation gap, crack, bubble, dirt, density unevenness.

[0030] External causes: mechanical vibration / frequent switching leading to poor conductor contact, excessive temperature leading to reduced insulation performance, excessive humidity leading to condensation on the insulation surface, excessive voltage and current leading to excessive local electric field, unreasonable structure.

[0031] The embodiment of the present application provides a partial discharge monitoring operation and maintenance method, system and device, which can realize online monitoring of power grid terminals, early monitoring and early warning of problems, and reduction of work pressure of front-line live detection; meanwhile, equipment abnormalities can be found in the first time, the contingency and hysteresis of periodic live detection are avoided, and the safety operation risk of equipment is reduced.

[0032] Figure 1 The first flowchart of the partial discharge monitoring operation and maintenance method based on an intelligent agent provided by the embodiment of the present application.

[0033] Figure 2 The second flowchart of the partial discharge monitoring operation and maintenance method based on an intelligent agent provided by the embodiment of the present application.

[0034] As shown in Figure 1 and Figure 2 The partial discharge monitoring operation and maintenance method based on an intelligent agent provided by the embodiment of the present application can be applied to a cloud intelligent agent, and specifically can include the following steps S100-S400.

[0035] S100: periodically collecting target data reported by at least one edge-side device, the target data being determined based on at least one power grid terminal covered by the edge-side device, and the target data including monitoring data and pre-diagnosis data, the pre-diagnosis data being generated when a partial discharge signal and / or a discharge interference signal is identified in the monitoring data.

[0036] The power grid terminal can be a switch cabinet in a power distribution room. The monitoring data can be obtained by periodically collecting the switch cabinet and other power grid terminals by a sensing terminal. The monitoring data can include partial discharge monitoring data and environmental monitoring data. Further, the edge-side device can be a station-side node, and one station-side node can be used to manage one power distribution room and related sensing terminals.

[0037] In some implementations, a plurality of switch cabinets can be arranged in the power distribution room, and the actual number of switch cabinets and the number of power distribution rooms depend on the actual power system arrangement, and the embodiments of the present application do not make specific limitations. The sensing terminal can include a partial discharge sensor and a temperature and humidity sensor, and each power distribution room is provided with a temperature and humidity sensor to collect environmental monitoring data of the power distribution room, such as temperature and humidity. In addition, each switch cabinet in each power distribution room is provided with a partial discharge sensor to sample partial discharge signals or discharge interference signals.

[0038] In some implementations, the sensing terminal can specifically collect data of the switch cabinet of the power distribution room, and can also collect data of the incoming line cabinet, the feeder cabinet, the bus coupler cabinet, the bus coupler isolation cabinet, the potential transformer (PT) cabinet, and the in-station transformer (in-station transformer).

[0039] It should be further pointed out that the pre-diagnosis is an edge-level calculation, which can be implemented by an edge-side device, and the steps of the pre-diagnosis can include positioning and excluding external interference, generating station-side partial discharge and interference alarms, and specific steps will be described in detail below, which will not be repeated here. In this way, the computing pressure of the cloud can be reduced.

[0040] Further, in the embodiments of the present application, the frequency of periodic collection can be preset, for example, 2h / time or 4h / time, and the embodiments of the present application do not make specific limitations.

[0041] In some implementations, the edge-side device and the cloud intelligent agent can communicate through wireless local area network authentication and privacy infrastructure (WLAN Authentication and Privacy Infrastructure, wapi), operator 5G or wired Ethernet network communication mode.

[0042] S200: When the target data is obtained in each cycle, the data analysis model is used to analyze the target data to obtain real-time analysis results.

[0043] The embodiments of the present application can perform real-time partial discharge analysis based on pre-diagnosis data and monitoring data. The data analysis model, also known as a partial discharge diagnosis model, can include an AI atlas classification model (also known as an atlas classification and recognition model), a mechanism feature analysis model (also known as a mechanism model), etc. The AI atlas classification model can be a CNN classification model (also known as a partial discharge atlas CNN classification model). It can be understood that step S200 is a single sampling analysis calculation step, which can include AI atlas classification and recognition and mechanism model feature calculation.

[0044] In this way, the target data reported by each edge-side device can be analyzed in real time to diagnose the partial discharge of the power grid terminal covered by the edge-side device and discover the partial discharge fault in time.

[0045] S300: For each edge-side device, the real-time analysis results and the target data in an analysis interval are aggregated to obtain a data group.

[0046] The analysis interval refers to an interval corresponding to the Mth collection period to the M+Nth collection period, and M and N are preset values. For example, the analysis interval can refer to an interval corresponding to the 1st collection period to the 14th collection period, where M is equal to 1 and N is equal to 13. In this way, the time collected in a period of time can be aggregated for further partial discharge diagnosis. It can be understood that step S300 can include AI atlas classification result aggregation and mechanism model partial discharge type analysis result aggregation.

[0047] S400: Analyzing the data group by using a data analysis model to obtain a partial discharge diagnosis result, the partial discharge diagnosis result including an abnormal probability and a partial discharge type of the power grid terminal existing partial discharge abnormality.

[0048] The partial discharge diagnosis result can include a partial discharge type (PD Type) and an abnormal probability of the power grid terminal existing partial discharge abnormality. For example, the partial discharge diagnosis result includes a partial discharge type and an abnormal probability of the switch cabinet in the power distribution room existing partial discharge abnormality.

[0049] Further, different defects result in different partial discharge types, and the partial discharge type can include a target partial discharge type and a particle discharge type. The target partial discharge type can include an air gap discharge type, a surface discharge type, a suspension discharge type, and a corona discharge type. The abnormal probability is obtained based on the data analysis model, and the specific analysis steps will be described below, which will not be repeated here.

[0050] In some implementations, step S400 can also perform partial discharge diagnosis based on historical collection data of the power grid terminal. The historical collection data can refer to data collected the day before or the days before for the same power grid terminal.

[0051] In this way, through the cooperation of cloud-edge-end, real-time monitoring of partial discharge abnormality of the power grid terminal (such as the switch cabinet in the power distribution room) can be realized, the abnormality identification is accurate and timely, and manpower is saved, which can improve the stability of equipment operation and ensure the safe operation of the power grid terminal.

[0052] From the above, the embodiment of the application provides a method for diagnosing partial discharge based on an intelligent agent, which comprises the following steps: periodically collecting target data reported by at least one edge-side device, the target data being determined based on at least one power grid terminal covered by the edge-side device, the target data comprising monitoring data and pre-diagnosis data, the pre-diagnosis data being generated when a partial discharge signal and / or a discharge interference signal is identified in the monitoring data; when the target data is obtained in each period, performing data analysis on the target data by using a data analysis model to obtain real-time analysis results; for each edge-side device, converging the real-time analysis results and the target data in an analysis interval to obtain a data group; wherein the analysis interval refers to an interval corresponding to the Mth collection period to the M+Nth collection period, and M and N are preset values; performing analysis on the data group by using the data analysis model to obtain a partial discharge diagnosis result, the partial discharge diagnosis result comprising an abnormal probability of the power grid terminal having a partial discharge abnormality and a partial discharge type. The method can realize intelligent evaluation, diagnosis and risk prediction of the partial discharge state of the power grid terminal, realize the transition from "diagnosing problems relying on personal experience" to "intelligent diagnosis based on model reasoning", and achieve less false positives, fast diagnosis and high result reliability of the partial discharge problem.

[0053] It should be further pointed out that in the traditional partial discharge detection and diagnosis mode, the user mainly relies on a single on-site live detection to diagnose whether there is partial discharge, and the single detection result is easily affected by incidental factors, resulting in a judgment error. In addition, the diagnosis of the partial discharge relies on personal experience, and may also be affected by personal skills and personnel changes, resulting in a judgment error. In order to avoid the judgment error, the user must continuously perform on-site retesting, which is low in efficiency.

[0054] After the method and system based on the intelligent agent provided by the application are put into use, firstly, the comprehensive diagnosis of multiple partial discharge monitoring results can be realized based on historical online monitoring data, the influence of incidental factors is avoided, and the reliability of diagnosis is improved; secondly, the diagnosis is changed from pure artificial diagnosis to system intelligent diagnosis based on the combination of mechanism model and AI model, the limitations of personal ability and experience are avoided, and the accuracy of diagnosis is improved; and the general partial discharge data is analyzed and diagnosed in association with temperature and humidity, load and account data, and the accuracy of the disposal measures is improved. Thirdly, relying on continuous online monitoring and diagnosis, the workload of on-site retesting is reduced, and the work efficiency of partial discharge diagnosis is improved.

[0055] The steps of pre-diagnosis of the edge-side device are described in detail below.

[0056] Figure 3 A flowchart of the pre-diagnosis steps provided by the embodiment of the application is shown.

[0057] As shown in Figures 1 to 3 The method for diagnosing partial discharge based on the intelligent agent provided by the embodiment of the application can further comprise the following steps S501-S504.

[0058] S501: receiving the collection data reported by the perception terminal, performing time alignment based on the time stamp carried in the collection data to obtain all collection data in the same time window, and the same time window corresponds to one collection period.

[0059] It can be understood that when the signal sampling is performed and the collection data is generated, the perception terminal can add a time stamp in the collection data, and the time stamp is used to represent a specific sampling time. Since the perception terminal is periodically sampled, all collection data of one power distribution station in one collection period can be determined based on the time stamp. The key of time alignment is to unify these different collection data to one common time standard, so that all data can be ensured to be in the same time window, time synchronization is achieved, and data analysis is facilitated.

[0060] S502: identifying the collection data in the same time window, judging whether it includes partial discharge monitoring data of all power grid terminals covered by the edge side device, and / or judging whether there is an abnormal value.

[0061] It can be understood that in this step, the data in the current time window can be checked, and first, it is judged whether it contains partial discharge monitoring data of all power grid terminals to be monitored. The partial discharge monitoring data is one of the key data for evaluating whether the power grid terminal has a potential fault. If the data lacks the monitoring results of some devices, the application embodiment can mark these data missing parts. In addition, the application embodiment can also perform abnormal value detection on the partial discharge monitoring data. For example, some values in the partial discharge monitoring data may deviate due to sensor failure, communication problems or other interference factors. At this time, the application embodiment can identify these abnormal values, which is convenient for subsequent correction.

[0062] S503: if the partial discharge monitoring data of all power grid terminals covered by the edge side device is not included, supplementing the missing partial discharge monitoring data, and / or if the abnormal value is included, correcting the abnormal value.

[0063] In the application embodiment, if the partial discharge monitoring data of some power grid terminals (such as switch cabinets) is missing, the numerical supplement calculation can be performed according to the data of other power grid terminals or the historical partial discharge monitoring data of the power grid terminal. In addition, it can also be realized in various ways, such as model-based prediction, monitoring data calculation of adjacent devices, etc. The purpose of supplement is to ensure that the partial discharge monitoring data of all power grid terminals can be completely reflected in the current time window.

[0064] For the data marked as abnormal, the embodiments of the present application can perform abnormal correction. For example, the incorrect data is replaced by a reasonable value through data filtering, calibration algorithm and other methods, so that the corrected data will more accurately reflect the real state of the device.

[0065] S504: Pre-diagnosis is performed on the supplemented and / or corrected partial discharge monitoring data in the same time window by using an ultrasonic algorithm, and when a target signal is identified, pre-diagnosis data is generated.

[0066] The target signal includes a partial discharge signal and / or a cabinet external discharge interference signal, and the pre-diagnosis data includes partial discharge alarm information and / or interference alarm information. The partial discharge alarm information at least includes the coordinates corresponding to the generation of the partial discharge signal, and the interference alarm information at least includes the coordinates corresponding to the generation of the partial discharge interference signal. In the embodiments of the present application, the "coordinates" can refer to the accurate positioning information of the partial discharge signal or the partial discharge interference signal in three-dimensional space, which can be the specific physical position of the power grid terminal (such as a switch cabinet or other equipment) where the partial discharge signal or the partial discharge interference signal appears, or the specific physical position of the partial discharge sensor that captures the partial discharge signal or the partial discharge interference signal. The "coordinates" can include X, Y, and Z three-dimensional coordinate values.

[0067] In some implementations, the "coordinates" can refer to the accurate positioning information of the partial discharge signal or the partial discharge interference signal in the time dimension, specifically the time stamp of capturing the partial discharge signal or the partial discharge interference signal.

[0068] In the embodiments of the present application, the partial discharge monitoring data is formed by a partial discharge monitoring signal. The partial discharge monitoring signal can be a pulse, for example, which is formed by sampling the superimposed signal of the partial discharge signal and / or the cabinet external interference signal in the collection period. Therefore, when the partial discharge signal and / or the cabinet external interference signal is monitored in the partial discharge monitoring signal, it can be determined that the switch cabinet in the power distribution station room has an abnormality, and the purpose of external interference marking and alarm reporting and data reporting is achieved.

[0069] In some implementations, when the partial discharge monitoring signal exhibits the characteristics of the partial discharge signal or exhibits the characteristics of the interference signal as a pulse, it can be determined that the target signal is identified in the partial discharge monitoring signal.

[0070] Further, if the target signal is detected, the embodiments of the present application can generate partial discharge alarm information according to the characteristics of the signal. The partial discharge alarm information can include the source position (such as the device coordinates of the specific power grid terminal) and the occurrence time of the signal, thereby helping the maintenance personnel to locate the faulty equipment.

[0071] Further, the ultrasonic algorithm can also distinguish partial discharge signals from other possible interference signals. Interference signals outside the cabinet can be caused by external environmental factors (such as electrical noise, wind noise, etc.). The ultrasonic algorithm can identify whether there is a partial discharge signal or an interference signal by analyzing the frequency and amplitude difference of the signal. Partial discharge signals usually have specific pulse characteristics, and interference signals are usually continuous noise. The embodiments of the present application can identify these interference signals and generate interference alarm information, and record the coordinates of the partial discharge interference signals for further analysis and elimination of false positives.

[0072] The steps of cloud agent data analysis are described in detail below.

[0073] In the embodiments of the present application, the cloud agent can call the data analysis model to analyze and calculate based on the 9 mechanism characteristics of the partial discharge monitoring data to obtain real-time analysis results. The 9 mechanism characteristics are as follows: ①Amplitude dispersion: analyze the signal amplitude range on each phase, judge its dispersion, and judge its characteristic compliance with different partial discharge types, for example, the amplitude dispersion of suspended discharge is small.

[0074] ②Pulse balance degree: analyze and count the pulse frequency (satisfying a certain amplitude intensity) in each power frequency cycle of the sampling data to obtain a sequence of 50 pulse frequencies of 1 sampling; based on the pulse count of the sequence, analyze its balance degree.

[0075] ③Phase spectrum waveform feature analysis: analyze the peak-valley morphology of the phase spectrum, analyze the single-peak, double-peak, and multi-peak characteristics, and judge their characteristic compliance with different partial discharge types, for example, suspended discharge presents double-peak characteristics.

[0076] ④Polarity characteristic: analyze and calculate whether the peak-valley morphology of the phase spectrum conforms to the polarity effect.

[0077] ⑤Phase aggregation characteristic: analyze and calculate the pulse aggregation in the phase spectrum, whether the signal is concentrated in a small number of phases or is very dispersed, for example, the aggregation degree of suspended discharge is high.

[0078] ⑥50Hz component significance: evaluate the 50Hz component significance degree by comparing the 50Hz frequency component of the ultrasonic signal with the signal peak amplitude.

[0079] ⑦100Hz component significance: evaluate the 100Hz component significance degree by comparing the 100Hz frequency component of the ultrasonic signal with the signal peak amplitude.

[0080] ⑧50 / 100Hz component numerical relationship (or 50-100Hz numerical relationship characteristic): analyze the data size relationship between the two, and judge their characteristic compliance with different discharge types, for example, the 50Hz component of suspended discharge is obviously smaller than the 100Hz component.

[0081] 9. Flight pattern characteristics: used to characterize particle discharge.

[0082] Further, the real-time analysis result output by the cloud agent can include waveform characteristics, analysis characteristics, and / or flight pattern characteristics. The waveform characteristics can be single-peak, double-peak, multi-peak, or flat peak, which can be determined based on phase spectrum waveform characteristic analysis. The analysis characteristics can include amplitude dispersion characteristics, pulse balance degree characteristics, polarity characteristics, phase aggregation characteristics, 50 Hz component significance characteristics, 100 Hz component significance characteristics, and 50 / 100 Hz component numerical relationship characteristics, a total of 7 kinds.

[0083] As shown in FIG. 2, step S200 can include steps S201-S202. Figures 1 to 3

[0084] S201: Perform spectrum mapping on the partial discharge monitoring data to obtain an analysis spectrum.

[0085] The analysis spectrum can include a partial discharge pulse signal (PRPS) spectrum and a partial discharge phase analysis (PRPD) spectrum. The PRPS spectrum presents the amplitude, phase, and time period signal of the partial discharge monitoring signal in a three-dimensional graph. The PRPD spectrum generally combines 50 power frequency cycle sampling data according to 0-360° phase, counts the same amplitude and phase, and presents in a scatter plot (different counts correspond to different colors). In actual application, the spectrum can be mapped based on the sampling data (sampling amplitude) of the PRPS data in a complete electrical cycle (360°). Generally, the sampling amplitude in a complete electrical cycle is generally 3600, that is, the sampling points are 3600. 3600 sampling points can achieve at least one sampling point per 0.1° phase, meeting the accuracy requirements of the analysis spectrum mapping.

[0086] S202: Perform waveform analysis on the partial discharge pulse signal spectrum using a spectrum classification and recognition model to determine the waveform characteristics of the analysis spectrum. The waveform characteristics include one or more of waveform type, peak information, and valley information. The waveform type includes single-peak, double-peak, multi-peak, and flat peak.

[0087] This step is a step of calculating the phase spectrum waveform characteristics. The peak information and the valley information can specifically include the effective peak and its left and right adjacent valleys, the start phase, the end phase, and the peak value, which are not limited by the embodiments of the present application.

[0088] ​In the embodiment of the present application, the atlas classification and identification model can be obtained by optimizing a basic model based on an AI classification model architecture (such as SVM, CNN, etc.) for the partial discharge scene. In the training stage, first, the PRPS data containing single-peak, double-peak and other waveforms are preprocessed, and the preprocessing steps include: data denoising, amplitude standardization, normalization, labeling waveform type and / or peak / trough feature (such as phase range, peak value, etc.). Then, the preprocessed PRPS data is used as training data to supervise the learning and training of the basic model, and finally a model capable of accurately identifying waveform features is obtained. Further, step S201 can further include steps S203-S204.

[0089] S203: using a mechanism model to calculate the features of the partial discharge pulse signal atlas, to obtain analysis features.

[0090] S204: using a mechanism model to calculate the features of the partial discharge phase analysis atlas, to obtain flight map features.

[0091] It is worth noting that the embodiment of the present application can divide each feature into features, for example, the amplitude dispersion feature can be divided into three types of high, medium and low, the pulse balance degree can also be divided into three types of high, medium and low, the phase atlas waveform feature can be divided into single-peak, double-peak, multi-peak and stable peak, the polarity feature can be divided into three types of not obvious, relatively obvious and obvious, the phase aggregation can be divided into three types of high, medium and low, the 50Hz component can be divided into three types of high, medium and low, the 100Hz component can be divided into three types of high, medium and low, and the 50 / 100Hz component numerical relationship feature can include 50Hz component >> 100Hz component or 100Hz component >> 50Hz component, wherein ">>" is a much greater than symbol. The flight map feature includes similarity representation and flight map feature similarity, and the similarity representation at least includes presenting a significant feature and presenting a non-significant feature. The flight map feature presenting a significant feature and presenting a non-significant feature correspond to different flight map feature similarities. The flight map feature presenting a significant feature can represent the existence of particle discharge, and presenting a non-significant feature can represent the non-existence of particle discharge. Further, step S400 can include steps S401-S404.

[0092] S401: for each edge side device, based on the data set, determining whether there is a discharge interference signal in the power grid terminal covered by the edge side device.

[0093] In the embodiment of the present application, the step of determining whether there is an interference signal based on the data set can specifically be determining whether the pre-diagnosis data includes partial discharge alarm information for the interference signal.

[0094] S402: if there is no discharge interference signal, determining that the partial discharge type is abnormal, and determining that the abnormal probability is equal to 0.

[0095] In this way, it can be determined that there is no partial discharge anomaly in the power grid terminal covered by the edge side device.

[0096] In the embodiments of the present application, if it is determined that there is no partial discharge anomaly, the routine monitoring can be resumed, and step S100 is continued to execute.

[0097] S403: If there is a discharge interference signal, it is judged whether the number of partial discharge alarm information in the data group is greater than a preset threshold.

[0098] The preset threshold is, for example, equal to 1 or 2, and the embodiments of the present application do not make specific limitations on this.

[0099] S404: If the number of partial discharge alarm information in the data group is not greater than the preset threshold, it is determined that the partial discharge type is no anomaly, and the anomaly probability is equal to 0.

[0100] In this way, it can be determined that there is no partial discharge anomaly in the power grid terminal covered by the edge side device.

[0101] Further, step S400 can further include steps S405-S407.

[0102] S405: If the number of partial discharge alarm information in the data group is greater than the preset threshold, a rule table is obtained from the database; the rule table includes signal feature information corresponding to different target partial discharge types; the target partial discharge type includes but is not limited to air gap discharge type, surface discharge type, floating discharge type and / or corona discharge type.

[0103] For example, the rule table is shown in Table 1.

[0104] Table 1 Rule table

[0105] The rows of the table represent each mechanism feature, and the columns of the table represent each partial discharge type. The signal feature information corresponding to each partial discharge type is shown in the cell composed of a row and a column.

[0106] In some implementations, the rule table can be saved in the database to quickly modify the rules. The content of the database is read each time the diagnosis is performed, and the probability calculation is performed according to the latest rules.

[0107] S406: According to the rule table, it is judged that the waveform feature and the analysis feature of the analysis spectrum corresponding to each data group match the signal feature information of each target partial discharge type in the rule table, and the feature probability of each waveform feature and each analysis feature is determined based on the judgment result.

[0108] Wherein, for each target partial discharge type, if the waveform feature or the analysis feature of the analysis graph corresponding to the data group matches the signal feature information corresponding to the target partial discharge type, the feature probability corresponding to the waveform feature or the analysis feature is 1, and if the waveform feature or the analysis feature does not match the signal feature information, the feature probability corresponding to the waveform feature or the analysis feature is 0.

[0109] The embodiment of the present application can determine the type probability Xi according to the 8 types of features (i.e. waveform features and analysis features) in the rule table. In the N real-time analysis results in the data group, if a certain type of waveform feature and analysis feature matches the corresponding type in the rule table, Xi=1, and if the type does not match, Xi=0.

[0110] S407: calculating the type probability P corresponding to each target partial discharge type of each data group based on a first formula, wherein the first formula is: ; Wherein, Xi represents the feature probability corresponding to the waveform feature or the analysis feature in the data group, q is equal to the total number of waveform features and analysis features in the data group, and 0

[0111] It can be understood that when the data group corresponds to the Mth to the M+Nth acquisition period, q=N The total number of waveform features and analysis features corresponding to the data of a single acquisition period is 8. For example, when N=13, q=13 8=104.

[0112] In the embodiment of the present application, steps S405-S407 can be executed after step S403. In this way, the probability of the existence of each target partial discharge type of partial discharge anomaly can be determined.

[0113] Further, step S406 can further include steps S408-S410.

[0114] S408: if the waveform feature and the analysis feature of the analysis graph corresponding to the data group do not match the signal feature information corresponding to each target partial discharge type, determining whether the flight map feature presents a significant feature according to the flight map feature similarity.

[0115] wherein the flight chart feature presenting a significant feature corresponds to a different flight chart feature similarity when presenting a non-significant feature. For example, if the wave peak-valley analysis is multi-peak and the trend feature is overall enhancement, the flight chart feature presents a significant feature (or "has obvious features"), and the flight chart feature similarity = 0.99; if the wave peak-valley analysis is multi-peak or the trend feature is overall enhancement, the flight chart feature presents a significant feature, and the flight chart feature similarity = 0.4; if neither is satisfied, the flight chart feature presents a non-significant feature (or "has no obvious features"), and the flight chart feature similarity = 0.

[0116] It should be noted that the value of the flight chart feature similarity can be a fixed value set based on operation and maintenance experience, or a dynamic value adaptively adjusted based on actual conditions and requirements. For example, each flight chart similarity feature can have a floating value within its corresponding interval, which is not limited in the embodiments of the present application.

[0117] Based on this, the embodiments of the present application can further determine whether there is particle discharge.

[0118] It can be understood that, since the 8 types of features of particle discharge are similar to the features of no partial discharge anomaly, it cannot be determined whether there is particle discharge based on the 8 types of features; therefore, if the current waveform feature and the analysis feature do not match the air gap discharge type, the surface discharge type, the suspended discharge type, the corona discharge type and the like in the rule table, it can be further determined whether there is particle discharge based on the flight chart feature.

[0119] S409: If the flight chart feature presents a significant feature, the flight chart feature similarity when presenting a significant feature is determined as the type probability corresponding to the particle discharge type.

[0120] If the flight chart feature presents a significant feature, the particle discharge is obvious, and it can be determined that there is a particle discharge anomaly. At this time, the type probability corresponding to the particle discharge type can be the similarity of the flight chart feature.

[0121] In this way, the probability of the particle discharge partial discharge anomaly can be obtained.

[0122] S410: Based on the type probability corresponding to each partial discharge type of each data group, a partial discharge diagnosis result is obtained.

[0123] In this way, the partial discharge diagnosis result of the edge side device corresponding to each data group can be obtained, and the partial discharge diagnosis result of each edge side device is used to represent the abnormal probability and the partial discharge type of the power grid terminal covered by the edge side device.

[0124] In some implementations, steps S401-S410 can be performed by a data analysis model.

[0125] Further, step S410 can specifically include steps S4101-S4103.

[0126] S4101: For each data group, determine the maximum probability value among the type probabilities corresponding to each partial discharge type, and count the number of type probabilities equal to the maximum probability value.

[0127] It can be understood that the specific value of the maximum probability value depends on the actual calculation situation.

[0128] S4102: If the number of type probabilities equal to the maximum probability value is one, determine that the partial discharge type corresponding to the data group is the target partial discharge type or the particle discharge type corresponding to the maximum probability value, and determine that the abnormal probability is equal to the maximum probability value, to obtain the partial discharge diagnosis result of the edge-side device corresponding to the data group.

[0129] That is, the embodiment of the present application can take the partial discharge type with the maximum type probability among the five types of partial discharge types (target partial discharge type and particle discharge type) as the final analysis result.

[0130] S4103: If the number of type probabilities equal to the maximum probability value is multiple, determine that the partial discharge type corresponding to the data group is a random one among the target partial discharge type or the particle discharge type corresponding to the maximum probability value, and determine that the abnormal probability is equal to the maximum probability value, to obtain the partial discharge diagnosis result of the edge-side device corresponding to the data group.

[0131] That is, if multiple partial discharge types have the same type probability, a random one of them is taken as the final analysis result.

[0132] It needs to be supplemented that, continuing to refer to Figure 2 The embodiment of the present application can also use the data analysis model to comprehensively diagnose the batch sampling data (data group), that is, batch sampling diagnosis, to eliminate the occasional risk. Specifically, the embodiment of the present application can input the data group formed by the same edge-side device in continuous multiple sampling periods into the data analysis model as a whole, the data analysis model is a pre-trained model, can perform joint analysis on the mechanism characteristics of all samples in the data group at one time, directly output the partial discharge type and abnormal probability of the power grid terminal covered by the edge-side device, thereby avoiding the accidental error caused by single sampling, and realizing stable diagnosis of batch data.

[0133] It can be seen that the embodiment of the application provides a voting mechanism, each real-time analysis result in the data set can participate in the voting process of the type probability P, the type probability result of multiple collection periods is obtained, and the partial discharge type corresponding to the maximum type probability is taken as the final diagnosis result, and the maximum type probability is directly set as the abnormal probability of the partial discharge of the power grid terminal, so that the partial discharge type and the risk probability are output once.

[0134] In some implementations, the embodiment of the application can analyze the real-time analysis result and the target data corresponding to a single collection period, and determine whether the waveform feature or the analysis feature of the analysis graph corresponding to the single collection period meets the signal feature information corresponding to the target partial discharge type. If the feature probability meets the feature, the value of the feature probability is 1, and if the feature probability does not meet the feature, the value of the feature probability is 0. Then, the type probability P' of the data corresponding to various target partial discharge types of the single collection period is calculated based on a second formula. The second formula is: ; wherein Xi' represents the feature probability corresponding to the waveform feature or the analysis feature in the data of the single collection period, t is the total number of the waveform feature and the analysis feature, and 0

[0135] In the embodiment of the application, t is equal to 8, i = 0, 1,..., 7, and the second formula is specifically 。

[0136] In some implementations, the embodiment of the application can calculate the type probability corresponding to the particle discharge type when the waveform feature and the analysis feature of the analysis graph corresponding to the single collection period do not meet the signal feature information corresponding to each target partial discharge type.

[0137] Then, the partial discharge type corresponding to the maximum type probability (or the maximum probability value) is taken as the partial discharge type corresponding to the collection period (if the number of the maximum type probability is more than one, a random one of the target partial discharge type or the particle discharge type corresponding to the maximum type probability can be randomly selected), and the maximum probability value is determined as the abnormal probability corresponding to the collection period.

[0138] In this way, the abnormal probability evaluation of the partial discharge monitoring sampling can be realized, and the partial discharge type can be subdivided.

[0139] In some implementations, the embodiment of the application can collect the partial discharge type and the abnormal probability of N collection periods. If the number of a certain partial discharge type appearing in the N collection periods exceeds a preset number threshold, and / or the number of the abnormal probability exceeding a preset probability threshold exceeds a preset number threshold, the partial discharge type can be taken as the final partial discharge analysis result, and the average of all type probabilities corresponding to the partial discharge type is taken as the corresponding abnormal probability.

[0140] Further, the method provided by the embodiment of the application can further include the following step S601: gathering the partial discharge types and abnormal probabilities corresponding to each edge side device to form a monitoring list.

[0141] It can be understood that after the partial discharge types and abnormal probabilities are determined, the data can be summarized and integrated by the embodiment of the application. The partial discharge types and abnormal probabilities of the power grid terminal covered by each edge side device are gathered together to form a detailed monitoring list. Comprehensive and intuitive information is provided for the operation and maintenance personnel, which facilitates monitoring and management of the operation of the power grid terminal covered by each edge side device.

[0142] In some implementations, continuing to refer to Figure 2 , the operation and maintenance personnel, maintenance personnel and management personnel of the user side can view the partial discharge monitoring, diagnosis and early warning data by logging in the web page (page) of the cloud intelligent agent, that is, the diagnosis results and treatment suggestions can be viewed, and manual auditing can be performed. In this way, the user can view the overall situation of the power grid terminal partial discharge monitoring, for example, whether there is a power grid terminal with high partial discharge abnormal probability, whether there is a power grid terminal with excessively high humidity, whether there is a power grid terminal with confirmed partial discharge and being processed, and the like, that is, the station room that needs to be focused on can be viewed.

[0143] In some implementations, in addition to generating the monitoring list, the embodiment of the application can also generate an early warning for the power grid terminal with a partial discharge abnormality, divide the early warning levels according to the probability size, give an abnormal ranking and a treatment suggestion, and provide a terminal list that needs to be focused on by the customer.

[0144] In some implementations, continuing to refer to Figure 2 , the embodiment of the application can also perform partial discharge influencing factor analysis, for example, analyzing the relationship between partial discharge and temperature and humidity, or partial discharge-temperature influencing factor analysis, which can be analyzed based on the actual situation by calling a data analysis model, and the embodiment of the application does not make specific limitation on this.

[0145] In some implementations, continuing to refer to Figure 2 , cabinet interference analysis and account influencing factor analysis can also be performed, and the embodiment of the application does not make specific limitation on this.

[0146] In some implementations, the embodiment of the application can also perform the following step S602: for each edge side device, generating a treatment suggestion based on the partial discharge type, abnormal probability, partial discharge monitoring data, environment monitoring data and / or account data of the power grid terminal.

[0147] Ledger data records the basic information, operating status, and / or maintenance history of power grid terminals, and can cover the entire lifecycle of power grid terminals. For example, this data may include basic equipment information (such as model), operating parameters, maintenance records, fault records, and environmental monitoring data.

[0148] In some implementations, the embodiments of the present application may further perform the following steps S603-S604.

[0149] S603: Obtain a judgment result for the partial discharge diagnosis result to form a positive sample and a negative sample; a positive sample corresponds to a judgment result that is consistent with the partial discharge diagnosis result, and a negative sample corresponds to a judgment result that is inconsistent with the partial discharge diagnosis result.

[0150] Understandably, continue to see Figure 2 The user can make a judgment on the diagnosis result, that is, determine whether there is partial discharge. If it is a false alarm, that is, there is no partial discharge abnormality, the embodiment of the present application can record this false alarm and use its related data as a negative sample training model. If it is not a false alarm, that is, there is a partial discharge abnormality, the diagnosis result can be used as a positive sample training model, and the user can perform manual disposal actions for the partial discharge abnormality.

[0151] S604: Using positive samples and negative samples to train a data analysis model.

[0152] By combining system diagnosis with manual diagnosis, sample data is continuously accumulated, and the accuracy of the partial discharge diagnosis model is improved through regular automatic learning and training.

[0153] For further information, see Figure 2 , embodiments of the present application can determine whether frequency adjustment is necessary, including adjusting the partial discharge monitoring frequency (or monitoring frequency) and adjusting the signal sampling frequency. For example, if a possible partial discharge anomaly is detected, monitoring can be strengthened, that is, the perception terminal signal sampling frequency, as well as the edge device monitoring frequency and data reporting frequency can be increased. Alternatively, normal frequency can be restored. This adjustment action can be notified by the cloud intelligent agent to the edge device, which then issues a command to control the perception terminal to adjust the sampling frequency.

[0154] Figure 4 Schematic diagram of the intelligent question-answering process provided in an embodiment of the present application.

[0155] like Figure 4 As shown, the agent-based partial discharge monitoring operation and maintenance method provided in the embodiment of the present application can be used to implement an intelligent operation and maintenance assistant, which can specifically include four levels.

[0156] The first level is the user interaction layer: for operation and maintenance personnel / inspectors / managers, they can interact with the assistant by logging into the web page of the cloud agent. Questions can be raised on this interface, and the system-generated answers can be viewed.

[0157] The second level is the system administrator operation layer. The system administrator can regularly import relevant knowledge or experience documents into the system to continuously update and expand the system's knowledge base, ensuring that the system can provide accurate and up-to-date information.

[0158] After updating the knowledge base, the system administrator can initiate model training to ensure that the system can understand and process new knowledge content.

[0159] The third level is the upper application of the agent, which is used for question transmission and answer display.

[0160] The fourth level is the text large model, which can be used to process and understand a large amount of text data. After receiving a question, the text large model will use the knowledge obtained through training to generate an answer.

[0161] Specifically, the text large model is trained based on preset training documents, which include partial discharge monitoring data, environmental monitoring data, terminal account data of the power grid, and / or partial discharge knowledge.

[0162] The method provided by the embodiment of the present application can further include steps S701-S702.

[0163] S701: In response to a text question raised by a user through a web page, call a text large model to analyze the question and generate an answer.

[0164] S702: Display the answer on the web page.

[0165] In this way, it is ensured that the user can interact with the cloud agent through an intuitive web interface and obtain satisfactory answers.

[0166] Figure 5 The first structure diagram of the partial discharge monitoring and operation system based on the agent provided by the embodiment of the present application.

[0167] Figure 6 The second structure diagram of the partial discharge monitoring and operation system based on the agent provided by the embodiment of the present application.

[0168] As Figure 5 and Figure 6As shown, the embodiments of the present application also provide an agent-based partial discharge monitoring operation and maintenance system, which can include a power grid terminal 100, a perception terminal 200, an edge side device 300 and a cloud agent 400. Among them, the power grid terminal 100 can be a switch cabinet, and multiple power grid terminals 100 can be located in the same power distribution room. One power distribution room can be configured with multiple perception terminals 200. The perception terminal 200 can be located at the end side and belong to the end side device. The edge side device 300 can be a station end node, used to manage the power grid terminal 100 and its related perception terminal in a power distribution room, and the edge side device 300 can be located at the edge side. Further, the cloud agent 400 can be located in the cloud. In this way, a cloud-edge-end collaborative architecture can be formed.

[0169] Further, the cloud can include an application layer, a model layer and a data layer. The application layer can be used to provide partial discharge monitoring, partial discharge early warning, partial discharge diagnosis and partial discharge auxiliary decision-making. The model layer can be used to provide partial discharge comprehensive diagnosis of partial discharge atlas CNN classification model, partial discharge mechanism feature analysis model and partial discharge influence factor analysis model, and can also be used to provide LLM text large model, and can be used to provide manual training of the model. The data layer can be used to provide monitoring data access and monitoring cycle remote adjustment instruction issuing, which specifically involves partial discharge monitoring data, environmental monitoring data, equipment account data, diagnostic decision data and partial discharge operation and maintenance knowledge.

[0170] The power grid terminal 100 is, for example, a 10kV switch cabinet of a power distribution room, an incoming line cabinet, a feeder cabinet, a bus coupler cabinet, a bus coupler isolation cabinet, a PT cabinet and / or an on-site transformer.

[0171] In some implementations, the perception terminal 200 can include a partial discharge sensor 201 and a temperature and humidity sensor 202. Each power distribution room can be configured with one temperature and humidity sensor 202 to collect environmental monitoring data of the power distribution room, for example, temperature and humidity. In addition, each switch cabinet 101 in each power distribution room is configured with one partial discharge sensor 201, which includes a double ultrasonic partial discharge sensor, or the partial discharge sensor 201 can also be a visual ultrasonic monitoring device. The visual ultrasonic monitoring device can include temperature, humidity and ozone sensors, and can capture ultrasonic waves and cooperate with camera image superposition to achieve the purpose of visualization.

[0172] It is worth noting that the perception terminal 200 can specifically collect data from the switch cabinet of the power distribution room, or from the incoming line cabinet, the feeder cabinet, the bus coupler cabinet, the bus coupler isolation cabinet, the PT cabinet and the on-site transformer.

[0173] Further, the edge-side device 300 is configured to generate target data and report the cloud agent 400; the target data is determined based on at least one power grid terminal 100 covered by the edge-side device 300, and the target data includes monitoring data and pre-diagnosis data, the pre-diagnosis data is generated when the partial discharge signal and / or discharge interference signal is identified in the monitoring data; The cloud agent 400 is configured to periodically collect target data reported by at least one edge-side device 300; when the target data is obtained in each period, the data analysis model is used to analyze the target data to obtain real-time analysis results; for each edge-side device 300, the real-time analysis results and the target data in an analysis interval are converged to obtain a data group; wherein the analysis interval refers to the interval corresponding to the Mth collection period to the M+Nth collection period, M and N are preset values; the data group is analyzed by using the data analysis model to obtain a partial discharge diagnosis result, the partial discharge diagnosis result includes an abnormal probability that the power grid terminal has a partial discharge anomaly and a partial discharge type.

[0174] In some implementations, the monitoring data includes partial discharge monitoring data; the cloud agent 400 is specifically configured to: perform atlas mapping on the partial discharge monitoring data to obtain an analysis atlas; the analysis atlas includes a partial discharge pulse signal atlas; a waveform analysis is performed on the partial discharge pulse signal atlas by using a graph classification and recognition model to determine the waveform features of the analysis atlas; wherein the waveform features include one or more of waveform type, peak information, and valley information, and the waveform type includes single peak, double peak, multi-peak, and flat peak; the real-time analysis result includes the waveform features.

[0175] In some implementations, the analysis atlas further includes a partial discharge phase analysis atlas, and the cloud agent 400 is further configured to: perform feature calculation on the partial discharge pulse signal atlas by using a mechanism model to obtain analysis features, the analysis features include amplitude dispersion feature, pulse balance degree feature, polarity feature, phase aggregation feature, 50Hz component significance feature, 100Hz component significance feature, and 50 / 100Hz component numerical relationship feature; perform feature calculation on the partial discharge phase analysis atlas by using the mechanism model to obtain flight map features; the real-time analysis result includes the analysis features and the flight map features.

[0176] In some implementations, the pre-diagnosis data includes partial discharge alarm information; the partial discharge types include a target partial discharge type and a particle discharge type; the cloud agent 400 is further configured to: determine whether the number of partial discharge alarm information in the data group is greater than a preset threshold; if the number of partial discharge alarm information in the data group is greater than the preset threshold, obtain a rule table from the database; the rule table includes signal feature information corresponding to different target partial discharge types; the target partial discharge types include, but are not limited to, air gap discharge type, surface discharge type, suspension discharge type, and corona discharge type; according to the rule table, determine the matching degree of the waveform feature and the analysis feature of the analysis graph corresponding to each data group with the signal feature information of each target partial discharge type in the rule table, and determine the feature probability of each waveform feature and each analysis feature based on the determination result; wherein, for each target partial discharge type, if the waveform feature or the analysis feature of the analysis graph corresponding to the data group meets the corresponding signal feature information, the feature probability corresponding to the waveform feature or the analysis feature is 1, and if it does not meet the signal feature information, the feature probability corresponding to the waveform feature or the analysis feature is 0; based on a first formula, calculate the type probability P of each data group corresponding to each target partial discharge type, and the first formula is: ; wherein Xi represents the feature probability corresponding to the waveform feature or the analysis feature in the data group, q is equal to the total number of waveform features and analysis features in the data group, and 0 < i ≤ q; If the waveform feature and the analysis feature of the analysis graph corresponding to the data group do not match the signal feature information corresponding to each target partial discharge type, determine whether the flight graph feature presents a significant feature according to the flight graph feature similarity; if the flight graph feature presents a significant feature, the flight graph feature similarity corresponding to the significant feature is determined as the type probability corresponding to the particle discharge type; based on the type probability of each data group corresponding to each partial discharge type, obtain the partial discharge diagnosis result.

[0177] In some implementations, the cloud agent 400 is further configured to: for each data group, determine the maximum probability value in the type probability of each partial discharge type corresponding to the data group, and count the number of type probabilities equal to the maximum probability value; if the number of type probabilities equal to the maximum probability value is one, determine that the partial discharge type corresponding to the data group is the target partial discharge type or the particle discharge type corresponding to the maximum probability value, and determine that the abnormal probability is equal to the maximum probability value, to obtain the partial discharge diagnosis result of the edge side device corresponding to the data group; if the number of type probabilities equal to the maximum probability value is multiple, determine that the partial discharge type corresponding to the data group is a random one of the target partial discharge type or the particle discharge type corresponding to the maximum probability value, and determine that the abnormal probability is equal to the maximum probability value, to obtain the partial discharge diagnosis result of the edge side device corresponding to the data group.

[0178] In some implementations, the cloud agent 400 is further configured to: aggregate the partial discharge type and the abnormal probability corresponding to each edge-side device 300 to form a monitoring list; and / or, for each edge-side device 300, generate a treatment suggestion based on the partial discharge type, the abnormal probability, the partial discharge monitoring data, the environmental monitoring data, and / or the account data of the power grid terminal 100, the account data being data recording basic information, operating state, and / or maintenance history of the power grid terminal; and / or, obtain a decision result for the partial discharge diagnosis result to form positive samples and negative samples, the positive samples corresponding to the decision result being consistent with the partial discharge diagnosis result, and the negative samples corresponding to the decision result being inconsistent with the partial discharge diagnosis result; and train a data analysis model using the positive samples and the negative samples.

[0179] In some implementations, the cloud agent 400 is further configured to: in response to a text question raised by a user through a web page, call a text large model to analyze the question and generate an answer; wherein the text large model is trained based on a preset training document, and the preset training document includes the partial discharge monitoring data, the environmental monitoring data, the account data of the power grid terminal 100, and / or partial discharge knowledge; and display the answer on the web page.

[0180] In some implementations, the pre-diagnosis data includes partial discharge alarm information, and the cloud agent 400 is further configured to: for each edge-side device 300, determine whether there is a discharge interference signal in the power grid terminal 100 covered by the edge-side device 300 based on the data group; if there is no discharge interference signal, determine that the partial discharge type is normal, and determine that the abnormal probability is equal to 0; if there is a discharge interference signal, determine whether the number of partial discharge alarm information in the data group is greater than a preset threshold; if the number of partial discharge alarm information in the data group is not greater than the preset threshold, determine that the partial discharge type is normal, and determine that the abnormal probability is equal to 0.

[0181] In summary, the method and system for monitoring and operating partial discharge based on an agent provided by the embodiments of the present application include the following three parts.

[0182] (1) Perception terminal 200 layer (end): using sensors to realize real-time monitoring of partial discharge signals of the power grid terminal 100 (such as a switch cabinet 101) on site; and by importing the live detection work report into the operation and maintenance system, realizing unified management of online monitoring and live detection data.

[0183] (2) Edge computing layer (edge): based on the edge-side device 300 (such as a station-side node), realizing centralized access to the perception terminal 200 (such as various monitoring devices in the distribution station room), and realizing overall management, time synchronization, and partial discharge pre-diagnosis of the local perception terminal 200 with a certain computing power.

[0184] (3) Cloud platform level (cloud): the cloud agent 400 relies on the power grid terminal 100 to place online monitoring data, temperature and humidity monitoring data, power grid terminal 100 (for example, switch cabinet 101) account data, local discharge knowledge and expert experience, and constructs two models of data analysis model and text large model, to realize local discharge monitoring, diagnosis, early warning and end-to-end application of disposal.

[0185] The local discharge monitoring and operation method and system based on an agent provided by the embodiments of the present application can realize real-time monitoring and fault early warning of indoor electrical equipment of a distribution station through cloud-edge-end cooperation, which helps to improve the operation reliability and maintenance efficiency of electrical equipment.

[0186] Figure 7 The structure diagram of the local discharge monitoring and operation device based on an agent provided by the embodiments of the present application is shown.

[0187] As shown in Figure 7 The local discharge monitoring and operation device based on an agent provided by the embodiments of the present application is applied to a cloud agent, and the device comprises: The acquisition module 1001 is configured to periodically acquire target data reported by at least one edge-side device, the target data being determined based on at least one power grid terminal covered by the edge-side device, and the target data comprising monitoring data and pre-diagnosis data, the pre-diagnosis data being generated when a local discharge signal and / or a discharge interference signal are identified in the monitoring data; The first analysis module 1002 is configured to, when the target data is acquired in each period, perform data analysis on the target data by using a data analysis model to obtain real-time analysis results; The aggregation module 1003 is configured to, for each edge-side device, aggregate the real-time analysis results and the target data in one analysis interval to obtain a data group; wherein one analysis interval refers to an interval corresponding to the Mth acquisition period to the M+Nth acquisition period, and M and N are preset values. The second analysis module 1004 is configured to analyze the data group by using the data analysis model to obtain a local discharge diagnosis result, the local discharge diagnosis result comprising an abnormal probability of a local discharge abnormality existing in the power grid terminal and a local discharge type.

[0188] In specific implementation, the present application further provides a computer storage medium, wherein the computer storage medium can store a program, and the program can include some or all steps in each embodiment of the local discharge monitoring and operation method based on an agent provided by the present application when executed. The storage medium can be a magnetic disc, an optical disc, a read-only memory (ROM) or a random access memory (RAM), etc.

[0189] It is easy to understand that the skilled in the art can combine, split, recombine, etc. the embodiments of the present application on the basis of the several embodiments provided in the present application to obtain other embodiments, and these embodiments do not exceed the protection scope of the present application.

[0190] The above detailed description of the embodiments of the present application has further detailed the purposes, technical solutions and beneficial effects of the embodiments of the present application. It should be understood that the above is only a specific implementation of the embodiments of the present application, and is not used to limit the protection scope of the embodiments of the present application. Any modification, equivalent replacement, improvement, etc. made on the basis of the technical solutions of the embodiments of the present application should be included in the protection scope of the embodiments of the present application.

Claims

1. A partial discharge monitoring and operation method based on an intelligent agent, characterized in that: Applied to a cloud agent, the method includes: Periodically collecting target data reported by at least one edge-side device, where the target data is determined based on at least one grid terminal covered by the edge-side device, and the target data includes monitoring data and pre-diagnostic data. The pre-diagnostic data is generated when a partial discharge signal and / or a discharge interference signal is identified in the monitoring data. When the target data is acquired in each cycle, the target data is analyzed using a data analysis model to obtain real-time analysis results; For each edge device, the real-time analysis results and the target data within an analysis interval are aggregated to obtain a data group; wherein the analysis interval refers to an interval corresponding to the Mth collection period to the M+Nth collection period, where M and N are preset values; The data group is analyzed using the data analysis model to obtain a partial discharge diagnosis result, wherein the partial discharge diagnosis result includes an abnormal probability and a partial discharge type of the partial discharge abnormality in the power grid terminal.

2. The agent-based partial discharge monitoring and operation method according to claim 1, characterized in that: The monitoring data includes partial discharge monitoring data; and the step of using a data analysis model to analyze the target data to obtain real-time analysis results includes: Plotting the partial discharge monitoring data to obtain an analysis spectrum; the analysis spectrum includes a partial discharge pulse signal spectrum; A waveform analysis is performed on the partial discharge pulse signal spectrum using a spectrum classification and recognition model to determine the waveform characteristics of the analysis spectrum; wherein the waveform characteristics include one or more of waveform type, peak information, and trough information, and the waveform types include single peak, double peak, multiple peaks, and stable peaks; and the real-time analysis results include the waveform characteristics.

3. The agent-based partial discharge monitoring and operation method according to claim 2, characterized in that: The analysis spectrum also includes a partial discharge phase analysis spectrum. After the step of plotting the partial discharge monitoring data to obtain the analysis spectrum, the method further includes: Using the mechanism model, characteristic calculation is performed on the partial discharge pulse signal spectrum to obtain analysis characteristics, wherein the analysis characteristics include amplitude dispersion characteristics, pulse balance characteristics, polarity characteristics, phase aggregation characteristics, 50Hz component significance characteristics, 100Hz component significance characteristics, and 50 / 100Hz component numerical relationship characteristics; The mechanism model is used to perform feature calculation on the partial discharge phase analysis spectrum to obtain a flight map feature; the real-time analysis result also includes the analysis feature and the flight map feature.

4. The agent-based partial discharge monitoring and operation method according to claim 3, characterized in that: The pre-diagnosis data includes partial discharge alarm information; the partial discharge type includes a target partial discharge type and a particle discharge type; and the step of performing data analysis on the data group using a data analysis model to obtain a partial discharge diagnosis result includes: Determining whether the amount of the partial discharge alarm information in the data group is greater than a preset threshold; If the number of the partial discharge alarm information in the data group is greater than the preset threshold, obtaining a rule table from a database; the rule table includes signal feature information corresponding to different target partial discharge types; the target partial discharge types include but are not limited to air gap discharge type, surface discharge type, suspension discharge type, and corona discharge type; According to the rule table, determining the degree of match between the waveform characteristics and analysis characteristics of the analysis spectrum corresponding to each data group and the signal characteristic information of each target partial discharge type in the rule table, and determining the characteristic probability of each waveform characteristic and each analysis characteristic based on the determination result; wherein, for each target partial discharge type, if the waveform characteristics or analysis characteristics of the analysis spectrum corresponding to the data group meet the corresponding signal characteristic information, the characteristic probability corresponding to the waveform characteristics or analysis characteristics is 1; if it does not meet the signal characteristic information, the characteristic probability corresponding to the waveform characteristics or analysis characteristics is 0; The type probability P of each data set corresponding to each target partial discharge type is calculated based on the first formula, which is: ； Wherein, Xi represents the feature probability corresponding to the waveform feature or analysis feature in the data set, q is equal to the total number of waveform features and analysis features in the data set, 0<i≤q; If the waveform characteristics and analysis characteristics of the analysis spectrum corresponding to the data group do not match the signal characteristic information corresponding to each of the target partial discharge types, determining whether the flight chart characteristics present significant characteristics based on the flight chart characteristic similarity; If the flight pattern features present significant features, determining the flight pattern feature similarity corresponding to when the significant features are present as the type probability corresponding to the particle discharge type; Based on the type probability of each data set corresponding to each partial discharge type, a partial discharge diagnosis result is obtained.

5. The agent-based partial discharge monitoring and operation method according to claim 4, characterized in that: The step of obtaining a partial discharge diagnosis result based on the type probability of each data group corresponding to each partial discharge type includes: For each data group, determining a maximum probability value among the type probabilities corresponding to each partial discharge type in the data group, and counting the number of type probabilities equal to the maximum probability value; If the number of type probabilities equal to the maximum probability value is one, determining that the partial discharge type corresponding to the data group is the target partial discharge type or the particle discharge type corresponding to the maximum probability value, and determining that the abnormal probability is equal to the maximum probability value, and obtaining a partial discharge diagnosis result of the edge-side device corresponding to the data group; If there are multiple type probabilities equal to the maximum probability value, determine that the partial discharge type corresponding to the data group is a random one of the target partial discharge type or particle discharge type corresponding to the maximum probability value, and determine that the abnormal probability is equal to the maximum probability value, and obtain the partial discharge diagnosis result of the edge side device corresponding to the data group.

6. The agent-based partial discharge monitoring and operation method according to claim 2, characterized in that: The monitoring data also includes environmental monitoring data; the method further includes: Gathering the partial discharge type and abnormal probability corresponding to each edge-side device to form a monitoring list; and / or, For each edge-side device, generating a treatment recommendation based on the partial discharge type, the abnormality probability, the partial discharge monitoring data, the environmental monitoring data, and / or the ledger data of the power grid terminal, where the ledger data is data recording basic information, operating status, and / or maintenance history of the power grid terminal; and / or, Obtaining a judgment result for the partial discharge diagnosis result to form a positive sample and a negative sample; the positive sample corresponds to a judgment result that is consistent with the partial discharge diagnosis result, and the negative sample corresponds to a judgment result that is inconsistent with the partial discharge diagnosis result; The data analysis model is trained using the positive samples and the negative samples.

7. The agent-based partial discharge monitoring and operation method according to claim 6, characterized in that: The method further comprises: In response to a text question raised by a user via a web page, a large text model is invoked to parse the question and generate an answer; wherein the large text model is trained based on a preset training document, and the preset training document includes the partial discharge monitoring data, the environmental monitoring data, the ledger data of the power grid terminal, and / or partial discharge knowledge; The answer is displayed on the web page.

8. The agent-based partial discharge monitoring and operation method according to claim 1, characterized in that: The pre-diagnosis data includes partial discharge alarm information; The step of analyzing the data group using the data analysis model to obtain a partial discharge diagnosis result further includes: For each of the edge-side devices, determining, based on the data group, whether the grid terminal covered by the edge-side device has the discharge interference signal; If the discharge interference signal does not exist, determining that the partial discharge type is normal and the abnormality probability is equal to 0; If the discharge interference signal exists, determining whether the number of the partial discharge alarm information in the data group is greater than a preset threshold; If the number of the partial discharge alarm information in the data group is not greater than the preset threshold, the partial discharge type is determined to be normal, and the abnormality probability is determined to be equal to 0.

9. An agent-based partial discharge monitoring and operation system, characterized in that: The system includes: an edge-side device and a cloud intelligent agent; The edge-side device is configured to: generate target data and report it to the cloud intelligent agent; the target data is determined based on at least one power grid terminal covered by the edge-side device, the target data includes monitoring data and pre-diagnosis data, and the pre-diagnosis data is generated when a partial discharge signal and / or a discharge interference signal is identified in the monitoring data; The cloud agent is configured to: Periodically collect target data reported by at least one edge device; When the target data is acquired in each cycle, the target data is analyzed using a data analysis model to obtain real-time analysis results; For each edge device, the real-time analysis results and the target data within an analysis interval are aggregated to obtain a data group; wherein the analysis interval refers to an interval corresponding to the Mth collection period to the M+Nth collection period, where M and N are preset values; The data group is analyzed using the data analysis model to obtain a partial discharge diagnosis result, wherein the partial discharge diagnosis result includes an abnormal probability and a partial discharge type of the partial discharge abnormality in the power grid terminal.

10. A partial discharge monitoring and operation and maintenance device based on an intelligent agent, characterized in that: Applied to a cloud intelligent entity, the device comprises: a collection module configured to: periodically collect target data reported by at least one edge-side device, the target data being determined based on at least one grid terminal covered by the edge-side device, the target data including monitoring data and pre-diagnostic data, the pre-diagnostic data being generated when a partial discharge signal and / or a discharge interference signal is identified in the monitoring data; The first analysis module is configured to: when the target data is acquired in each cycle, perform data analysis on the target data using a data analysis model to obtain real-time analysis results; an aggregation module configured to aggregate the real-time analysis results and the target data within an analysis interval for each edge device to obtain a data group; wherein the analysis interval refers to an interval corresponding to the Mth collection period to the M+Nth collection period, where M and N are preset values; The second analysis module is configured to: analyze the data group using the data analysis model to obtain a partial discharge diagnosis result, wherein the partial discharge diagnosis result includes an abnormality probability and a partial discharge type of the power grid terminal.

Images