CN115239058A - Generator state detection evaluation method and system based on DCS data deep mining - Google Patents

Generator state detection evaluation method and system based on DCS data deep mining Download PDF

Info

Publication number
CN115239058A
CN115239058A CN202210570304.6A CN202210570304A CN115239058A CN 115239058 A CN115239058 A CN 115239058A CN 202210570304 A CN202210570304 A CN 202210570304A CN 115239058 A CN115239058 A CN 115239058A
Authority
CN
China
Prior art keywords
generator
data
state
dcs
evaluation
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
Application number
CN202210570304.6A
Other languages
Chinese (zh)
Inventor
杨玉磊
张树铭
王安保
王萍
陈延云
赵淼
张二龙
郑泽蔚
方瑞明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Datang Corp Science and Technology Research Institute Co Ltd
Datang Boiler Pressure Vessel Examination Center Co Ltd
East China Electric Power Test Institute of China Datang Corp Science and Technology Research Institute Co Ltd
Original Assignee
China Datang Corp Science and Technology Research Institute Co Ltd
Datang Boiler Pressure Vessel Examination Center Co Ltd
East China Electric Power Test Institute of China Datang Corp Science and Technology Research Institute Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by China Datang Corp Science and Technology Research Institute Co Ltd, Datang Boiler Pressure Vessel Examination Center Co Ltd, East China Electric Power Test Institute of China Datang Corp Science and Technology Research Institute Co Ltd filed Critical China Datang Corp Science and Technology Research Institute Co Ltd
Priority to CN202210570304.6A priority Critical patent/CN115239058A/en
Publication of CN115239058A publication Critical patent/CN115239058A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00—Error detection; Error correction; Monitoring
    • G06F11/30—Monitoring
    • G06F11/3065—Monitoring arrangements determined by the means or processing involved in reporting the monitored data
    • 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
    • G06Q10/20—Administration of product repair or maintenance
    • 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Electricity, gas or water supply

Abstract

The invention provides a method and a system for detecting and evaluating the state of a generator based on DCS data deep mining, wherein the method comprises the following steps: collecting generator DCS data of each power plant and mirroring the data to a server in real time; capturing generator monitoring data, unifying the format of the generator monitoring data and storing the generator monitoring data; setting not less than 2 generator state evaluation sub-modules for processing the generator monitoring data to obtain generator state detection evaluation data; acquiring early warning level data, processing the generator state detection evaluation data to obtain generator state detection level early warning data, and sending the generator state detection level early warning data to a generator state detection center of a preset classification region; historical working condition data of the generator are obtained from a generator state detection center of a preset classification area, and operation and maintenance scheme data of the generator are obtained through processing, so that operation and maintenance suggestion information is generated and output. The invention solves the technical problems of low applicability, dependence on manual operation and low operation, maintenance and overhaul efficiency of the generator.

Description

Generator state detection evaluation method and system based on DCS data deep mining
Technical Field
The invention relates to the field of state detection and diagnosis of electrical equipment, in particular to a method and a system for detecting and evaluating the state of a generator based on DCS data deep mining.
Background
The large-scale steam turbine generator is used as key equipment of a power plant, and the safe operation of the large-scale steam turbine generator plays a crucial role in actual production. Meanwhile, the normal and stable operation of the excitation system, the water-hydrogen cooling system, the air cooling system and other unit systems plays a decisive role in the safe operation of the unit. Therefore, it is important to enhance the management of the steam turbine unit, improve the safety, stability and life of the unit, perform effective state monitoring on the steam turbine unit, and diagnose or predict various abnormal states in time.
The thermal power industry is developing at a high speed, the complexity and the dynamic property of a steam turbine generator unit are gradually improved, the traditional steam turbine planned maintenance mode is not enough to deal with various abnormal and fault conditions encountered in the production process, and real-time monitoring and early warning cannot be achieved. Because the complexity of the steam turbine is increased, the dynamic performance of the operation condition of the steam turbine is enhanced, and it is difficult to establish an accurate steam turbine analytic mathematical model for fault analysis, the traditional pure mechanism modeling method is not suitable for use. In recent years, in order to meet the requirement of state monitoring, a turbo generator set is provided with hundreds of measuring points, a large amount of operation data is generated along with the development of a digital information era, each large power plant is advancing to a digital power plant, a novel information System such as a Distributed Control System (DCS) is popularized in each power plant, mass production data such as sensors and mobile terminals are stored, and research of intelligent diagnosis is also entered into the large data era along with the development of the digital information era. Massive operation data truly reflects the operation boundary, equipment characteristics and operation state of the unit, and a solid data foundation is provided for realizing unit operation optimization, abnormal detection, fault diagnosis and state maintenance. The method comprises the following steps of establishing an index system corresponding to each technical supervision project in the prior invention patent application document CN112580836A, classifying indexes in the index system, and determining corresponding data measuring points of each index in the thermal power generating unit; acquiring a real-time value of a data measuring point, and judging whether an event corresponding to the real-time value belongs to an equipment fault symptom event or not according to the real-time value of the data measuring point and a corresponding alarm criterion; if so, converting the event into an operation problem according to a pre-configured problem conversion rule, and counting the operation problem; and sending the counted running problems to a user, and receiving a running problem correction result fed back by the user. As is clear from the detailed description of the present patent application document, paragraph 1, the technical solution in this document is not based on DCS information system data, and this document does not disclose specific technical features of the multiple independent generator state evaluation submodules of the present invention, and this document mainly solves the technical problems of classifying each index and processing abnormality tracking, and unlike the technical problems solved by the present application, the technical effects of the present application cannot be achieved.
Although the steam turbine state monitoring technology is developed rapidly and applied to many power plants at home and abroad, the mainstream state monitoring technology and related software at present mainly monitor the running state of a steam turbine generator unit object in a form of a draft of an equipment monitoring system and a regular manual inspection, so that the steam turbine generator unit state monitoring system has two main defects: on one hand, the manual inspection method has large limitation and hysteresis, and generally can only find the obvious fault of the steam turbine equipment and cannot find and early warn in time at the initial stage of abnormity or fault occurrence; on the other hand, the method focuses more on monitoring of the signal level, and is lack of extracting and monitoring state information, so that the comprehensive and effective monitoring of the state of the steam turbine equipment cannot be carried out. As is known, deep peak shaving and variable load operation conditions faced by large steam turbine generators are increasingly common scenarios. Due to the influence of the flexible operation of the generator set, the turn-to-turn short circuit fault of the rotor is one of the major defects influencing the safe and stable operation of the large-scale turbonator.
As typical complex equipment, a steam turbine generator unit has complex working conditions and strong correlation among different operation data, frequently faces the risk of coexistence of multiple faults and multiple reasons, and is difficult to establish an accurate and perfect physical model; explosive dynamic data increase, and equipment tests, equipment information and the like are added, so that the intelligent diagnosis information data volume of the steam turbine generator unit is huge, and the equipment running state is difficult to be judged timely and accurately only by manpower. In addition, the state information of the unit equipment is distributed in the databases of each service system, and the problems of serious data heterogeneity, high redundancy, poor communication among different databases, difficult transverse sharing, difficult longitudinal communication and the like exist, so that the utilization rate of information and resources is insufficient, the difficulty of data integration and fusion analysis is high, and the efficiency and the effect of intelligent diagnosis of the unit are influenced. How to utilize mass steam turbine data and efficiently and accurately discover the value of the data becomes a research focus in the field of steam turbine state monitoring and fault diagnosis, and is also a research subject with important theoretical significance and economic value.
In summary, for safe and stable operation of large-sized steam turbine units of various power generation enterprises, research on key technologies of state monitoring and intelligent operation and maintenance of the steam turbine units, especially research on identification technologies of abnormal disturbance and early failure states affecting safety and stability of the steam turbine units, is urgently needed. The prior art has the technical problems of low applicability, dependence on manual operation and low maintenance efficiency of the generator.
Disclosure of Invention
The invention aims to solve the technical problems of low applicability, dependence on manual operation and low operation and maintenance efficiency of a generator in the prior art.
The invention adopts the following technical scheme to solve the technical problems: the generator state detection evaluation method based on DCS data deep mining comprises the following steps:
s1, collecting generator DCS data of each power plant by using a preset DCS information system, and mirroring the generator DCS data to a server in real time;
s2, capturing generator monitoring data from the server, unifying the format of the generator monitoring data and storing the generator monitoring data to a preset storage device;
s3, setting at least 2 generator state evaluation sub-modules by utilizing a preset extensible evaluation algorithm, and processing the generator monitoring data to obtain generator state detection evaluation data, wherein the generator state detection evaluation data comprises: rotor inter-turn insulation state data, electronic internal humidity data and generator state data;
s4, acquiring early warning level data, processing the generator state detection evaluation data to obtain generator state detection level early warning data, and sending the generator state detection level early warning data to a generator state detection center of a preset classification region;
and S5, acquiring historical working condition data of the generator from the generator state detection center of the preset classification area, and extracting operation and maintenance scheme data of the generator according to an obtained automatic early warning and forecasting mechanism so as to generate and output operation and maintenance recommendation information.
The invention establishes the service for ensuring and promoting the healthy, stable and efficient operation of the turbo generator set, thereby improving the operation and monitoring level of the set, optimizing the existing operation and preventive maintenance system, ensuring the safe and reliable operation of the system, reducing the whole life cycle cost and the failure rate of the system, and meeting the development requirement of the predicted maintenance, thereby realizing the implementation of large-scale energy resource optimization configuration in the system by using the constraint conditions of least consumed resources, least environmental cost and best economic benefit.
In a more specific technical solution, the step S1 includes:
s11, distributing DCS acquisition ends of the DCS information system to generator sets of the power plants;
s12, collecting the DCS data of the generator from the generator set by using the DCS collecting end;
and S13, transmitting the generator DCS data to the server in a mirror image mode.
According to the invention, a large data deep mining thinking mode is adopted, and based on the mass monitoring data of a DCS (distributed control system) of a large-scale turbonator, four organic layers of modeling around the space-time dynamic behavior of the generator, critical characteristic characterization, quantitative evaluation and defect early warning are researched to research the key technology related to the state monitoring and intelligent operation and maintenance of the generator, so that the final aim is to establish a multi-dimensional full-situation perception system of the generator based on the DCS data deep mining, and the state detection and health evaluation cloud platform of the generator at a power plant level or a region level is realized.
In a more specific embodiment, the generator DCS data in step S12 includes: generating equipment data, power plant data and generator DCS real-time image data.
In a more specific technical solution, the step S2 includes:
s21, acquiring the generator DCS data from the server through a preset network;
s22, extracting the relevant data of the generator running state from the DCS data of the generator by using a data capture mode;
s23, carrying out normalization processing on the relevant data of the running state of the generator to obtain normalized generator state data;
and S24, storing the normalized motor state data into the preset storage equipment for the calling of the generator state evaluation submodule.
In a more specific technical solution, the preset storage device in step S2 is provided in a local server.
In a more specific technical solution, the step S3 includes:
s31, acquiring space-time dynamic data of the generator, obtaining a generator state evaluation submodule according to modeling, and representing a key state variable of the generator at any time i by the following observation vector matrix:
X(i)＝[x 1i x 2i …x ni ] T ；
s32, extracting historical monitoring data from the generator monitoring data, wherein generators corresponding to the generator monitoring data are independent of one another;
s33, processing the historical monitoring data by each generator state evaluation submodule according to preset evaluation logic to obtain the generator state detection evaluation data, wherein the step S33 further comprises the following steps:
s331, collecting the operation data of the generator by using a preset sliding time window, and forming a random matrix by using historical data with the length of T-1 before the current time in the operation data:
X N×T ＝[X(t i-T+1 )...X(t i )]
s332, processing the original matrix in the random matrix by utilizing the following logic standardization
Figure BDA0003659994030000041
To obtain hermitian matrix of NxT dimensions
Figure BDA0003659994030000042
Figure BDA0003659994030000043
In the formula:
Figure BDA0003659994030000044
is a matrix
Figure BDA0003659994030000045
The standard deviation of (a);
Figure BDA0003659994030000046
is a matrix
Figure BDA0003659994030000047
The mean value of (a);
Figure BDA0003659994030000048
is a matrix
Figure BDA0003659994030000049
Standard deviation of
Figure BDA00036599940300000410
Is a matrix
Figure BDA00036599940300000411
Is and
Figure BDA00036599940300000412
1≤i≤N，1≤j≤T；
and S333, obtaining the generator state detection evaluation data according to the Hermite matrix.
The invention characterizes and combines the mass DCS data generated when the generators with different types and different capacities operate through a high-dimensional random matrix theory, so that the processed data is highly independent of the actual characteristic information of the generator set, the defect that the traditional expert diagnosis algorithm or system cannot have universality and cloud monitoring due to the fact that one machine is adopted or the same machine is adopted is avoided, and a unified data basis is provided for the extraction of the evolution characteristics of the physical layer of the turbonator, the identification of the state and the like by adopting a digital algorithm and an intelligent tool subsequently.
In a more specific technical solution, the generator state evaluation submodule in step S31 includes: the system comprises a rotor inter-turn insulation evaluation submodule, a stator internal temperature early warning submodule and a generator running state evaluation submodule, wherein the generator state evaluation submodule is an expandable module.
In a more specific technical solution, the preset evaluation logic of the generator state evaluation submodule in step S31 includes: a self-optimizing algorithm, a self-evolving algorithm, and a self-learning algorithm, wherein the step S33 further includes:
s311, based on the normal historical operation data of each state variable acquired by the DCS of the generator, normalizing the data of each state variable by using the following logic:
Figure BDA0003659994030000051
wherein x is the original data, x max Is the maximum value, x, of the set of state variables min Is the minimum value of the set of state variables, and x' is the normalized data;
s312, calculating a gray correlation coefficient of each state variable and the monitored physical quantity, wherein the calculation formula is as follows:
Figure BDA0003659994030000052
wherein x is 0 As a reference state variable, x i Rho is a resolution coefficient and is a value between (0 and 1) for a state variable of the degree of association to be solved;
s313, calculating the gray correlation degree according to the gray correlation coefficient by the following logic:
Figure BDA0003659994030000053
wherein r is i The grey correlation degree is obtained;
and S314, screening the state variables with higher grey correlation degrees as the key state variables.
The motor state evaluation submodule modules of the invention are self-optimizing, self-evolving and self-learning aiming at the self historical monitoring data of any generator, and different generators are mutually independent, so the invention naturally has a cloud evaluation function, solves the problem that the traditional expert system depends on a large number of accident cases and can not be used universally among generators with different types and different capacities, and is easy to carry out universal monitoring and state evaluation on different types of generator sets of different power generation enterprises.
In a more specific technical solution, the step S4 includes:
s41, performing unified evaluation analysis on the generator state detection evaluation data to obtain generator unified evaluation data;
s42, acquiring the real-time data of the generator DCS, processing the real-time data of the generator DCS through preset logic, and accordingly obtaining a change trend of the real-time data of the DCS, wherein the step S42 comprises the following steps:
s421, detecting the real-time running state of the generator by using a preset region level generator detection center;
s422, accessing an SIS system of the power plant to acquire real-time data of the generator DCS;
s423, performing characterization combination according to the multi-information-source heterogeneous data source, the monitoring data and the typical scene simulation data in the real-time DCS data of the generator by using a big data characterization method to construct a unit state characterization high-dimensional random matrix;
s424, analyzing the spectral distribution and the ring rate of the characteristic vector in the space-time evolution process according to the high-dimensional random matrix to obtain the change trend of the DCS real-time data;
s43, combining and processing the DCS real-time data change trend and the generator unified evaluation data to obtain historical working condition data of the generator;
and S44, acquiring generator defect state data and accident grade prediction data according to the generator historical working condition data, setting and realizing three-level early warning evaluation on the generator state according to 3 labels of general early warning, serious early warning and special serious early warning.
The invention relates the information mining of abnormal monitoring parameters, the state evaluation of the abnormal development trend of the unit, the unit fault risk and the monitoring parameter change, and establishes a turbo generator unit state monitoring and defect early warning technical system which aims at fault monitoring and realizes continuous monitoring of equipment faults on the basis of characteristic parameter monitoring.
The invention realizes three-level early warning evaluation on the state of the generator, reminds the area-level or plant-level generator state detection center to pay attention to the continuous change of the abnormal parameters of the equipment, prepares for maintenance or defect elimination in advance and provides suggestions for the subsequent operation and maintenance of the generator.
In a more specific technical scheme, the generator state detection and evaluation system based on DCS data deep mining comprises:
the DCS information module is used for acquiring generator DCS data of each power plant by using a preset DCS information system and mirroring the generator DCS data to a server in real time;
the generator monitoring data storage module is used for capturing generator monitoring data from the server, unifying the format of the generator monitoring data and storing the generator monitoring data to preset storage equipment, and is connected with the DCS information module;
the generator state detection and evaluation processing module is used for setting at least 2 generator state evaluation submodules by utilizing a preset extensible evaluation algorithm and processing the generator monitoring data to obtain generator state detection and evaluation data, wherein the generator state detection and evaluation data comprise: the rotor inter-turn insulation state data, the electronic internal humidity data and the generator state data are connected with the generator monitoring data storage module through the generator state detection evaluation processing module;
the system comprises a classification region generator detection data module, a generator state detection evaluation processing module, a classification region generator detection data module and a generator state detection evaluation processing module, wherein the classification region generator detection data module is used for acquiring early warning level data, processing the generator state detection evaluation data to obtain generator state detection level early warning data, and sending the generator state detection level early warning data to a preset classification region generator state detection center;
and the operation and maintenance module is used for acquiring historical working condition data of the generator from the preset classification region generator state detection center, processing the historical working condition data to obtain operation and maintenance scheme data of the generator, and generating and outputting operation and maintenance suggestion information, and is connected with the classification region generator detection data module.
Compared with the prior art, the invention has the following advantages: the invention establishes the service for ensuring and promoting the healthy, stable and efficient operation of the turbo generator set, thereby improving the operation and monitoring level of the set, optimizing the existing operation and preventive maintenance system, ensuring the safe and reliable operation of the system, reducing the whole life cycle cost and the failure rate of the system, and meeting the development requirement of the predicted maintenance, thereby realizing the implementation of large-scale energy resource optimization configuration in the system by using the constraint conditions of least consumed resources, least environmental cost and best economic benefit.
According to the invention, a large data deep mining thinking mode is adopted, and based on the mass monitoring data of a DCS (distributed control system) of a large-scale turbonator, four organic layers of modeling around the space-time dynamic behavior of the generator, critical characteristic characterization, quantitative evaluation and defect early warning are researched to research the key technology related to the state monitoring and intelligent operation and maintenance of the generator, so that the final aim is to establish a multi-dimensional full-situation perception system of the generator based on the DCS data deep mining, and the state detection and health evaluation cloud platform of the generator at a power plant level or a region level is realized.
The invention characterizes and combines the mass DCS data generated when the generators with different types and different capacities operate through a high-dimensional random matrix theory, so that the processed data is highly independent of the actual characteristic information of the generator set, the defect that the traditional expert diagnosis algorithm or system cannot have universality and cloud monitoring due to the fact that one machine is adopted or the same machine is adopted is avoided, and a unified data basis is provided for the extraction of the evolution characteristics of the physical layer of the turbonator, the identification of the state and the like by adopting a digital algorithm and an intelligent tool subsequently.
The motor state evaluation submodule modules of the invention are self-optimizing, self-evolving and self-learning aiming at the self historical monitoring data of any generator, and different generators are mutually independent, so the invention naturally has a cloud evaluation function, solves the problem that the traditional expert system depends on a large number of accident cases and can not be used universally among generators with different types and different capacities, and is easy to carry out universal monitoring and state evaluation on different types of generator sets of different power generation enterprises.
The invention relates the information mining of abnormal monitoring parameters, the state evaluation of the abnormal development trend of the unit, the unit fault risk and the monitoring parameter change, and establishes a turbo generator unit state monitoring and defect early warning technical system which aims at fault monitoring and realizes continuous monitoring of equipment faults on the basis of characteristic parameter monitoring.
The invention realizes three-level early warning evaluation on the state of the generator, reminds the area-level or plant-level generator state detection center to pay attention to the continuous change of the abnormal parameters of the equipment, prepares for maintenance or defect elimination in advance and provides suggestions for the subsequent operation and maintenance of the generator. The problem of among the prior art exist the suitability low, rely on manual operation and generator operation and maintenance to overhaul technical problem with low efficiency is solved.
Drawings
Fig. 1 is a schematic flow diagram of a method for detecting and evaluating a state of a generator based on DCS data deep mining in embodiment 1 of the present invention;
fig. 2 is a schematic diagram of a power generator state detection and health evaluation cloud platform based on DCS data deep mining according to embodiment 1 of the present invention;
fig. 3 is a schematic diagram of a data processing step of a turn-to-turn insulation diagnosis algorithm module of a generator rotor in embodiment 2 of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention, and it is obvious that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example 1
As shown in fig. 1, the method for detecting and evaluating the state of a generator based on DCS data deep mining of the present invention includes:
s1, collecting generator DCS data of each power plant by using a preset DCS information system, and mirroring the generator DCS data to a server in real time; the invention can realize the combing and the mining of mass DCS data generated when the generator operates, and can timely find the trend degradation phenomena in the aspects of electricity, temperature, pressure, vibration and the like which possibly exist.
S2, capturing generator monitoring data from the server, unifying the format of the generator monitoring data and storing the generator monitoring data to a preset storage device; in this embodiment, since the DCS real-time screen data is transmitted in a mirror image manner, the operating data cannot be directly obtained, and the DCS systems of the power generation enterprises are different from one another. According to the invention, all data related to the running state of the generator are extracted in a data capture mode, and the data are stored in the local server after normalization and unification processing, so that the data can be called by the subsequent state evaluation submodule.
S3, setting not less than 2 generator state evaluation sub-modules by using a preset extensible evaluation algorithm, and processing the generator monitoring data to obtain generator state detection evaluation data, wherein the generator state detection evaluation data comprises: rotor inter-turn insulation state data, electronic internal humidity data and generator state data; in this embodiment, after data acquisition is completed, modeling is performed around the time-space dynamics behavior of the generator based on the mass monitoring data of the DCS system of the large-scale steam turbine generator, and various generator state evaluation sub-modules are developed, such as a rotor inter-turn insulation evaluation module, a stator internal temperature early warning module, a generator state evaluation module, and the like. The module carries out self-optimization, self-evolution and self-learning aiming at the self historical monitoring data of any generator, and different generators are mutually independent, so that the module naturally has a cloud evaluation function, solves the problem that the traditional expert system depends on a large number of accident cases and cannot be used universally among generators with different types and different capacities, is easy to carry out universal monitoring and state evaluation on different types of generator sets of different power generation enterprises, and is a core technology of a project.
S4, acquiring early warning level data, processing the generator state detection evaluation data to obtain generator state detection level early warning data, and sending the generator state detection level early warning data to a generator state detection center of a preset classification region;
and S5, acquiring historical working condition data of the generator from the generator state detection center of the preset classification area, and extracting operation and maintenance scheme data of the generator according to an automatic early warning and forecasting mechanism obtained by processing so as to generate and output operation and maintenance suggestion information.
In the embodiment, according to the unified evaluation and analysis of the in-service generator state evaluation module, the DCS real-time data change trend is combined, according to the development speed of the generator equipment defect or adverse trend and the severity of the generator damage accident, the three-level early warning evaluation of the generator state is realized according to 3 labels of general early warning, serious early warning and special serious early warning, the region-level or plant-level generator state detection center is reminded to pay attention to the continuous change of the abnormal parameters of the equipment, the preparation for maintenance or elimination is made in advance, and the suggestion is provided for the subsequent operation and maintenance of the generator.
Example 2
In this embodiment, taking the generator rotor inter-turn insulation diagnosis algorithm module as an example, the following description is given:
whether a short-circuit fault exists between turns of the rotor can be judged by adopting an exciting current amplification detection method. After the rotor has turn-to-turn short circuit fault, under the influence of a loop formed by short-circuit point current, under the same operating condition (active power and reactive power are similar), the exciting current after the fault is obviously increased compared with that before the fault. Therefore, the excitation current of the generator is calculated in real time through the relevant electrical operation parameters, and the calculated value of the excitation current is compared with the actual value, so that the early turn-to-turn short circuit defect of the rotor can be found.
According to the above diagnostic method, the embodiments of this patent are as follows:
step S1': and selecting related electric quantities which possibly influence the excitation current of the generator by utilizing the established regional generator monitoring center. According to the mapping relation between the on-line monitoring time sequence data of the operation of the generator and the exciting current, the grey correlation degree of each state variable and the exciting current is calculated through a grey correlation degree analysis method, and the key state variable capable of representing the fault is screened out.
Step S2': and according to the screened key state variables, using normal historical operating data acquired by the generator to be diagnosed under different working conditions in the operating process to construct a memory matrix of the NSET model by a method combining equidistant sampling and segmented sampling.
Step S3': inputting the test set into the prediction model established in the step 2, and taking the Manhattan distance as a nonlinear operator of the model to complete the prediction of the excitation current.
Step S4': and comparing the predicted value of the exciting current with the actual running exciting current value of the generator, and judging the turn-to-turn insulation state of the rotor.
In this embodiment, the generator state evaluation submodule is obtained by modeling by using the space-time dynamics data of the generator;
by means of a big data characterization method, aiming at the current situations that a multi-information-source heterogeneous data source exists in a DCS (distributed control system) of a large-scale steam turbine generator unit and massive monitoring data and typical scene simulation data are involved in research, a high-dimensional random matrix theory is utilized to characterize and combine time sequences of massive data of the DCS, a high-dimensional random matrix for dynamically characterizing the running state of the unit is constructed, the spectral distribution and the ring rate of feature vector data in the time-space evolution process are analyzed, and the variation trend of key performance of the unit is researched. The mass DCS data generated when the generators of different types and different capacities operate are characterized and combined through a high-dimensional random matrix theory, so that the processed data is highly independent of the actual characteristic information of the generator set, the defect that the traditional expert diagnosis algorithm or system cannot have universality and cloud monitoring due to the fact that one machine and one strategy or the same machine and the same strategy are adopted is avoided, and a unified data basis is provided for the extraction of the evolution characteristics of the physical layer of the turbonator, the identification of the state and the like by adopting a digital algorithm and an intelligent tool subsequently.
In this embodiment, a generator in the power plant generating set to be monitored is taken, n different state variables are in the generator, and at any time i, the observation of the n different state variables can be represented as an observation vector matrix.
X(i)＝[x 1i x 2i …x ni ] T
In order to facilitate real-time calculation, a sliding time window is used for collecting data of the generator in operation, the collection time length, namely the width of each sampling window is T, and when data at the moment i are collected, historical data with the length of T-1 before the moment are fully utilized to form a random matrix. The N state variables intercepted by the time window sliding to time i form a random matrix with spatio-temporal characteristics. And (4) adopting a sliding time window, moving a time point backwards after sampling every time, and monitoring the dynamic process of state quantity change.
X N×T ＝[X(t i-T+1 )...X(t i )]
For any one NxT original matrix
Figure BDA0003659994030000101
The treatment process is as follows: less than or equal to
First, an original matrix of N × T is expressed by
Figure BDA0003659994030000111
And (5) carrying out standardization processing.
Figure BDA0003659994030000112
In the formula:
Figure BDA0003659994030000113
is a matrix
Figure BDA0003659994030000114
Standard deviation of (d);
Figure BDA0003659994030000115
is a matrix
Figure BDA0003659994030000116
The mean value of (a);
Figure BDA0003659994030000117
is a matrix
Figure BDA0003659994030000118
Standard deviation of
Figure BDA0003659994030000119
Is a matrix
Figure BDA00036599940300001110
Is and
Figure BDA00036599940300001111
i is more than or equal to 1 and less than or equal to N, and j is more than or equal to 1 and less than or equal to T. The processed matrix becomes a Hermite matrix with NxT dimensions
Figure BDA00036599940300001112
In the embodiment, the method is closely combined with basic theories and methods such as complex networks, mathematical statistics, association analysis and the like, and is used for counting and comparing topological structure attributes of a physical layer of a unit, key performance of the unit and a time-space association mechanism of a monitoring information layer aiming at mass monitoring data of a DCS (distributed control system) of the unit, and analyzing and extracting a measure form and a representation method of the time-space association relationship based on a dynamic grey association method.
Based on normal historical operation data of each state variable acquired by a DCS (distributed control system) of the generator, the state variable comprises but is not limited to parameters such as a timestamp, stator current, stator voltage, excitation current, excitation voltage, active power, reactive power and vibration displacement, and the acquired state variable data is normalized, and the calculation formula is as follows:
Figure BDA00036599940300001113
wherein x is the original data, x max Is the maximum value, x, of the set of state variables min Is the minimum value of the set of state variables, and x' is the normalized data. Calculating the grey correlation coefficient of each state variable and the monitored physical quantity, wherein the calculation formula is as follows:
Figure BDA00036599940300001114
wherein x is 0 As a reference state variable, x i Rho is a resolution coefficient for the state variable of the degree of association to be solved, and is taken as a value between (0 and 1), wherein the smaller rho is, the stronger the distinguishing capability is.
And calculating the association degree according to the following calculation formula:
Figure BDA00036599940300001115
wherein r is i Is the degree of association.
And screening the state variable with higher grey correlation degree as a key state variable. By means of an artificial intelligence algorithm, aiming at the requirements of dynamic evaluation under a dynamic operation specific scene, according to data of a time space section corresponding to a unit DCS, a support vector machine algorithm is adopted to predict the evolution trend of a state characteristic vector, a dynamic reference sequence is established, dynamic network marks are screened, panoramic multi-dimensional dynamic quantitative evaluation is carried out on the operation state of the unit, early-stage defects are early warned in time, fault depth analysis and the influence on reliability are further carried out, and technical support is improved for realizing a unit predictive overhaul technology.
In conclusion, the invention establishes the service for ensuring and promoting the healthy, stable and efficient operation of the steam turbine generator unit, thereby improving the operation and monitoring level of the unit, optimizing the existing operation and preventive maintenance system, ensuring the safe and reliable operation of the system, reducing the whole life cycle cost and the failure rate, and meeting the development requirement of the predicted maintenance, thereby realizing the implementation of large-scale energy resource optimal configuration in the system by using the constraint conditions of minimum consumption resources, minimum environmental cost and best economic benefit.
According to the invention, a big data deep excavation thinking mode is adopted, based on the mass monitoring data of the DCS of the large-scale steam turbine generator, four organic layers of modeling around the space-time dynamic behavior of the generator, critical characteristic characterization, quantitative evaluation and defect early warning are researched, the key technology related to the state monitoring and intelligent operation and maintenance of the generator is researched, and the final aim is to establish a generator multi-dimensional full-situation sensing system based on DCS data deep excavation, so that the state detection and health evaluation cloud platform of the generator at a power plant level or a regional level is realized.
The motor state evaluation submodule modules of the invention are self-optimizing, self-evolving and self-learning aiming at the self historical monitoring data of any generator, and different generators are mutually independent, so the invention naturally has a cloud evaluation function, solves the problem that the traditional expert system depends on a large number of accident cases and can not be used universally among generators with different types and different capacities, and is easy to carry out universal monitoring and state evaluation on different types of generator sets of different power generation enterprises.
The invention characterizes and combines the mass DCS data generated when the generators with different types and different capacities operate through a high-dimensional random matrix theory, so that the processed data is highly independent of the actual characteristic information of the generator set, the defect that the traditional expert diagnosis algorithm or system cannot have universality and cloud monitoring due to the fact that one machine is adopted or the same machine is adopted is avoided, and a unified data basis is provided for the extraction of the evolution characteristics of the physical layer of the turbonator, the identification of the state and the like by adopting a digital algorithm and an intelligent tool subsequently.
The invention relates the information mining of abnormal monitoring parameters, the state evaluation of the abnormal development trend of the unit, the unit fault risk and the monitoring parameter change, and establishes a turbo generator unit state monitoring and defect early warning technical system which aims at fault monitoring and realizes continuous monitoring of equipment faults on the basis of characteristic parameter monitoring.
The invention realizes three-level early warning evaluation on the state of the generator, reminds the area-level or plant-level generator state detection center to pay attention to the continuous change of the abnormal parameters of the equipment, prepares for maintenance or defect elimination in advance and provides suggestions for the subsequent operation and maintenance of the generator. The problem of among the prior art exist the suitability low, rely on manual operation and generator operation and maintenance to overhaul technical problem with low efficiency is solved.
The above examples are only intended to illustrate the technical solution of the present invention, and not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims (10)

1. The method for detecting and evaluating the state of the generator based on DCS data deep mining is characterized by comprising the following steps:
s1, collecting generator DCS data of each power plant by using a preset DCS information system, and mirroring the generator DCS data to a server in real time;
s2, capturing generator monitoring data from the server, unifying the format of the generator monitoring data and storing the generator monitoring data to a preset storage device;
s3, setting at least 2 generator state evaluation sub-modules by utilizing a preset extensible evaluation algorithm, and processing the generator monitoring data to obtain generator state detection evaluation data, wherein the generator state detection evaluation data comprises: rotor inter-turn insulation state data, electronic internal humidity data and generator state data;
s4, acquiring early warning level data, processing the generator state detection evaluation data to obtain generator state detection level early warning data, and sending the generator state detection level early warning data to a generator state detection center of a preset classification region;
and S5, acquiring historical working condition data of the generator from the generator state detection center of the preset classification area, and processing the historical working condition data to obtain operation and maintenance scheme data of the generator so as to generate and output operation and maintenance suggestion information.
2. The method for detecting and evaluating the state of the generator based on DCS data deep mining of claim 1, wherein the step S1 comprises:
s11, distributing DCS acquisition ends of the DCS information system to generator sets of the power plants;
s12, collecting the DCS data of the generator from the generator set by using the DCS collecting end;
and S13, transmitting the generator DCS data to the server in a mirror image mode.
3. The method of claim 2, wherein the generator DCS data in the step S12 comprises: the data of the power generation equipment, the data of the power plant and the data of the real-time images of the generator DCS.
4. The method for detecting and evaluating the state of the generator based on DCS data deep mining of claim 1, wherein the step S2 comprises:
s21, acquiring the generator DCS data from the server through a preset network;
s22, extracting the relevant data of the generator running state from the DCS data of the generator by using a data capture mode;
s23, carrying out normalization processing on the generator running state related data to obtain normalized generator state data;
and S24, storing the normalized motor state data into the preset storage equipment for the calling of the generator state evaluation submodule.
5. The method of claim 4, wherein the preset storage device in step S2 is provided in a local server.
6. The method of claim 1, wherein the step S3 comprises:
s31, acquiring space-time dynamic data of the generator, obtaining a generator state evaluation submodule according to modeling, and representing a key state variable of the generator at any time i by the following observation vector matrix:
X(i)＝[x 1i x 2i …x ni ] T ；
s32, extracting historical monitoring data from the generator monitoring data, wherein generators corresponding to the generator monitoring data are independent of one another;
s33, processing the historical monitoring data by each generator state evaluation submodule according to preset evaluation logic to obtain the generator state detection evaluation data, wherein the step S33 further comprises the following steps:
s331, collecting operation data of the generator by using a preset sliding time window, and forming a random matrix by using historical data with the length of T-1 before the current time in the operation data:
X N×T ＝[X(t i-T+1 )...X(t i )]
s332, processing the original matrix in the random matrix by using the following logic standardization
Figure FDA0003659994020000021
To obtain hermitian matrix of NxT dimensions
Figure FDA0003659994020000022
Figure FDA0003659994020000023
In the formula:
Figure FDA0003659994020000024
Figure FDA0003659994020000025
is a matrix
Figure FDA0003659994020000026
The standard deviation of (a);
Figure FDA0003659994020000027
is a matrix
Figure FDA0003659994020000028
The mean value of (a);
Figure FDA0003659994020000029
is a matrix
Figure FDA00036599940200000210
Standard deviation of
Figure FDA00036599940200000211
Figure FDA00036599940200000212
Is a matrix
Figure FDA00036599940200000213
Mean value of
Figure FDA00036599940200000214
And S333, obtaining the generator state detection evaluation data according to the Hermite matrix.
7. The method of claim 6, wherein the generator state evaluation submodule in step S31 comprises: the device comprises a rotor inter-turn insulation evaluation submodule, a stator internal temperature early warning submodule and a generator running state evaluation submodule, wherein the generator state evaluation submodule is an expandable module.
8. The method of claim 6, wherein the preset evaluation logic of the generator status evaluation submodule in step S33 includes: a self-optimizing algorithm, a self-evolving algorithm, and a self-learning algorithm, wherein the step S33 further includes:
s311, based on the normal historical operation data of each state variable acquired by the DCS of the generator, normalizing the data of each state variable by using the following logic:
Figure FDA0003659994020000031
wherein x is the original data, x max Is the maximum value, x, of the set of state variables min Is the minimum value of the set of state variables, and x' is the normalized data;
s312, calculating a gray correlation coefficient of each state variable and the monitored physical quantity, wherein the calculation formula is as follows:
Figure FDA0003659994020000032
wherein x is 0 As a reference state variable, x i Rho is a resolution coefficient and is a value between (0 and 1) for a state variable of the degree of association to be solved;
s313, calculating the gray correlation degree according to the gray correlation coefficient by the following logic:
Figure FDA0003659994020000033
wherein r is i The grey correlation degree is obtained;
and S314, screening the state variables with higher grey correlation degrees as the key state variables.
9. The method of claim 1, wherein the step S4 comprises:
s41, performing unified evaluation analysis on the generator state detection evaluation data to obtain generator unified evaluation data;
s42, acquiring real-time data of the generator DCS, processing the real-time data of the generator DCS through preset logic, and accordingly obtaining change trend of the real-time data of the DCS, wherein the step S42 comprises the following steps:
s421, detecting the real-time running state of the generator by using a preset region level generator detection center;
s422, accessing an SIS system of the power plant to acquire real-time data of the generator DCS;
s423, performing characterization combination according to a multi-source heterogeneous data source, monitoring data and typical scene simulation data in the real-time DCS data of the generator by using a big data characterization method to construct a unit state characterization high-dimensional random matrix;
s424, analyzing the spectral distribution and the ring rate of the characteristic vector in the space-time evolution process according to the high-dimensional random matrix to obtain the change trend of the DCS real-time data;
s43, combining and processing the DCS real-time data change trend and the generator unified evaluation data to obtain historical working condition data of the generator;
and S44, acquiring generator defect state data and accident grade prediction data according to the generator historical working condition data, setting and realizing three-level early warning evaluation on the generator state according to 3 labels of general early warning, serious early warning and special serious early warning.
10. Generator state detection evaluation system based on DCS data degree of depth excavation, characterized in that, the system includes:
the DCS information module is used for acquiring generator DCS data of each power plant by using a preset DCS information system and mirroring the generator DCS data to a server in real time;
the generator monitoring data storage module is used for capturing generator monitoring data from the server, unifying the format of the generator monitoring data and storing the generator monitoring data to preset storage equipment, and is connected with the DCS information module;
the generator state detection and evaluation processing module is used for setting at least 2 generator state evaluation submodules by utilizing a preset extensible evaluation algorithm and processing the generator monitoring data to obtain generator state detection and evaluation data, wherein the generator state detection and evaluation data comprise: the rotor inter-turn insulation state data, the electronic internal humidity data and the generator state data are connected with the generator monitoring data storage module through the generator state detection evaluation processing module;
the classified region generator detection data module is used for acquiring early warning level data, processing the generator state detection evaluation data to obtain generator state detection level early warning data, and sending the generator state detection level early warning data to a preset classified region generator state detection center, and the classified region generator detection data module is connected with the generator state detection evaluation processing module;
and the operation and maintenance module is used for acquiring historical working condition data of the generator from the preset classification region generator state detection center, processing the historical working condition data to obtain operation and maintenance scheme data of the generator, and generating and outputting operation and maintenance suggestion information, and is connected with the classification region generator detection data module.
CN202210570304.6A 2022-05-24 2022-05-24 Generator state detection evaluation method and system based on DCS data deep mining Pending CN115239058A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210570304.6A CN115239058A (en) 2022-05-24 2022-05-24 Generator state detection evaluation method and system based on DCS data deep mining

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210570304.6A CN115239058A (en) 2022-05-24 2022-05-24 Generator state detection evaluation method and system based on DCS data deep mining

Publications (1)

Publication Number Publication Date
CN115239058A true CN115239058A (en) 2022-10-25

Family

ID=83667675

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210570304.6A Pending CN115239058A (en) 2022-05-24 2022-05-24 Generator state detection evaluation method and system based on DCS data deep mining

Country Status (1)

Country Link
CN (1) CN115239058A (en)

Similar Documents

Publication Publication Date Title
CN109146093B (en) Power equipment field investigation method based on learning
Zaher et al. Online wind turbine fault detection through automated SCADA data analysis
EP3902992A1 (en) Scalable system and engine for forecasting wind turbine failure
CN112803592B (en) Intelligent fault early warning method and system suitable for distributed power station
CN108627720B (en) Power equipment state monitoring method based on Bayesian algorithm
CN115048591B (en) Visual intelligent display analysis system of distribution network holographic data based on artificial intelligence
CN112947290B (en) Edge cloud cooperation-based equipment state monitoring method and system and storage medium
CN116069079B (en) Intelligent heat dissipation control method and system for intelligent switch cabinet
CN116308304B (en) New energy intelligent operation and maintenance method and system based on meta learning concept drift detection
CN111209934A (en) Fan fault prediction and alarm method and system
CN115563873A (en) Digital twin simulation system and method of power network
CN111555776B (en) Fusion sensing and joint diagnosis method, system and device for power transmission line
CN114755515A (en) Energy storage power station key equipment fault diagnosis and early warning method based on data mining
CN116418117A (en) Equipment detection system for intelligent power grid
CN117289085A (en) Multi-line fault analysis and diagnosis method and system
CN116800199A (en) Distributed photovoltaic operation electric energy quality monitoring and analyzing method
CN115313930A (en) Method for predicting excitation current of large-scale steam turbine generator in real time
CN111929579B (en) Generator online fault diagnosis method and device and computer device
CN117390944A (en) Substation operation condition simulation system
CN112769235A (en) Index monitoring system and method for power distribution network grid
CN116644291A (en) Energy storage battery multi-risk type early warning method based on cloud edge cooperation
CN115239058A (en) Generator state detection evaluation method and system based on DCS data deep mining
CN116221037A (en) Wind turbine generator monitoring method and device
CN111539642B (en) Object-oriented power plant data acquisition and processing system and method thereof
CN114442543A (en) Computer monitoring method suitable for early warning of hydropower station fault

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