CN113268590A - Power grid equipment running state evaluation method based on equipment portrait and integrated learning - Google Patents
Power grid equipment running state evaluation method based on equipment portrait and integrated learning Download PDFInfo
- Publication number
- CN113268590A CN113268590A CN202110369354.3A CN202110369354A CN113268590A CN 113268590 A CN113268590 A CN 113268590A CN 202110369354 A CN202110369354 A CN 202110369354A CN 113268590 A CN113268590 A CN 113268590A
- Authority
- CN
- China
- Prior art keywords
- equipment
- power grid
- monitoring
- data
- signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 23
- 238000012544 monitoring process Methods 0.000 claims abstract description 112
- 238000004458 analytical method Methods 0.000 claims abstract description 48
- 238000000034 method Methods 0.000 claims abstract description 47
- 238000004140 cleaning Methods 0.000 claims abstract description 9
- 230000007547 defect Effects 0.000 claims description 46
- 230000002159 abnormal effect Effects 0.000 claims description 31
- 238000012549 training Methods 0.000 claims description 18
- 238000012423 maintenance Methods 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 14
- 230000009471 action Effects 0.000 claims description 12
- 230000036541 health Effects 0.000 claims description 11
- 238000004422 calculation algorithm Methods 0.000 claims description 10
- 238000007726 management method Methods 0.000 claims description 10
- 238000005259 measurement Methods 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 9
- 230000001364 causal effect Effects 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 8
- 238000005065 mining Methods 0.000 claims description 5
- 239000003990 capacitor Substances 0.000 claims description 4
- 238000011161 development Methods 0.000 claims description 4
- 230000010354 integration Effects 0.000 claims description 4
- 230000000903 blocking effect Effects 0.000 claims description 3
- 238000007405 data analysis Methods 0.000 claims description 3
- 230000007613 environmental effect Effects 0.000 claims description 3
- 238000007418 data mining Methods 0.000 abstract description 2
- 230000011664 signaling Effects 0.000 description 7
- 238000011160 research Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 230000005856 abnormality Effects 0.000 description 4
- 230000001932 seasonal effect Effects 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 239000011248 coating agent Substances 0.000 description 3
- 238000000576 coating method Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 238000003058 natural language processing Methods 0.000 description 3
- 238000009826 distribution Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 208000025274 Lightning injury Diseases 0.000 description 1
- 108010076504 Protein Sorting Signals Proteins 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 238000013523 data management Methods 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 238000004141 dimensional analysis Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000008014 freezing Effects 0.000 description 1
- 238000007710 freezing Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000003340 mental effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000011158 quantitative evaluation Methods 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 238000012916 structural analysis Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Educational Administration (AREA)
- Software Systems (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Public Health (AREA)
- Evolutionary Computation (AREA)
- Primary Health Care (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Water Supply & Treatment (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention relates to a power grid equipment running state evaluation method based on equipment portrait and integrated learning, and belongs to the technical field of power regulation and control. The method comprises the following steps: standardizing power grid operation monitoring signals and identifying and cleaning monitoring data; modeling the operation monitoring data of the power grid with equipment as a center; analyzing the logic relation between the power grid topology and the monitoring signals; monitoring alarm information clustering by using equipment with a power grid event as a center; device portrayal and ensemble learning based on monitoring information; and evaluating the running state of the power grid equipment. The method provided by the invention realizes intelligent analysis of the monitoring data of the substation equipment, and improves the management and big data mining efficiency of the monitoring data of the substation equipment, so that the regulation and control integrated operation efficiency is improved, and the method is easy to popularize and apply.
Description
Technical Field
The invention belongs to the technical field of electric power regulation and control, and particularly relates to a power grid equipment running state evaluation method based on equipment portrait and integrated learning.
Background
In recent years, with the continuous expansion of the scale of a power grid, the intelligentization level of the power grid is continuously improved, and the quantity of substations, power grid equipment and alarm information which are connected to a regulation and control mechanism to realize centralized monitoring is increased dramatically. Hundreds of substations and related lines are controlled by a small number of operators in a centralized manner, the operators monitor the operation state of the power grid equipment by means of remote signaling and remote measuring information, and the working pressure and the mental pressure of the operators are increased day by day. The health state of the power grid equipment is closely related to the reliability and economy of equipment operation.
The monitoring information data volume of the current regulation and control center is huge, various operation data and historical data of a transformer substation are included, valuable early warning information and knowledge experience are formed through multi-dimensional analysis on mass data, and the problem to be solved is urgent. In the power grid operation management process, problems and deterioration trends of equipment and even the power grid operation state can be fed back by remote signaling alarm, remote measurement out-of-limit, power transmission and transformation online monitoring, field inspection and test data, but the following defects generally exist:
1. the control center lacks deep analysis and excavation of massive monitoring data, and the problem that monitoring information is nonstandard and irregular simultaneously cannot effectively monitor the running state of equipment.
2. The existing monitoring information analysis is basically manual processing and analysis, and a monitoring information normative analysis tool and an intelligent processing tool are lacked.
3. Due to the fact that the monitoring equipment is large in information quantity, screening and analysis of the monitoring information are mostly processed by excel tables, structured association of equipment signals is not established, and the phenomenon of important monitoring information omission easily occurs.
Therefore, how to overcome the defects of the prior art is a problem to be solved urgently in the technical field of power regulation at present.
Disclosure of Invention
The invention aims to solve the defects of the prior art, and provides a power grid equipment running state evaluation method based on equipment portrait and integrated learning, which comprises the steps of extracting, cleaning, mining, associating and labeling power grid data, then carrying out portrait modeling on the equipment, mainly mining the correlation between the health state of the equipment and various analysis factors, continuously correcting and perfecting the model by adopting an integrated learning method in machine learning, automatically identifying the health state condition of the equipment, evaluating the running state, realizing intelligent analysis of monitoring data of transformer substation equipment on the basis, improving the monitoring data management and big data mining efficiency of the transformer substation equipment, improving the regulation and control integrated operation efficiency, expanding the depth and breadth of professional management and accelerating the practical construction of technical support means, the safe and efficient operation of the regulation and control integration is guaranteed.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a power grid equipment running state evaluation method based on equipment representation and integrated learning comprises the following steps:
step (1), standardizing power grid operation monitoring signals and identifying and cleaning monitoring data;
step (2), modeling the operation monitoring data of the power grid with equipment as the center;
step (3), analyzing the logic relation between the power grid topology and the monitoring signals, and establishing a multi-dimensional power grid operation mode identification model, an associated signal judgment model, a trip event analysis model and a monitoring information causal relation model;
step (4), equipment monitoring alarm information clustering taking a power grid event as a center;
step (5), equipment portrait and ensemble learning based on monitoring information;
and (6) evaluating the running state of the power grid equipment.
Further, preferably, in the step (1), the structured data and the text data generated by the power grid operation monitoring are subjected to natural language analysis, the monitoring signals are analyzed item by item, the meaning of the monitoring signals is subjected to standard processing, and suspicious data are identified and cleaned.
Further, preferably, in the step (2), by extracting the device keyword in the operation management data, establishing association with the device, and using the device ID as a center, and by adopting a mode of positioning the master device at intervals, the association relationship between the monitoring information, the alarm information, the measurement data, the operation data, the position status, and the switch, the disconnecting link, and the bus device is established, so as to implement data integration based on the device.
Further, it is preferable that the specific method of step (3) is as follows:
analyzing the running state of equipment and the condition of an accident based on the opening and closing states of a breaker and a disconnecting link and in combination with the wiring mode, the running mode and the topological relation of a power grid, checking an uploading signal according to an accident signal standard sequence, and checking whether signal loss and redundancy exist or not to obtain whether the conditions of missing report and false report exist or not;
according to the equipment retrieval, the associated maintenance application, the operation ticket and the fault and defect abnormal information, the equipment is classified according to the equipment manufacturer and the equipment model, and an equipment file and a package are established for each equipment
According to the telemetering data and the power grid topology, when the active power and the reactive power of the equipment in the running state are 0, the equipment is defined as abnormal in the running state, and the load flow balance of each end of the line, the load flow balance of three sides of a main transformer and the load flow balance of each outgoing line of a bus are analyzed; when active and reactive remote measurement is 0, carrying out power grid topology analysis through intervals according to corresponding intervals of equipment, combining related equipment, establishing a multi-dimensional power grid operation mode identification model, and carrying out comprehensive judgment on abnormal power grid operation states;
classifying equipment non-accident signals and signals which do not influence the actual power grid operation according to the topological relation of the power grid equipment, the signal meaning, the signal sending time and the signal resetting time, and establishing an associated signal judgment model; stripping the associated signal from the trip event, and not taking the stripped signal as a basis for analyzing the trip event;
thirdly, a trip event analysis model: judging the type, development process and possible missed signal of the trip event according to the transformer substation, voltage class, interval and equipment;
the tripping event is based on tripping of a line, a bus, a main transformer, a capacitor and a reactor, and tripping classification is carried out based on switch associated equipment; when the switch is changed from closing to opening, the front and the back of tripping are analyzed; firstly, based on an overhaul application ticket, if the switch is overhauled, overhauling and mistakenly sending signals are filtered, and the filtered alarm signals are used as analysis sources for analysis; if the switch is not maintained, the switch is defined as a tripping event, the tripping event is confirmed by taking the standard sequence of the tripping event as a basis and the interval accident total and protection outlet signals as identification starting points and combining the switching-off signals in the tripping event, and reclosing action, equipment switching-on, interval accident signals, total station accident total and signal resetting important signals are analyzed in sequence to confirm the tripping event; the trip classification is distinguished based on instantaneous faults and permanent faults, the instantaneous faults are defined when the switch is reclosed successfully, and the permanent faults are defined when the switch is not reset after the switch is tripped again after the switch is reclosed; establishing a line trip, main transformer trip and bus trip event analysis rule model based on signal analysis, confirmed trip and classified judgment;
monitoring information causal relationship model: analyzing the equipment alarm information, automatically associating and acquiring the equipment sending the information or the interval of the equipment, analyzing possible reasons generated by the equipment in combination with relevant maintenance information, operation information, log information, weather and thunder monitoring information within 15 minutes of signal sending, and in combination with the relation between a specific signal and actual primary and secondary equipment, a power grid model and a current operation mode, and giving out a plurality of reasons to speculate the signal meaning, reason analysis and consequence of each monitoring event for modeling.
Further, preferably, in the step (4), based on the data identification and cleaning of the monitoring data, a data association relationship between the original signal and the standard signal and an association relationship between the standard signal and the evened rule base are established, and meanwhile, according to the evened rule base, a multi-level association relationship is automatically established for a large amount of originally discrete monitoring data, and the monitoring data are packaged into trip information, operation and association, maintenance and debugging, AVC information, monitoring defects and abnormal monitoring events in a blocking manner, so that equipment monitoring alarm information clustering taking the power grid event as a center is realized.
Further, preferably, in the step (5), the warning signal is analyzed by using a natural language, the device defect, the operation condition, the operation age, the family defect, the online monitoring and the weather are selected as characteristic quantities to portray the device, then an Adaboost algorithm in an ensemble learning algorithm is used for training the classifier, various parameters such as the device defect, the operation data and the environmental data are accumulated for calculation to obtain a strong classifier of various parameters, and the strong classifier is used for carrying out multi-dimensional data analysis on the primary device to give the health state of the device.
Further, preferably, in the step (6), the operation condition of the power grid is analyzed, and the macroscopic safety of the power grid is evaluated as a whole: dividing the whole power grid into a plurality of sub-areas according to the specified area according to actual contact, analyzing and identifying transformer substation stations with problems in the sub-areas, positioning the problems of a certain device in the transformer substation, and analyzing the state of the power grid to specific devices through multi-level mining; meanwhile, weak links in a grid structure of the power grid are analyzed and positioned in combination with a wiring mode, an operation mode and power grid tide of the power grid, the influence range and the accident level of various devices when the devices break down are analyzed, the tripping probability of the devices is combined for integral evaluation, the safety state of the power grid is comprehensively analyzed, and a final evaluation value and a fault handling suggestion are given.
Compared with the prior art, the invention has the beneficial effects that:
the invention provides a method for realizing quantitative evaluation and health degree division of a power grid device operation state on the basis of analyzing power grid monitoring operation big data, which has the following effects: and establishing a multi-dimensional portrait of the equipment according to the correlation and correlation analysis of the equipment information and various models, management, operation, real-time and production data, wherein the multi-dimensional portrait comprises the interval of the equipment, the voltage level, the wiring form, the operation age, the fault frequency, the operation and maintenance frequency and the equipment use number. Through quantitative analysis of the health state of the equipment, the health state of the equipment and the logic relation among various health-affecting factors can be displayed more visually, a theoretical basis is provided for establishing evaluation and early warning of the operation state of the power grid equipment, an equipment operation state evaluation result is provided for an equipment management department, equipment management personnel can pay attention to the equipment operation state evaluation result and arrange maintenance as soon as possible, and meanwhile, a reference basis is provided for subsequent type selection of the substation equipment.
Drawings
FIG. 1 is a flow chart of a method for evaluating the operating condition of a power grid device based on device representation and ensemble learning;
fig. 2 is a flow chart of a companion signal analysis method.
Detailed Description
The present invention will be described in further detail with reference to examples.
It will be appreciated by those skilled in the art that the following examples are illustrative of the invention only and should not be taken as limiting the scope of the invention. The examples do not specify particular techniques or conditions, and are performed according to the techniques or conditions described in the literature in the art or according to the product specifications. The materials or equipment used are not indicated by manufacturers, and all are conventional products available by purchase.
A power grid equipment running state evaluation method based on equipment representation and integrated learning comprises the following steps:
step (1), standardizing power grid operation monitoring signals and identifying and cleaning monitoring data;
step (2), modeling the operation monitoring data of the power grid with equipment as the center;
step (3), analyzing the logic relation between the power grid topology and the monitoring signals, and establishing a multi-dimensional power grid operation mode identification model, an associated signal judgment model, a trip event analysis model and a monitoring information causal relation model;
step (4), equipment monitoring alarm information clustering taking a power grid event as a center;
step (5), equipment portrait and ensemble learning based on monitoring information;
and (6) evaluating the running state of the power grid equipment.
Preferably, in the step (1), the structured data and the text data generated by the power grid operation monitoring are analyzed in natural language, the monitoring signals are analyzed item by item, the meaning of the monitoring signals is subjected to standard processing, and suspicious data are identified and cleaned.
Preferably, in the step (2), by extracting the device keywords in the operation management data, establishing association with the device, and taking the device ID as a center, and adopting a mode of positioning the master device at intervals, the association relationship between the monitoring information, the alarm information, the measurement data, the operation data, the position state and the switch, the disconnecting link and the bus device is established, so as to realize data integration based on the device.
Preferably, the specific method of step (3) is as follows:
analyzing the running state of equipment and the condition of an accident based on the opening and closing states of a breaker and a disconnecting link and in combination with the wiring mode, the running mode and the topological relation of a power grid, checking an uploading signal according to an accident signal standard sequence, and checking whether signal loss and redundancy exist or not to obtain whether the conditions of missing report and false report exist or not; the standard signal sequence takes the single-phase instantaneous fault of a line interval as an example, and the conditions of missing, redundant, missing report and false report of an alarm signal are checked according to the sequence.
Single-phase instantaneous fault signal standard sequence:
xxx switch/protection 1 outlet actions
Xxx switch/protection 2 outlet actions
Opening of xxx switch
Accidental opening of xxx switches
Xxx line/Accident Total Signal action
Xxx/total station accident Total Signal action
Xxx switch/protection 1 outlet reset
Xxx switch/protection 2 outlet reset
Xxx switch/first set control loop disconnect action
Xxx/second set of control loops open
Xxx switch/first set control loop disconnection reset
Xxx switch/second set control loop disconnection reset
Xxx/total station incident total signal regression
Xxx switch/accident total signal recovery
Xxx switch/recloser outlet action
Xxx switch/recloser outlet action
Xxx switch closing
Xxx switch/recloser outlet reset
Xxx switch/recloser outlet reset
Xxx Fault oscillograph Start action
Xxx fault recorder start-up reset
According to the equipment retrieval, the associated maintenance application, the operation ticket and the fault and defect abnormal information, the equipment is classified according to the equipment manufacturer and the equipment model, an equipment file is established for each equipment, the equipment file comprises the specific operation, fault, maintenance and defect times of the equipment in the whole life cycle, and the equipment file is shown in a table 1.
TABLE 1 device Profile example
According to the telemetering data and the power grid topology, when the active power and the reactive power of the equipment in the running state are 0, the equipment is defined as abnormal in the running state, and the load flow balance of each end of the line, the load flow balance of three sides of a main transformer and the load flow balance of each outgoing line of a bus are analyzed; when active and reactive remote measurement is 0, carrying out power grid topology analysis through intervals according to corresponding intervals of equipment, combining related equipment, establishing a multi-dimensional power grid operation mode identification model, and carrying out comprehensive judgment on abnormal power grid operation states;
classifying equipment non-accident signals and signals which do not influence the actual power grid operation according to the topological relation of the power grid equipment, the signal meaning, the signal sending time and the signal resetting time, and establishing an associated signal judgment model; stripping the associated signal from the trip event, and not taking the stripped signal as a basis for analyzing the trip event;
thirdly, a trip event analysis model: judging the type, development process and possible missed signal of the trip event according to the transformer substation, voltage class, interval and equipment;
the tripping event is based on tripping of a line, a bus, a main transformer, a capacitor and a reactor, and tripping classification is carried out based on switch associated equipment; when the switch is changed from closing to opening, the front and the back of tripping are analyzed; firstly, based on an overhaul application ticket, if the switch is overhauled, overhauling and mistakenly sending signals are filtered, and the filtered alarm signals are used as analysis sources for analysis; if the switch is not maintained, the switch is defined as a tripping event, the tripping event is confirmed by taking the standard sequence of the tripping event as a basis and the interval accident total and protection outlet signals as identification starting points and combining the switching-off signals in the tripping event, and reclosing action, equipment switching-on, interval accident signals, total station accident total and signal resetting important signals are analyzed in sequence to confirm the tripping event; the trip classification is distinguished based on instantaneous faults and permanent faults, the instantaneous faults are defined when the switch is reclosed successfully, and the permanent faults are defined when the switch is not reset after the switch is tripped again after the switch is reclosed; establishing a line trip, main transformer trip and bus trip event analysis rule model based on signal analysis, confirmed trip and classified judgment;
monitoring information causal relationship model: analyzing the alarm information of the equipment, automatically associating and acquiring the equipment sending the information or the interval of the equipment, analyzing possible reasons of the equipment in combination with relevant overhaul information, operation information, log information, weather and thunder monitoring information within 15 minutes of signal sending, the relation between a specific signal and actual primary and secondary equipment, a power grid model and a current operation mode, giving a plurality of reasons, and presuming the meaning of the signal of each monitoring event, the analysis of the reasons and the modeling of the consequences.
Preferably, in the step (4), based on data identification and cleaning of the monitoring data, a data association relationship between an original signal and a standard signal and an association relationship between a standard signal and an evened rule base are established, and meanwhile, according to the evened rule base, a multi-level association relationship is automatically established for a large amount of originally discrete monitoring data, and the monitoring data are packaged into trip information, operation and association, overhaul and debugging, AVC information, monitoring defects and abnormal monitoring events in a blocking manner, so that equipment monitoring alarm information clustering taking a power grid event as a center is realized.
Preferably, in the step (5), the warning signal is analyzed by using a natural language, the device defect, the operation condition, the operation age, the family defect, the online monitoring and the weather are selected as characteristic quantities to portray the device, then an Adaboost algorithm in an integrated learning algorithm is used for training a classifier, various parameters such as the device defect, the operation data and the environmental data are accumulated for calculation to obtain a strong classifier of various parameters, and the strong classifier is used for carrying out multi-dimensional data analysis on primary devices to give the health state of the device.
Preferably, in the step (6), the operation condition of the power grid is analyzed, and the macroscopic safety of the power grid is evaluated integrally: dividing the whole power grid into a plurality of sub-areas according to the specified area according to actual contact, analyzing and identifying transformer substation stations with problems in the sub-areas, positioning the problems of a certain device in the transformer substation, and analyzing the state of the power grid to specific devices through multi-level mining; meanwhile, weak links in a grid structure of the power grid are analyzed and positioned in combination with a wiring mode, an operation mode and power grid tide of the power grid, the influence range and the accident level of various devices when the devices break down are analyzed, the tripping probability of the devices is combined for integral evaluation, the safety state of the power grid is comprehensively analyzed, and a final evaluation value and a fault handling suggestion are given.
Examples of the applications
A power grid equipment running state evaluation method based on equipment representation and integrated learning comprises the following steps:
step (1), standardizing power grid operation monitoring signals and identifying and cleaning monitoring data;
step (2), modeling the operation monitoring data of the power grid with equipment as the center;
step (3), analyzing the logic relation between the power grid topology and the monitoring signals, and establishing a multi-dimensional power grid operation mode identification model, an associated signal judgment model, a trip event analysis model and a monitoring information causal relation model;
step (4), equipment monitoring alarm information clustering taking a power grid event as a center;
step (5), equipment portrait and ensemble learning based on monitoring information;
and (6) evaluating the running state of the power grid equipment.
The specific implementation method of the step (1) comprises the following steps:
(1) the structured data and the text data are analyzed based on natural language, each monitoring signal is analyzed in a syntactic manner to obtain an accurate monitoring signal meaning, the meaning is automatically analyzed to an equipment level, the data and the text are processed into contents which can be recognized by a computer, and the standardization of the power grid operation monitoring data is improved.
(2) And adopting univariate data identification and cross data identification methods to identify the conditions of incomplete access data, abnormal quality, invalid information, associated information, equipment state, non-correspondence to remote signaling alarm and remote measuring alarm, and non-correspondence to switch or disconnecting link position.
(3) Establishing and perfecting an electric power corpus according to the characteristics of primary and secondary equipment and monitoring information, carrying out grammar inference and syntactic analysis on each monitoring signal based on a Natural Language Processing (NLP), and screening and filtering invalid contents by utilizing segmentation and reconstruction. In addition, the power grid topology is also required to be combined, error data in remote signaling and remote measurement are subjected to cross identification, and operation data and management data in unstructured data are cleaned.
The specific implementation method of the step (2) is as follows: by extracting the device keywords in the operation management data, establishing association with the devices, and taking the device ID as the center, establishing a data model based on the devices by adopting a mode of positioning the main devices at intervals. The power grid equipment model mainly comprises a switch, a disconnecting link, a trolley switch, a grounding disconnecting link, a circuit, a bus, a transformer, a voltage transformer and a current transformer. The associated data comprises alarm information, equipment operation data, protection information, fault recording information, remote measurement remote signaling remote control information, other data, overhaul information, fault information, equipment defects, primary equipment ledger, operation ticket information and station information.
The specific implementation method of the step (3) is as follows: by analyzing the power grid operation mode and the equipment topological relation, the logical relation between the power grid topology and the monitoring signals is analyzed, and the following model is established:
(1) the multi-dimensional power grid operation mode identification model comprises the following steps: on the basis of the power grid operation mode analysis model, in order to prevent data abnormality existing in the remote signaling state information, a multi-dimensional power grid operation mode identification model is established, cross identification is carried out on related data, and the operation mode of the power grid is comprehensively judged.
Analyzing the running state of equipment and the condition of an accident based on the opening and closing states of a breaker and a disconnecting link and in combination with the wiring mode, the running mode and the topological relation of a power grid, checking an uploading signal according to an accident signal standard sequence, checking whether signal loss and redundancy exist or not, and obtaining whether the conditions of missing report and false report exist or not;
according to equipment retrieval, associated maintenance application, operation tickets, fault and defect abnormal information, refining and classifying the equipment according to equipment manufacturers and equipment models, and establishing an equipment file for each equipment, wherein the equipment file comprises the specific operation, fault, maintenance and defect times of the equipment in the whole life cycle;
according to the telemetering data and the power grid topology, when the active power and the reactive power of the equipment in the running state are 0, the equipment is defined as abnormal in the running state, and the load flow balance of each end of the line, the load flow balance of three sides of a main transformer and the load flow balance of each outgoing line of a bus are analyzed; when active and reactive remote measurement is 0, carrying out power grid topology analysis through intervals according to corresponding intervals of equipment, combining related equipment, establishing a multi-dimensional power grid operation mode identification model, and carrying out comprehensive judgment on abnormal power grid operation states;
(2) an associated signal judgment model: and establishing an associated signal judgment model according to the topological relation of the power grid equipment, the signal meaning, the signal sending time and the signal resetting time. Stripping the associated signal from the trip event, and not taking the stripped signal as a basis for analyzing the trip event;
associated signal reasoning principle: the first choice needs to be the signals sent by the same equipment, or certain topological relation exists between the equipment related to the signals; and secondly, limiting a certain time range according to the on-off operation time of the switch and the disconnecting link, and analyzing an associated signal. The main characteristic points are as follows:
1) the signal corresponds to a standard signal;
2) the method is suitable for a wiring mode and an operation mode;
3) relationships to operating equipment (e.g., inter-bay disconnecting links, inter-bay connecting master devices, inter-bay link buses);
4) a limit earlier than the operating time;
5) later than the limit of the operating time.
(3) And (3) analyzing and modeling a trip event: judging the type, development process and possible missed signal of the trip event according to the transformer substation, voltage class, interval and equipment;
the tripping event is based on tripping of a line, a bus, a main transformer, a capacitor and a reactor, and tripping classification is carried out based on switch associated equipment; when the switch is changed from closing to opening, the front and the back of tripping are analyzed; firstly, based on an overhaul application ticket, if the switch is overhauled, overhauling and mistakenly sending signals are filtered, and the filtered alarm signals are used as analysis sources for analysis; if the switch is not maintained, the switch is defined as a tripping event, the tripping event is confirmed by taking the standard sequence of the tripping event as a basis and the interval accident total and protection outlet signals as identification starting points and combining the switching-off signals in the tripping event, and reclosing action, equipment switching-on, interval accident signals, total station accident total and signal resetting important signals are analyzed in sequence to confirm the tripping event; the trip classification is distinguished based on instantaneous faults and permanent faults, the instantaneous faults are defined when the switch is reclosed successfully, and the permanent faults are defined when the switch is not reset after the switch is tripped again after the switch is reclosed; and establishing a line trip, main transformer trip and bus trip event analysis rule model based on signal analysis, confirmed trip and classified judgment.
(4) Monitoring information causal relationship model: analyzing the equipment alarm information, automatically associating and acquiring the equipment sending the information or the interval of the equipment, analyzing possible reasons generated by the equipment in combination with relevant maintenance information, operation information, log information, weather and thunder monitoring information within 15 minutes of signal sending, and in combination with the relation between a specific signal and actual primary and secondary equipment, a power grid model and a current operation mode, and giving out a plurality of reasons to speculate the signal meaning, reason analysis and consequence of each monitoring event for modeling.
The specific implementation method of the step (4) is as follows: based on the processed monitoring data, establishing a data association relation between an original signal and a standard signal and an association relation between the standard signal and an evened rule base, automatically establishing an atom → molecule → cell association relation for the original discrete and large amount of monitoring data according to the evened rule base, classifying the monitoring data of the monitoring signals according to the trip information, operation and association, maintenance and debugging, AVC information, monitoring defects and other subject monitoring events after the original signal is standardized, and realizing equipment monitoring alarm information clustering taking the power grid event as the center through the trip information. And a machine learning technology is adopted to automatically analyze the full class of events of the subsequent incremental monitoring operation data, so as to promote the analysis and research of the correctness of the accident and abnormal signal and realize the automatic identification of the power grid fault and abnormal 'event'.
The specific implementation method of the step (5) is as follows:
(1) study of equipment portrait based on monitoring information: after removing the signals in the normal event, the remaining valid alarm signals are analyzed to render an image of the device. The effective alarm signals comprise missed-sending signals, false-sending (suspicious) signals, frequently-sending and overtime unreset signals and telemetering data of current, voltage, active power, idle power and oil temperature.
a. Equipment anomaly index system: the service life, the defect and the fault information of the retired equipment are statistically displayed by taking the commissioning date, the retired date, the equipment type, the equipment model and the voltage grade as dimensions and taking the equipment running time, the fault times, the defect times, the effective alarm and the heavy overload condition as fact tables, so that data support is provided for equipment health analysis. The following is the classification of the deduction index system, and the specific contents of the items and deduction values are shown in tables 2 and 3.
Table 2 deduction factors of the transformer based on the device image:
table 3 deduction factors of the switch based on the device picture:
1) device defect
Classifying the defects of various kinds of spaced primary and secondary equipment of the transformer and the switch, establishing a corresponding index deduction system, and deducting according to the defects of the equipment. Sources of defect data include: and the OMS comprises a defect processing flow, effective alarm after monitoring signal eventing and online power transmission and transformation monitoring. For the same device, if the same type of defect information can be located from different data sources, the most serious condition is taken as the deduction of the item (to avoid repeated deduction). In the process, a multi-dimensional characteristic point pattern recognition technology is needed to perform structural analysis based on an abnormal index deduction system on defect and remote signaling data. And part of alarm information needs to consider the frequency of signal sending and the signal recovery time.
2) Operating conditions
And establishing a deduction index of the bad working condition, and comprehensively bringing the deduction index into an equipment abnormal index deduction system. The method comprises the following steps:
short circuit impact accumulation: considering factors including short circuit impact frequency, impact current and impact duration;
switching-on and switching-off times of the switch: when reaching a certain number of times, the maintenance is needed;
the switch and the disconnecting link are not operated for a long time;
the times and the accumulated duration of heavy overload of the line and the current heavy overload condition;
the times and the accumulated duration of the main variable weight overload and the current heavy overload condition;
the times and the accumulated duration of the oil temperature exceeding the limit, and the current oil temperature;
and the bus voltage is out of limit and the current voltage is obtained.
3) Operating life
According to the operation age, the equipment is classified according to the age, the middle age and the cyan, and the corresponding scores are directly deducted by the 'old' equipment. The primary equipment and the secondary equipment are set to have different ages according to different types.
4) Family defects
And directly deducting corresponding scores for manufacturers and models which are identified as familial defects according to the regulation of an abnormal index system. If the defect is not qualitative, after the manufacturer and the model of the equipment are standardized, the device number, the accumulated running time and the probability of different types of defects of the equipment with the same manufacturer and the same model are counted, the probability value is larger than a certain threshold, the suspected familial defect is set, and the deduction is carried out according to a certain proportion.
5) Factor of maintenance application
After the equipment is overhauled, the defect data generated in a period of time is considered to be cleared (mainly equipment defects analyzed from alarm signals; and information in the defect records can be judged according to whether the defects are eliminated or not).
And the detailed deduction rule corresponding to the specific equipment monitoring state index system refers to the attached table I and the attached table II.
b. The external factors affect: according to different types of natural disasters and meteorological conditions, the fault probability of equipment under different conditions is researched according to historical data. And aiming at various external factors, establishing an association relation with a line (through a tower coordinate) or equipment in the station according to the coordinate range of each external condition.
1) Thunder and lightning
And analyzing the lightning falling range, the lightning falling times and the lightning stroke time, and researching the tripping times of the related equipment in the time interval. For the lines, the analysis is performed in conjunction with the length of the line in the lightning zone. The analyzed content is mainly the mean failure probability of the equipment; for general equipment, calculating the fault probability according to the average fault probability in the lightning environment, and converting the fault probability into deduction; and for equipment which is easily influenced by lightning, calculating according to the individual probability model of the equipment. The calculation method is shown in (2) 1 in a of the tripping probability of the abnormal state equipment in the concrete implementation method of the step (6).
2) Mountain fire
And analyzing the range and the grade of the mountain fire, researching the trip times of the line in a time interval, and analyzing by combining the line length in a fire passing area. The analyzed content is mainly the mean failure probability of the equipment; for general equipment, calculating the fault probability according to the average fault probability in a mountain fire environment, and converting the fault probability into a deduction; for equipment which is easily affected by mountain fire, calculation is carried out according to an individual probability model of the equipment. The calculation method is shown in (2) 1 in a of the tripping probability of the abnormal state equipment in the concrete implementation method of the step (6).
3) Ice coating
And analyzing the freezing range and grade, researching the tripping times of the line in a time interval, and analyzing by combining the line length. And selectively researching the relation between the icing and tripping correlation model and the line parameters according to the condition of the parameters. For underground cables, the influence of ice coating is not considered. The calculation method is shown in (2) 1 in a of the tripping probability of the abnormal state equipment in the concrete implementation method of the step (6).
4) Typhoon (big wind)
The time, the path range and the wind power of the typhoon are researched, the probability of circuit tripping in the coverage range of the typhoon is analyzed, and the circuit tripping and the typhoon influence in the area range are associated by voltage grades according to the length of the circuit. For underground cables, the effect of typhoons is not taken into account. The calculation method is shown in (2) 1 in a of the tripping probability of the abnormal state equipment in the concrete implementation method of the step (6).
5) Weather (weather)
Under the premise of removing extreme weather, the correlation among weather types (rain and snow), air temperature, main transformer oil temperature and equipment tripping is researched. Based on weather conditions when equipment trips, including rain and snow, air temperature and main transformer temperature, relevance analysis is carried out according to time scales, a main transformer temperature and meteorological condition influence threshold value is established, and the relation between gear shifting, oil temperature out-of-limit and weather influence in operation of the main transformer is researched. The calculation method is shown in (2) in a 2) of the tripping probability of the abnormal state equipment in the concrete implementation method of the step (6).
6) Seasonal factors
On the premise of eliminating extreme weather, the seasonal rule of equipment fault tripping is researched, clustering is carried out according to tripping reasons in seasons, and research is respectively carried out by utilizing the voltage grade, the grid of the located area and the characteristic points of line types (overhead lines or cables). The calculation method is shown in (2) in a 2) of the tripping probability of the abnormal state equipment in the concrete implementation method of the step (6).
c. Comprehensively evaluating the hidden danger of the equipment: comprehensively considering the abnormality of the equipment (such as more equipment failure times and frequent tripping) and the influence of external factors (such as weather, thunder and lightning and ice coating), and comprehensively evaluating the failure probability of the equipment.
(2) Integrated learning research: adaboost is an iterative algorithm, and the core idea is to train different classifiers for the same training set, and then to assemble the weak classifiers to form a stronger classifier. The algorithm itself is implemented by changing the data distribution according to each sample x in each training setiWhether the classification of (1) is correct, and the accuracy of the last overall classification 1-emDetermining a weight D of each samplem(i) In that respect Will be provided withModifying the new data set of the over-weighted value to the lower classifier for training, and finally obtaining the classifier f obtained by each trainingm(x) Fused together as the final decision classifier h (x).
a. Sample(s)
Training sample set D { (x)i,yi)|i＝1,...,N}，xi∈X,yi∈Y＝{-1,+1}，xiDenotes the i-th element, y, in Xi1 denotes training sample xiIs a negative sample, yiWith +1 representing a training sample xiIs a positive sample.
b. Training process
1) Initializing training samples xiWeight D1(i) 1., N. If the number of positive and negative samples is consistent, then
If the number of positive and negative samples is N respectively+,N-Then, the positive sample:
negative sample:
2) for 1, M, training weak classifier fm(x)＝L(D,Dm) E { -1, +1}, estimate weak classifier fm(x) Classification error rate e ofm. Such as:
normalization:
obtaining a strong classifier:
the specific implementation method of the step (6) is as follows: the safety condition of a real-time power grid is analyzed, the macroscopic safety of the power grid is integrally evaluated, the weak links of the grid structure are analyzed, the influence range and the accident level of different devices when the devices break down are analyzed, and the safety state of the power grid is comprehensively analyzed. The content comprises the following steps:
(1) weak link of grid structure: the method is characterized in that the method is combined with the wiring mode, the operation mode and the tide distribution of the power grid, weak links in the grid structure of the power grid are analyzed, and the operation mode influencing the safety and stable operation of the power grid is analyzed and extracted: the main transformers running in parallel are inconsistent in gear, and run in an electromagnetic looped network and an isolated network.
(2) Abnormal state equipment trip probability: the method comprehensively considers the influence of the abnormality of the equipment and external factors, researches the failure probability of the equipment, and comprehensively evaluates the risks of equipment power failure, equipment heavy overload, isolated network operation and other successive failures caused by the tripping of the related equipment.
a. Obtaining the tripping probability of the equipment according to the evaluation result of the equipment abnormity index system (comprehensively considering the abnormity of the equipment and the influence of external factors); the equipment trip probability to be calculated here is actually a multiple of the probability of failure of the respective equipment to be evaluated with respect to the equipment without any abnormality. And (4) the abnormity of the equipment is firstly converted into the deduction of the equipment, and then the fault probability of the equipment is wholly evaluated. The equipment itself is abnormal and has been deducted by deduction system (principle of deduction: the problem existing in a certain aspect is converted into defect, corresponding to critical defect, deduction 41 points; corresponding to serious defect, deduction 11 points; corresponding to general defect, deduction 2 points).
1) Mountain fire, icing, typhoon, thunder and lightning harsh environment: calculating according to the severe environment grade and the line length within the severe environment range, and evaluating based on historical data (calculating for various severe environments and different voltage grades respectively):
the calculation formula between the line length of the environment, the corresponding deduction parameter Len and the real line length is as follows:
(maximum value of Len sets upper limit 9). Wherein LenbaseAs a basic parameter, an initial value is set to 1 km first, and then historical data is used for refreshing (in the case of insufficient data, parameter adjustment may not be performed first): grouping for the same type, same class of cases, LenbaseSequentially calculating the parameters Len by taking 0.1 kilometer as a starting point and 0.1 kilometer as a step length to obtain a plurality of groups of line length parameters; the real situation is divided into a group of relation data of 'line length VS trip probability' according to the line length (with 0.1 kilometer as a unit), the occurrence frequency and the real trip frequency, and the real situation probability with the shortest line length is used as a base number 1 to convert the real situation into a group of numerical values related to the line length. According to different LenbaseThe parameter set obtained by calculation corresponds to the wire length appearing in the real data situation, and a group of values (each Len) related to the wire length is also obtainedbaseOne for each group). Comparing theoretical parameters with actual parameters, using
Calculating, and superposing a set of numbers, each LenbaseCorresponding to a calculated value.
The calculation is carried out on each group (information with the same type and the same grade) to obtain each group of calculated values and the minimum LenbaseTo adjust the parameters and obtain the results.
And (3) the grades of the severe environment are divided into 1-3 grades from high to low, initial values are set to be 9, 3 and 1 respectively corresponding to the deduction coefficient LV (the subsequent refreshing is carried out through historical data, the adjustment process of the parameters similar to the wire length is carried out, the highest grade is kept to be 9, and the parameters of the other two grades are adjusted).
Is badThe environment corresponding deduction value is calculated according to the following formula:
Jumpprothe trip parameter of the time is initially determined to be 1, and is calculated according to the occurrence frequency of the problems in the historical data and the real trip frequency to obtain:
Jumpprothe value range is set to [0.1,9 ]]。
2) Weather, seasonal factors: on the premise of eliminating the abnormal and severe environment of the equipment, the seasonal and weather correlation rules of fault tripping of the equipment are researched, and clustering is carried out according to tripping reasons. In the clustering process, the voltage grade, the located area grid and the line type (overhead line or cable) are used as characteristic points for grouping, and the fault probability of the equipment under different conditions is respectively counted. And adding a weather and season factor knowledge base for the condition that the fault probability exceeds the average fault probability of the similar equipment by more than 2 times, wherein the deduction calculation formula is as follows:
the value interval of the deduction value of the single factor is [3, 11%]。
3) The fault probability multiplying power is as follows: the corresponding formula is: deduction x deduction impact factor. And the deduction influence factor is used as a parameter object to be adjusted, the initial value is set to be 0.25, and the parameter is adjusted according to the relation between the real trip and the equipment deduction and the duration time. The value interval of the multiplying power is set as [1,30 ].
b. And (3) comprehensively evaluating equipment trip influence: analyzing the wiring mode and the operation mode of the power grid, and researching the influence caused after the related equipment trips, wherein the method comprises the following steps: main transformer power failure (calculating loss load); double-circuit, parallel running transformers (including load transfer caused by a self-throwing device) and heavy overload of equipment in a section caused by equipment tripping; the disconnection and the loop disconnection of the power grid and the isolated network operation caused by the disconnection and the loop disconnection.
c. Risk of sequential failure: tracking switch opening locking information and protection fault information, and extracting successive faults possibly caused by refusing to operate; other equipment overload situations that may be caused by equipment tripping are studied to extract the successive faults that may be caused by equipment overload.
(3) The power grid safety state index system comprises: and establishing a power grid safety state index system, and comprehensively evaluating the safety and stability level of the running state of the power grid.
a. Based on weak links in the whole grid structure as key research objects, carrying out N-1 scanning on power grid equipment, and researching and evaluating the influence range and the accident level when a power grid fails; the study subjects included: the loss of load, the power failure of equipment or a transformer substation or a power plant, and the power failure of important users. And giving comprehensive assessment suggestions and treatment suggestions for the influence after evaluation according to the power grid operation mode and the condition of power transmission and outage equipment.
b. And according to the fault probability of the equipment (the evaluation result of an equipment abnormal index system), increasing the corresponding influence coefficient on the influence caused by the fault trip of the relevant equipment. The factors are summarized into possible power grid accident levels. And deducting according to the accident grade of the power grid (1 is deducted for 8-grade power grid accidents, the accident grade is improved by one grade and is deducted by 3 times of the previous grade), and superposing the accident grade deductions caused by single accident tripping in the range of the power grid. And comprehensively calculating a numerical value according to the deduction, and performing fault assessment and reminding through a defined value.
c. According to the probability of occurrence of the successive fault sets, overlapping corresponding accident grade deductions after the occurrence of the successive faults;
d. deducting abnormal operation modes;
e. and finally evaluating the safety state of the power grid, and considering the following factors: the total deduction condition of an evaluation system in the power grid range; and considering the scale of the power grid and the total deduction, evaluating in a mode of 'total deduction/power grid scale'. (the grid scale may be evaluated in terms of the number of substations, equipment, total load, etc. at different voltage levels);
f. the power grid safety state evaluation formula is as follows:
electric network scale: the scales of main transformers with different voltage grades are used as main reference standards, 1 35kV transformer is recorded in 1, a 110kV transformer is recorded in 3, a 220kV transformer is recorded in 9, a 500kV transformer is recorded in 27, and a 750kV (1000kV) transformer is recorded in 81.
Evaluating by adopting a mode of 'total deduction/power grid scale', wherein the weighted deduction value is more than or equal to 1, and the power grid state is red; the weighted deduction value is [0.5,1 ], and the power grid state is orange; the weighted deduction value is [0.1,0.5 ], and the power grid state is yellow; and the weighting deduction value is less than 0.1, and the power grid state is green (accounting is needed in a specific interval).
g. And (3) experimental verification:
the method comprises the steps of using power grid collection data from 2018 to 2019 in 10 months in a certain place as a training sample set D, training a classifier by using an Adaboost algorithm, obtaining a strong classifier H after N iterations, then classifying a part of test data sets in 2019 in 11 months by using the strong classifier H to obtain a classification result, and evaluating the accuracy of classification.
53562 parts of sample data, 52407 parts of training data and 1155 parts of test data, were actually used in the experiment. Preprocessing, word segmentation, semantic recognition, equipment association and NLP natural language analysis processing are carried out on each group of data before an experiment, the original data are subjected to standardized processing, an initially available training sample ' 110kV/v permanent open line 112 switch/d air pressure low total locking/s without telemetering data/i ' is obtained, plug-in faults/r ' are collected, then 1000 iterations are carried out by using the training sample to obtain a classifier constructed by an Adaboost algorithm, the recognition rate of corresponding unhealthy defects and fault occurrence results of the equipment obtained by testing the test sample data reaches 82.7%, and the error rate is 17.3%.
Experiments show that the method is used for evaluating the health state of the power grid equipment, and the obtained effect is better and better along with the increase of the accuracy of data and the iteration times.
The foregoing shows and describes the general principles, essential features, and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (7)
1. A power grid equipment running state evaluation method based on equipment representation and integrated learning is characterized by comprising the following steps:
step (1), standardizing power grid operation monitoring signals and identifying and cleaning monitoring data;
step (2), modeling the operation monitoring data of the power grid with equipment as the center;
step (3), analyzing the logic relation between the power grid topology and the monitoring signals, and establishing a multi-dimensional power grid operation mode identification model, an associated signal judgment model, a trip event analysis model and a monitoring information causal relation model;
step (4), equipment monitoring alarm information clustering taking a power grid event as a center;
step (5), equipment portrait and ensemble learning based on monitoring information;
and (6) evaluating the running state of the power grid equipment.
2. The method for evaluating the operation state of the power grid equipment based on the equipment representation and the ensemble learning of claim 1, wherein: in the step (1), the structured data and the text data generated by the operation monitoring of the power grid are analyzed in natural language, the monitoring signals are analyzed one by one, the meaning of the monitoring signals is subjected to standard processing, and suspicious data are identified and cleaned.
3. The method for evaluating the operation state of the power grid equipment based on the equipment representation and the ensemble learning of claim 1, wherein: in the step (2), by extracting the device keywords in the operation management data, establishing association with the device, and taking the device ID as the center, and adopting an interval positioning main device mode, establishing association relations between the monitoring information, the alarm information, the measurement data, the operation data and the position state and the switch, the disconnecting link and the bus device, and realizing data integration based on the device.
4. The method for evaluating the running state of the power grid equipment based on the equipment representation and the ensemble learning as claimed in claim 1, wherein the specific method in the step (3) is as follows:
analyzing the running state of equipment and the condition of an accident based on the opening and closing states of a breaker and a disconnecting link and in combination with the wiring mode, the running mode and the topological relation of a power grid, checking an uploading signal according to an accident signal standard sequence, and checking whether signal loss and redundancy exist or not to obtain whether the conditions of missing report and false report exist or not;
according to equipment retrieval, associated maintenance application, operation tickets, fault and defect abnormal information, the equipment is subjected to refined classification according to equipment manufacturers and equipment models, an equipment file is established for each piece of equipment, the equipment file comprises that when the specific operation, fault, active power and reactive power of the equipment in the whole life cycle are 0, the equipment file is defined as abnormal operation state, the tide balance of each end of a line and the tide balance of three sides of a main transformer are analyzed, and the tide balance of each outgoing line of a bus is analyzed; when active and reactive remote measurement is 0, carrying out power grid topology analysis through intervals according to corresponding intervals of equipment, combining related equipment, establishing a multi-dimensional power grid operation mode identification model, and carrying out comprehensive judgment on abnormal power grid operation states;
classifying equipment non-accident signals and signals which do not influence the actual power grid operation according to the topological relation of the power grid equipment, the signal meaning, the signal sending time and the signal resetting time, and establishing an associated signal judgment model; stripping the associated signal from the trip event, and not taking the stripped signal as a basis for analyzing the trip event;
thirdly, a trip event analysis model: judging the type, development process and possible missed signal of the trip event according to the transformer substation, voltage class, interval and equipment;
the tripping event is based on tripping of a line, a bus, a main transformer, a capacitor and a reactor, and tripping classification is carried out based on switch associated equipment; when the switch is changed from closing to opening, the front and the back of tripping are analyzed; firstly, based on an overhaul application ticket, if the switch is overhauled, overhauling and mistakenly sending signals are filtered, and the filtered alarm signals are used as analysis sources for analysis; if the switch is not maintained, the switch is defined as a tripping event, the tripping event is confirmed by taking the standard sequence of the tripping event as a basis and the interval accident total and protection outlet signals as identification starting points and combining the switching-off signals in the tripping event, and reclosing action, equipment switching-on, interval accident signals, total station accident total and signal resetting important signals are analyzed in sequence to confirm the tripping event; the trip classification is distinguished based on instantaneous faults and permanent faults, the instantaneous faults are defined when the switch is reclosed successfully, and the permanent faults are defined when the switch is not reset after the switch is tripped again after the switch is reclosed; establishing a line trip, main transformer trip and bus trip event analysis rule model based on signal analysis, confirmed trip and classified judgment;
monitoring information causal relationship model: analyzing the equipment alarm information, automatically associating and acquiring the equipment sending the information or the interval of the equipment, analyzing possible reasons generated by the equipment in combination with relevant maintenance information, operation information, log information, weather and thunder monitoring information within 15 minutes of signal sending, and in combination with the relation between a specific signal and actual primary and secondary equipment, a power grid model and a current operation mode, and giving out a plurality of reasons to speculate the signal meaning, reason analysis and consequence of each monitoring event for modeling.
5. The method for evaluating the operation state of the power grid equipment based on the equipment representation and the ensemble learning of claim 1, wherein: in the step (4), based on the data identification and cleaning of the monitoring data, a data association relationship between an original signal and a standard signal and an association relationship between a standard signal and an evened rule base are established, meanwhile, according to the evened rule base, a multi-level association relationship is automatically established for original discrete and large amount of monitoring data, and the monitoring data are packaged into monitoring events of trip information, operation and association, overhaul and debugging, AVC information, monitoring defects and abnormity in a blocking manner, so that equipment monitoring alarm information clustering taking a power grid event as a center is realized.
6. The method for evaluating the operation state of the power grid equipment based on the equipment representation and the ensemble learning of claim 1, wherein: in the step (5), the natural language is used for analyzing the alarm signal, the equipment defects, the operation conditions, the operation age, the family defects, the online monitoring and the weather are selected as characteristic quantities to portray the equipment, then an Adaboost algorithm in an integrated learning algorithm is used for training a classifier, various parameters such as the defects of the equipment, the operation data and the environmental data are accumulated for calculation to obtain a strong classifier of various parameters, and the strong classifier is used for carrying out multi-dimensional data analysis on the primary equipment to give the health state of the equipment.
7. The method for evaluating the operation state of the power grid equipment based on the equipment representation and the ensemble learning of claim 1, wherein: in the step (6), analyzing the operation condition of the power grid, and integrally evaluating the macroscopic safety of the power grid: dividing the whole power grid into a plurality of sub-areas according to the specified area according to actual contact, analyzing and identifying transformer substation stations with problems in the sub-areas, positioning the problems of a certain device in the transformer substation, and analyzing the state of the power grid to specific devices through multi-level mining; meanwhile, weak links in a grid structure of the power grid are analyzed and positioned in combination with a wiring mode, an operation mode and power grid tide of the power grid, the influence range and the accident level of various devices when the devices break down are analyzed, the tripping probability of the devices is combined for integral evaluation, the safety state of the power grid is comprehensively analyzed, and a final evaluation value and a fault handling suggestion are given.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110369354.3A CN113268590A (en) | 2021-04-06 | 2021-04-06 | Power grid equipment running state evaluation method based on equipment portrait and integrated learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110369354.3A CN113268590A (en) | 2021-04-06 | 2021-04-06 | Power grid equipment running state evaluation method based on equipment portrait and integrated learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113268590A true CN113268590A (en) | 2021-08-17 |
Family
ID=77228536
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110369354.3A Pending CN113268590A (en) | 2021-04-06 | 2021-04-06 | Power grid equipment running state evaluation method based on equipment portrait and integrated learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113268590A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114123192A (en) * | 2021-11-30 | 2022-03-01 | 广东电网有限责任公司 | Method, system, equipment and medium for checking operation state of power grid station bus |
| CN115239033A (en) * | 2022-09-26 | 2022-10-25 | 广东电网有限责任公司东莞供电局 | Method for generating causal model under corresponding power grid operation environment |
| CN115396289A (en) * | 2022-08-31 | 2022-11-25 | 广东电网有限责任公司 | Fault alarm determination method and device, electronic equipment and storage medium |
| CN116342093A (en) * | 2022-12-20 | 2023-06-27 | 国网浙江省电力有限公司金华供电公司 | Power inspection method based on fault reliability analysis |
| CN116384980A (en) * | 2023-05-25 | 2023-07-04 | 杭州青橄榄网络技术有限公司 | Repair reporting method and system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102508059A (en) * | 2011-10-18 | 2012-06-20 | 北京科东电力控制系统有限责任公司 | Topology error-preventing checking method based on intelligent identification of state of equipment |
| CN104281982A (en) * | 2014-10-08 | 2015-01-14 | 广东电网有限责任公司茂名供电局 | Substation equipment state evaluation method based on power grid topological structure |
| CN106532964A (en) * | 2016-12-29 | 2017-03-22 | 国网浙江省电力公司温州供电公司 | Alarm device and system for remote control information of power grid equipment |
| CN110932881A (en) * | 2019-10-16 | 2020-03-27 | 国网浙江省电力有限公司温州供电公司 | Equipment portrait method based on monitoring alarm signal |
| CN111768076A (en) * | 2020-05-28 | 2020-10-13 | 国网浙江省电力有限公司绍兴供电公司 | Monitoring alarm signal clustering method taking power grid event as center |
-
2021
- 2021-04-06 CN CN202110369354.3A patent/CN113268590A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102508059A (en) * | 2011-10-18 | 2012-06-20 | 北京科东电力控制系统有限责任公司 | Topology error-preventing checking method based on intelligent identification of state of equipment |
| CN104281982A (en) * | 2014-10-08 | 2015-01-14 | 广东电网有限责任公司茂名供电局 | Substation equipment state evaluation method based on power grid topological structure |
| CN106532964A (en) * | 2016-12-29 | 2017-03-22 | 国网浙江省电力公司温州供电公司 | Alarm device and system for remote control information of power grid equipment |
| CN110932881A (en) * | 2019-10-16 | 2020-03-27 | 国网浙江省电力有限公司温州供电公司 | Equipment portrait method based on monitoring alarm signal |
| CN111768076A (en) * | 2020-05-28 | 2020-10-13 | 国网浙江省电力有限公司绍兴供电公司 | Monitoring alarm signal clustering method taking power grid event as center |
Non-Patent Citations (1)
| Title |
|---|
| 高志等: "设备监控信息大数据与设备模型的互校验及实用化事件分析", 《电力大数据》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114123192A (en) * | 2021-11-30 | 2022-03-01 | 广东电网有限责任公司 | Method, system, equipment and medium for checking operation state of power grid station bus |
| CN115396289A (en) * | 2022-08-31 | 2022-11-25 | 广东电网有限责任公司 | Fault alarm determination method and device, electronic equipment and storage medium |
| CN115396289B (en) * | 2022-08-31 | 2024-03-22 | 广东电网有限责任公司 | Fault alarm determining method and device, electronic equipment and storage medium |
| CN115239033A (en) * | 2022-09-26 | 2022-10-25 | 广东电网有限责任公司东莞供电局 | Method for generating causal model under corresponding power grid operation environment |
| CN115239033B (en) * | 2022-09-26 | 2023-02-10 | 广东电网有限责任公司东莞供电局 | Method for generating causal model under corresponding power grid operation environment |
| CN116342093A (en) * | 2022-12-20 | 2023-06-27 | 国网浙江省电力有限公司金华供电公司 | Power inspection method based on fault reliability analysis |
| CN116342093B (en) * | 2022-12-20 | 2023-12-01 | 国网浙江省电力有限公司金华供电公司 | Power inspection method based on fault reliability analysis |
| CN116384980A (en) * | 2023-05-25 | 2023-07-04 | 杭州青橄榄网络技术有限公司 | Repair reporting method and system |
| CN116384980B (en) * | 2023-05-25 | 2023-08-25 | 杭州青橄榄网络技术有限公司 | Repair reporting method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113268590A (en) | Power grid equipment running state evaluation method based on equipment portrait and integrated learning | |
| CN106372735B (en) | Relay protection state evaluation method | |
| CN103439593B (en) | Distributed power grid methods of risk assessment based on electric loop fault characteristic | |
| CN105678469A (en) | Risk assessment method for relay protection equipment in intelligent substation | |
| CN110879327B (en) | 10KV line monitoring method by multi-data fusion | |
| CN113221931B (en) | Electricity stealing prevention intelligent identification method based on electricity utilization information acquisition big data analysis | |
| CN103996099B (en) | Method for conducting intelligent evaluation on student operation in training system | |
| CN113589098B (en) | Power grid fault prediction and diagnosis method based on big data driving | |
| CN112116276A (en) | Transformer substation operation risk assessment method considering time-varying state of electrical main equipment | |
| CN117614137A (en) | Power distribution network optimization system based on multi-source data fusion | |
| CN116468243A (en) | Power grid trend early warning method based on monitoring operation data | |
| Zhang et al. | Research on power grid fault diagnosis based on a quantitative representation of alarm information | |
| Sun et al. | A multi-model-integration-based prediction methodology for the spatiotemporal distribution of vulnerabilities in integrated energy systems under the multi-type, imbalanced, and dependent input data scenarios | |
| Sun et al. | Transmission Line Fault Diagnosis Method Based on Improved Multiple SVM Model | |
| CN118035910A (en) | Novel power system-oriented power grid resource clustering early warning method, system, equipment and storage medium | |
| CN115224684A (en) | Intelligent power distribution network risk state identification method and system based on immune hazard theory | |
| Ren et al. | Research on causes of transmission line fault based on decision tree classification | |
| GU et al. | Research on intelligent early warning algorithm for distribution network considering extreme climate conditions | |
| CN116365714A (en) | Analysis method for monitoring false alarm of alarm signal | |
| CN113097981B (en) | Method for judging missing of monitoring alarm signal of transformer substation | |
| CN111313355B (en) | Method for updating monitoring signal event rule under manual supervision | |
| CN115877145A (en) | Transformer overload working condition big data cross evaluation system and method | |
| Li et al. | Nonparametric kernel density estimation model of transformer health based on dissolved gases in oil | |
| CN112581021A (en) | Method for automatically evaluating superiority and inferiority of operation scheme based on analytic hierarchy process | |
| Yang et al. | Research on transformer condition evaluation based on oil chromatography differentiation threshold |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210817 |