Power grid fault joint intelligent diagnosis method based on multi-source data

Abstract

The invention discloses a power grid fault joint intelligent diagnosis method based on multi-source data, relates to the technical field of power grid fault diagnosis, and aims to solve the problem that the prior art affects rapid fault isolation and accident recovery. Comprising the following steps: machine account benchmark association: by taking a region, a plant station and a primary equipment machine account of a dispatching system as a benchmark model, identifying equipment machine accounts in a signal protection master station, an EMS system, a wave recording networking master station and a unified system by adopting a text intelligent identification mode, and associating the equipment machine accounts with inconsistent description in different systems to the benchmark model, the identifier of the same device in each system is ensured to be unique; multi-system information comprehensive acquisition and preprocessing: acquiring information of four systems to generate a comprehensive information packet; and information time sequence arrangement and recording analysis: analyzing the comprehensive information packet to obtain analysis data including fault elements, fault starting time, fault duration, reclosing action conditions, distance measurement results and the like. The method has the advantage of rapid and accurate power grid fault diagnosis.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power grid fault diagnosis, and more particularly to a power grid fault joint intelligent diagnosis method based on multi-source data. BACKGROUND

[0002] In recent years, with the rapid development of power grids, the power grid presents a complex form of multi-element cooperation of "source grid load storage", AC-DC hybrid, and high penetration of distributed power supply. The number of primary and secondary devices has increased dramatically, and the operation and maintenance scenarios have become increasingly diverse.

[0003] When a complex fault occurs in the power grid, a large number of action and alarm information will be sent. The existing technology uses manual analysis of this information, which not only has the problem of great difficulty, but also is prone to inaccurate fault positioning, which seriously affects the rapid isolation of faults and accident recovery. In view of this, we propose a power grid fault joint intelligent diagnosis method based on multi-source data. SUMMARY

[0004] The purpose of the present application is to provide a power grid fault joint intelligent diagnosis method based on multi-source data, which aims to solve the problem of the existing technology affecting the rapid isolation of faults and accident recovery.

[0005] To solve the above technical problems, the present application provides the following technical scheme: a power grid fault joint intelligent diagnosis method based on multi-source data, which comprises the following steps:

[0006] S1, account benchmark association: taking the region, station, and primary device account of the dispatching system as a benchmark model, using a text intelligent recognition method to recognize the device accounts in the main station of the protection and communication system, the EMS system, the main station of the wave recording network, and the unified and distributed system, and associating the inconsistent device accounts in different systems to the benchmark model to ensure that the same device has a unique identifier in each system;

[0007] S2, multi-system information comprehensive collection and preprocessing: when a fault occurs in the power grid, collect the protection action events, protection wave recording, switch quantity displacement, alarms, and device setting value information of the protection and communication system, the switch displacement, hard contact displacement, and fault time load change telemeter information of the EMS system, the wave recorder wave file of the wave recording network system, which contains fault wave recording, the device name, the station to which it belongs, the interval, and the voltage level of the unified and distributed system. If the fault is a line fault, collect the above information at both ends of the line. Use the associated device benchmark model to preprocess the device time and information receiving time of the same device information in different systems to eliminate time errors. The processed information is summarized to form a comprehensive information package;

[0008] S3, information timing arrangement and recording wave analysis: the comprehensive analysis of each information in the comprehensive information package is carried out, on the one hand, the action of each element of the protection device, the tripping of each switch, the information of load change is arranged in time sequence to obtain time sequence information, on the other hand, the collected protection recording wave and fault recording wave are analyzed to obtain analysis data containing fault element, fault starting time, fault duration, fault phase, fault type, reclosing action condition, ranging result.

[0009] More preferably, it also includes the time sequence information and recording wave analysis data obtained based on the above S3, combined with the primary topology relationship of the power grid and the action principle of each protection, analyzes the tripping reason of the equipment, for complex chain fault, combs the causal chain relationship of each equipment tripping, and clearly defines the cause, development process and final result of the fault, and according to the above fault analysis result, gives the fault disposal suggestion.

[0010] More preferably, it also includes using visualization display technology to dynamically display the whole process of power grid fault and detailed action information of each stage in the form of combination of text and picture, and reverses the fault occurrence process to assist professional personnel in power grid analysis and fault disposal.

[0011] More preferably, in the above S1, the text intelligent recognition method adopted is to extract features and match semantics from the text description of the protection signal master station, EMS system, recording wave networking master station and equipment account in the integrated system, accurately identify the differentiated expression of different systems to the same equipment, and then complete the association with the reference model, and ensure the consistency of equipment identification when calling multi-source data later.

[0012] More preferably, in the above S2, when eliminating time error, the clock of the dispatching system is taken as the standard reference to calculate the time offset of each device of the protection signal system, EMS system and recording wave networking system, and the time record of the collected information is uniformly corrected through timestamp correction algorithm to ensure the time dimension synchronization of all data in the comprehensive information package.

[0013] More preferably, in the above S3, the analysis of protection recording wave and fault recording wave is obtained by extracting recording wave file and combining Fourier transform algorithm to calculate electrical parameters at fault time.

[0014] More preferably, the combined primary topology relationship of the power grid includes the main wiring structure of the substation, the line connection path, and the device association logic, which cooperates with the action setting value and principle of the protection device.

[0015] Ideally, when performing feature extraction and semantic matching on the equipment ledger text description, three-dimensional feature weights of equipment model, factory number, and installation location are introduced, with equipment model accounting for 40%, factory number accounting for 35%, and installation location accounting for 25%. The similarity of equipment ledger text descriptions in different systems is calculated by weighting, and when the similarity exceeds 90%, they are determined to be the same equipment and the association is completed.

[0016] Ideally, a ledger update linkage mechanism should be established simultaneously. When the equipment information in the baseline ledger of the dispatching system changes, a ledger update instruction should be pushed to the main station of the information security system, the EMS system, the waveform recording network main station, and the unified distribution system to update the equipment ledger information in each system in a synchronous manner. This will avoid inconsistencies in equipment identification across multiple systems due to changes in the baseline ledger and ensure the long-term accuracy of multi-source data retrieval.

[0017] As can be seen from the above technical solution, the power grid fault joint intelligent diagnosis method based on multi-source data provided in this embodiment of the invention firstly involves the association of ledger benchmarks: using the regional, substation, and primary equipment ledgers of the dispatching system as benchmark models, and employing text intelligent recognition to identify the equipment ledgers of the information protection master station, EMS system, waveform recording network master station, and unified distribution system, associating inconsistent equipment ledgers from different systems with the benchmark model to ensure that the identifier of the same equipment is unique in each system; and secondly, the comprehensive collection and preprocessing of multi-system information: when a power grid fault occurs, the following are collected: protection action events, protection waveforms, switch changes, alarms, and equipment setting information from the information protection system; switch changes, hard contact changes, and load change telemetry information at the time of the fault from the EMS system; and waveform recording files from the waveform recording network system. The system includes fault recordings, equipment names, substations, bays, and voltage levels. If the fault is a line fault, the above information is collected from both ends of the line. Using the aforementioned associated equipment reference model, the device time and information reception time of the same equipment information in different systems are preprocessed to eliminate time errors. The processed information is then summarized into a comprehensive information package. Information timing organization and recording analysis: Comprehensive analysis is performed on the information in the comprehensive information package. On the one hand, the operation of each component of the protection device, the tripping of each switch, and the load change information are organized in chronological order to obtain timing information. On the other hand, the collected protection recordings and fault recordings are analyzed to obtain analytical data including the faulty component, fault start time, fault duration, fault phase, fault type, reclosing operation status, and ranging results. Compared with the prior art, the beneficial effects of this invention are:

[0018] Firstly, the present application takes the dispatching system account book as the benchmark, uses the text intelligent recognition to associate the multi-system account book with the three-dimensional feature weight, ensures the uniqueness of the equipment identification, collects the multi-system fault information and eliminates the time error, analyzes the fault by combining the Fourier transform, combs the cause-effect chain, replaces the manual analysis, reduces the difficulty and avoids the deviation, realizes the accurate positioning, and helps to quickly isolate the fault and recover;

[0019] Secondly, the present application establishes the account book updating linkage mechanism, synchronously pushes the updating instruction to each system when the benchmark account book is changed, avoids the data deviation, provides the reliable basis for the multi-source data calling, integration and analysis in the fault, and reduces the analysis error caused by the inconsistent data. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 The flowchart of the power grid fault joint intelligent diagnosis method based on multi-source data of the present application. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments.

[0022] Embodiment one:

[0023] In this embodiment, the regional, station and primary equipment account book of the dispatching system is taken as the benchmark model, the text intelligent recognition is adopted, the equipment account books in the main station of the signal protection, the EMS system, the main station of the wave recording networking, and the integrated and distributed system are recognized, the inconsistent equipment account books in different systems are associated to the benchmark model, and the identification of the same equipment in each system is ensured to be unique;

[0024] Among them:

[0025] The text intelligent recognition is a method for extracting features and matching semantics from the text description of the equipment account book in the main station of the signal protection, the EMS system, the main station of the wave recording networking, and the integrated and distributed system, accurately recognizing the differentiated expressions of the same equipment in different systems, and then completing the association with the benchmark model, ensuring the consistency of the equipment identification in the subsequent multi-source data calling. When extracting features and matching semantics from the text description of the equipment account book, the three-dimensional feature weight of the equipment model, the factory number and the installation position is introduced, wherein the weight proportion of the equipment model is 40%, the weight proportion of the factory number is 35%, and the weight proportion of the installation position is 25%. The similarity of the text description of the equipment account book in different systems is calculated by weighting, and when the similarity is more than 90%, it is determined that the same equipment is associated. The specific method is as follows:

[0026] Suppose that the equipment model weight proportion is 40% (denoted as w1=0.4), the factory number weight proportion is 35% (denoted as w2=0.35), and the installation position weight proportion is 25% (denoted as w3=0.25). The three-dimensional features of the equipment account in each system are extracted, and the feature vector of each equipment in each dimension is obtained. Suppose that the feature vector of a certain equipment in the system to be identified is X=(x1, x2, x3), and the feature vector of a certain equipment in the benchmark model is Y=(y1, y2, y3). x1 and y1 represent the model features of the equipment to be identified and the benchmark equipment, respectively. x2 and y2 represent the factory number features of the two, and x3 and y3 represent the installation position features of the two.

[0027] Then, the similarity of the equipment account text description in different systems is calculated by weighting. The similarity calculation formula is as follows: S=w1*sim(x1, y1)+w2*sim(x2, y2)+w3*sim(x3, y3). In the formula, S represents the similarity of the equipment to be identified and the benchmark equipment. sim(a, b) represents the semantic similarity of feature a and feature b, and the value range is [0, 1]. When two features are completely consistent, sim(a, b)=1, and when two features are completely inconsistent, sim(a, b)=0.

[0028] The calculated similarity S is compared with the set threshold (90%, i.e. 0.9). When S>0.9, it is determined that the equipment in the system to be identified and the equipment in the benchmark model are the same equipment, and the account of the equipment to be identified is associated with the benchmark model. When S≤0.9, no association is performed.

[0029] Meanwhile, a table account updating linkage mechanism is established. When the benchmark account of the dispatching system changes, the table account updating instruction is pushed to the main station of the signal maintenance center, the EMS system, the main station of the recording wave networking, and the unified and distributed system, and the equipment table account information in each system is updated synchronously, so as to avoid the inconsistency of the equipment identification in multiple systems caused by the change of the benchmark account, and to ensure the long-term accuracy of the multi-source data call.

[0030] Then when the power grid fails, the protection action event, protection recording, switch value change, alarm, device setting value information of the signal collection protection system, the switch value change, hard contact value change, fault time load change telemetry information of the EMS system, the recording wave file of the recording wave networking system, the recording wave file containing fault recording, the device name, the station to which the device belongs, the interval, the voltage level of the unified distribution system, if the fault is a line fault, the above information at both ends of the line is collected, the device time and information receiving time of the same device information in different systems are preprocessed using the aforementioned associated device reference model, the time error is eliminated, and the processed information is summarized to form a comprehensive information package (i.e., using the associated reference model, the device data of each system associated with the same reference model is used as source data, and the comprehensive information package is formed after preprocessing such as time error elimination).

[0031] Among them:

[0032] When eliminating the time error, the clock of the dispatching system is taken as the standard reference, the time offset of each device of the signal collection protection system, the EMS system and the recording wave networking system is calculated, the time record of the collected information is uniformly corrected through the time stamp correction algorithm, and the time dimension of all data in the comprehensive information package is synchronized.

[0033] The comprehensive analysis of each information in the comprehensive information package is carried out, on the one hand, the action of each element of the protection device, the tripping of each switch and the load change information are arranged in time sequence to obtain time sequence information, and on the other hand, the collected protection recording and fault recording are analyzed to obtain analysis data containing fault elements, fault start time, fault duration, fault phase, fault type, reclosing action, and ranging result;

[0034] Among them:

[0035] The analysis of the protection recording and the fault recording is obtained by extracting the recording file and calculating the electrical parameters at the fault time by combining the Fourier transform algorithm, specifically:

[0036] The protection recording and fault recording files are read, the data format and storage structure in the files are parsed, the voltage and current original waveform data in the files are extracted, the voltage waveform time sequence and the current waveform time sequence are obtained, the extracted voltage and current waveform data are preprocessed, the noise interference in the waveform is removed (using filtering algorithms such as low-pass filtering and Kalman filtering), and then the characteristic parameters of the waveform, i.e., the peak value, effective value, phase and harmonic component of the waveform, are extracted;

[0037] The Fourier transform algorithm is used to analyze the preprocessed voltage and current waveform data, and the electrical parameters at the fault time are calculated, specifically:

[0038] For a voltage signal u(t) with a period T, its Fourier series expansion is: wherein u(t) represents the voltage signal, t represents the time variable, for describing the change process of the voltage signal over time, a0 represents the DC component of the voltage signal u(t), the value of which is equal to the average value of the voltage signal in a period, reflecting the static offset level of the voltage signal, n represents the harmonic order, taking a positive integer (1, 2, 3,...), corresponding to the fundamental component when n = 1, and corresponding to each harmonic component when n ≥ 2, for distinguishing voltage components of different frequencies, ω0 represents the fundamental angular frequency of the voltage signal, and ω0 = 100π rad / s, a n represents the amplitude coefficient of the n-th harmonic cosine component in the voltage signal u(t), for determining the amplitude size of the n-th harmonic cosine component, the value of which is obtained by integral calculation, reflecting the contribution degree of the harmonic cosine component in the voltage signal, b n represents the amplitude coefficient of the n-th harmonic sine component in the voltage signal u(t), for determining the amplitude size of the n-th harmonic sine component, the value of which is obtained by integral calculation, reflecting the contribution degree of the harmonic sine component in the voltage signal;

[0039] Similarly, for the calculation of the current signal i(t), only replace u(t) with i(t);

[0040] According to the calculated electrical parameters, the fault type and fault phase are determined; combined with the calculated electrical parameters and the time information in the recording wave file, the fault starting time and fault duration are determined; according to the change characteristics of the electrical parameters and the topology relationship of the power grid, the fault element is identified; by analyzing the action signal of the protection device and the recording wave data, the reclosing action situation is judged; the voltage and current data in the fault recording wave and the impedance-based ranging algorithm are used to calculate the fault ranging result, and the distance from the fault point to the measuring point is determined.

[0041] The embodiment also includes based on the above-mentioned sorted time sequence information and recording wave analysis data, combined with the primary topology relationship of the power grid (including the main wiring structure of the substation, the line connection path, the device association logic, cooperating with the action setting value and principle of the protection device), the principle of each protection action, analyzing the device tripping reason, for the complex chain fault, combing the causal chain relationship of each device tripping, clarifying the cause, development process and final result of the fault, and according to the above-mentioned fault analysis result, giving the fault disposal suggestion, providing the basis for the operation and maintenance personnel to quickly troubleshoot the fault and restore the power supply of the power grid, and further improving the efficiency of the power grid fault disposal.

[0042] The embodiment also includes using the visualization display technology to dynamically display the whole process of the power grid fault and the detailed action information of each stage in the form of combination of text and pictures, reversing the fault occurrence process, and assisting the professional personnel in power grid analysis and fault disposal.

[0043] The embodiments of the present application disclose the preferred embodiments, but are not limited to the same. Those skilled in the art can easily understand the spirit of the present application according to the above embodiments, and make different inferences and changes, as long as they do not deviate from the spirit of the present application, and are within the protection scope of the present application.

Claims

1. A method for joint intelligent diagnosis of power grid faults based on multi-source data, characterized in that, The method comprises the following steps: S1, account base correlation: taking the region, station, and primary equipment account of the dispatching system as a base model, using text intelligent recognition, identifying the equipment account in the protection main station, EMS system, wave recording networking main station, and unified and distributed system, correlating the inconsistent equipment account in different systems to the base model, and ensuring that the same equipment is uniquely identified in each system; S2, multi-system information comprehensive collection and preprocessing: when a fault occurs in the power grid, the protection action event, protection recording wave, switch position change, alarm, and equipment setting value information of the protection system, the switch position change, hard contact position change, and fault time load change telemetry information of the EMS system, the wave recorder recording file of the wave recording networking system, which contains fault recording wave, and the device name, belonging station, interval, and voltage grade of the unified and distributed system are collected, the time error is eliminated by using the device information time and information receiving time of the same base model in each system, the processed information is summarized to form a comprehensive information package; S3, information time sequence arrangement and recording wave analysis: the comprehensive information package is analyzed, the protection device element action, switch tripping, and load change information are arranged in time sequence to obtain time sequence information, and the collected protection recording wave and fault recording wave are analyzed to obtain analysis data containing fault elements, fault start time, fault duration, fault phase, fault type, reclosing action, and ranging result. 2.The power grid fault joint intelligent diagnosis method based on multi-source data according to claim 1, characterized in that, Based on the time sequence information and recording wave analysis data obtained by S3, the device tripping reason is analyzed in combination with the power grid primary topology relationship and each protection action principle, the causal chain relationship of each device tripping is combed for complex cascading faults, the cause, development process, and final result of the fault are determined, and fault disposal suggestions are given according to the fault analysis result. 3.The power grid fault joint intelligent diagnosis method based on multi-source data according to claim 1, characterized in that, The visualization display technology is used to dynamically display the whole process of the power grid fault and the detailed action information of each stage in the form of combination of text and graphics, the fault occurrence is reversed, and professional personnel are assisted to analyze the power grid and dispose the fault. 4.The power grid fault joint intelligent diagnosis method based on multi-source data according to claim 1, characterized in that, In S1, the text intelligent recognition is achieved by extracting features and matching semantics from the text description of the equipment account in the protection main station, EMS system, wave recording networking main station, and unified and distributed system, accurately identifying the differentiated expressions of the same device in different systems, and then correlating the base model, to ensure the consistency of the device identification when calling multi-source data.

5. The method of claim 1, wherein, In S2, when eliminating the time error, the clock of the dispatching system is taken as a standard base, the time offset of each device of the protection system, EMS system, and wave recording networking system is calculated, the time record of the collected information is uniformly corrected by a timestamp correction algorithm, and the time dimension of all data in the comprehensive information package is synchronized.

6. The method of claim 1, wherein, In S3, the analysis of the protection recording wave and fault recording wave is achieved by extracting the recording file and calculating the electrical parameters at the fault time by the Fourier transform algorithm.

7. The method of claim 2, wherein, The combined power grid primary topology relationship includes substation main wiring structure, line connection path, device association logic, and the action setting value and principle of the protection device. 8.The method of claim 4, wherein, In the feature extraction and semantic matching of the equipment account text description, the three-dimensional feature weights of equipment model, factory number and installation location are introduced, among which the equipment model weight accounts for 40%, the factory number weight accounts for 35%, and the installation location weight accounts for 25%. The similarity of the equipment account text description in different systems is calculated by weighting, and when the similarity exceeds 90%, it is determined as the same equipment and the correlation is completed.

9. The method of claim 8, wherein, At the same time, the account updating linkage mechanism is established. When the reference account of the dispatching system changes, the account updating instruction is pushed to the main station of the signal maintenance system, the EMS system, the main station of the recording wave networking, and the unified and distributed system, and the equipment account information in each system is updated synchronously, so as to avoid the inconsistency of equipment identification in multiple systems caused by the change of the reference account, and ensure the long-term accuracy of multi-source data calling.

Images