Intelligent analysis system and method for abnormity of power equipment based on Internet of Things

Abstract

The invention discloses a power equipment abnormity intelligent analysis system and method based on the Internet of Things, and belongs to the technical field of power equipment abnormity analysis. The method comprises the following steps: S10: constructing a power plant data logic graph and an abnormal case graph; s20, analyzing the abnormal condition of the real-time operation data of the power plant; s30, determining a target abnormal case in the abnormal case graph; and S40, updating the atlas. Through the power plant data logic graph, the abnormal operation data can be rapidly judged, whether the operation data has an abnormal trend is analyzed, the abnormal power equipment can be effectively identified before the power equipment is abnormal, and the abnormal power equipment can be effectively identified when the power equipment is abnormal. And according to the matching degree between the abnormal operation data of the power equipment and the abnormal operation data recorded in each abnormal case, the target abnormal case is searched, so that maintenance personnel can quickly determine an abnormal processing scheme, and the abnormal analysis effect of the system is further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment abnormality analysis, and in particular to an intelligent power equipment abnormality analysis system and method based on the Internet of Things. Background Art

[0002] During the operation of the generator set, the power plant system and staff need to identify and process the data that does not conform to the normal operation mode as quickly as possible. These abnormal data may be caused by equipment failure, operating errors, external interference or other emergencies, which may affect the stability and reliability of the generator set operation. Therefore, timely and accurate processing of these instantaneous abnormal data is crucial to ensure the normal operation of power generation.

[0003] The existing system realizes real-time monitoring of instantaneous abnormal data through sensors installed on key equipment, compares the real-time monitoring results with the set thresholds, and realizes early warning of instantaneous abnormal data. It has the advantages of rapid response and comprehensive abnormal capture, but there are still some defects, including: 1. Misjudgment of instantaneous abnormal data; 2. Insensitivity to abnormal trends; 3. Vulnerabilities in system control logic; 4. Unreliability of abnormal accident handling (the accident handling process is too dependent on the status and technical level of the staff). Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent analysis system and method for abnormalities of power equipment based on the Internet of Things to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: an intelligent analysis method for abnormalities of power equipment based on the Internet of Things, the method comprising:

[0006] S10: Perform data cleaning on the historical operation data of the power plant units to obtain the normal historical operation data set and the abnormal historical operation data set of the power plant units. Combined with the control logic of the power plant units, define the entities, attributes, and relationships in the normal historical operation data set of the power plant units. Based on the defined entities, attributes, and relationships, convert the semi-structured data obtained from the power plant units into a structured data set, perform knowledge fusion processing on each structured data set, and construct the power plant data logic map and abnormal case map;

[0007] S20: acquiring the real-time operation data of the power plant, and analyzing the abnormal situation of the real-time operation data of the power plant according to the constructed power plant data logic map;

[0008] S30: determining a target abnormal case in the abnormal case map, and the operation and maintenance personnel perform maintenance processing on the power plant unit according to the abnormal processing plan in the target abnormal case;

[0009] S40: The actual abnormality handling scheme of the power plant units by the operation and maintenance personnel is obtained, and the target abnormality cases are integrated in combination with the operation data of the selected entities under abnormal conditions. Based on the integrated processing results, the abnormal case map and the power plant data logic map are updated.

[0010] Further, the S10 includes:

[0011] S101: performing data cleaning on normal historical operation data and abnormal historical operation data of power plant units to obtain normal historical operation data sets and abnormal historical operation data sets of power plant units. In the normal historical operation data sets of power plant units, historical operation data sets belonging to the same data type are defined as an entity, and the entity name is the name of the corresponding data type;

[0012] S102: According to the control logic of the power plant unit, the relationship between the entities is defined. The relationship between two entities is the control logic existing between the two entities. According to the data feature set of each entity, the attribute of each entity is defined. The attribute of the entity is the data feature set of the entity.

[0013] S103: Acquire semi-structured data in the power plant units, the semi-structured data includes the power plant's regulations and specifications, emergency plans, technical measures and accident reports, and use natural language processing technology to extract named entities, relationships between named entities and attributes of named entities in various types of semi-structured data. Based on the extracted named entities and the relationships between named entities, match the named entities with entities defined by the normal historical operation data set. If the match is successful, convert the corresponding type of semi-structured data into structured data, structured data = {named entity, named entity relationship, named entity attribute}, and the collection of various types of structured data acquired is called a structured data set of the corresponding type;

[0014] S104: Perform knowledge fusion processing on the normal historical operation data set of the power plant units and the standard and procedure data set, achieve entity alignment, attribute alignment and relationship alignment in the knowledge fusion process, and construct a logical graph of the power plant data based on the knowledge fusion result;

[0015] Through the entities defined in the normal historical operation data set, the relationships between entities and the attributes of each entity are defined in the abnormal historical operation data set, and the abnormal historical operation data set of the power plant unit and the emergency plan, technical measures, and accident report data set are subjected to knowledge fusion processing. In the process of knowledge fusion processing, entity alignment, attribute alignment, and relationship alignment are achieved. Based on the knowledge fusion results, an abnormal case map is constructed. Knowledge fusion can eliminate the ambiguity between indicators such as entities, relationships, and attributes and factual objects, which is conducive to improving the quality of the formed knowledge base.

[0016] Further, the S20 includes:

[0017] S201: According to the acquired real-time operation data of the power plant, the real-time operation data of each entity is determined, and according to the power plant data logic map, the standard operation data range of each entity is acquired, and an entity is randomly selected to determine whether the real-time operation data of the selected entity is within the standard operation data range of the selected entity. If so, it indicates that the selected entity is working normally, and if not, it indicates that the selected entity is working abnormally;

[0018] S202: When the selected entity is analyzed to be working normally, the historical operation data of the selected entity in the time period [tT, t] is obtained, and the historical operation data obtained in the time period [tT, t] is combined with the associated entities of the selected entity, where t represents the time when the selected entity is analyzed to be working normally, T represents the time interval value and T>0, and whether the selected entity has an abnormal trend is analyzed. The specific analysis method is:

[0019] According to the power plant data logic map, the associated entities of the selected entity are determined, and each associated entity is numbered. The numbering result is: i = 1, 2, ..., z; z represents the total number of associated entities of the selected entity. According to the order of the historical operation data collection time, the historical operation data of the selected entity is put into the set M, M = {w1, w2, ..., wn}, and the historical operation data of the associated entity numbered i is put into the set N i In, N i ={r1i,r2i,…,rni}, according to the logical relationship between the selected entity and each associated entity, the abnormal coefficient of each historical operation data of the selected entity is predicted. The specific prediction formula is:

[0020] F wp =(1 / z)*Σ[|Q wpi -rpi| / Q wpi ];

[0021] Wherein, p = 1, 2, ..., n, which indicates that the collection time points of each historical operation data are numbered in chronological order, n indicates the total number, wp indicates the historical operation data of the selected entity collected at the collection time point numbered p, rpi indicates the historical operation data of the associated entity numbered i collected at the collection time point numbered p, Q wpi It means that when the historical operation data of the selected entity is wp, according to the logical relationship between the selected entity and the associated entity numbered i, the theoretical historical operation data corresponding to the associated entity numbered i is obtained, F wpIt represents the abnormal coefficient corresponding to the historical operation data wp of the selected entity, the subscript of Σ is i=1 and the superscript is z;

[0022] The coordinates (p, F wp ) is represented in a plane rectangular coordinate system, and each coordinate point represented in the plane rectangular coordinate system is connected in sequence using a smooth curve. If the connection curve shows an overall upward trend or the connection curve is parallel to the x-axis as a whole, it means that the selected entity has an abnormal trend. If the connection curve shows an overall downward trend, it means that the selected entity does not have an abnormal trend.

[0023] Through the logical relationship between the selected entity and each associated entity, when the historical operating data of the selected entity is determined, the theoretical historical operating data of each associated entity is calculated, and the abnormal coefficient of the historical operating data of the selected entity is predicted based on the changes between the theoretical historical operating data of each associated entity and the actual historical operating data of each associated entity. This is conducive to analyzing the operating conditions of the selected entity through the operating conditions of the associated entities, and is conducive to improving the analysis accuracy of the abnormal trend of the selected entity.

[0024] Furthermore, when the selected entity has an abnormal trend or the selected entity works abnormally, the target abnormal case is determined in the abnormal case map according to the historical operation data of the selected entity. The specific determination method is as follows:

[0025] When 0≤F wp ≤U, the historical operation data wp of the selected entity is deleted from the set M to obtain the set M1, M1={x1,x2,…,xv}, where 0<U≤0.5, and the eigenvalues ​​of the set M1 are predicted. The specific prediction formula is:

[0026]

[0027] Wherein, c=1,2,…,v, represents the number corresponding to each storage location in the set M1, v represents the total number of storage locations in the set M1, xc represents the historical operation data stored in the c-th storage location in the set M1, and S represents the characteristic value of the set M1;

[0028] In the abnormal case graph, the abnormal operation data in each abnormal case is obtained, and the obtained abnormal operation data is put into the set G b Based on the above method of calculating the eigenvalues ​​of set M1, b The eigenvalue E b Calculation is performed, b = 1, 2, ..., B, which represents the number corresponding to each abnormal case, and B represents the total number of abnormal cases in the abnormal case map;

[0029] According to K b=A b *{β1*[(SG b ) / S]+β2*[(T b -t) / t]} predicts the matching degree between the abnormal cause of the abnormal case numbered b and the abnormal cause of the selected entity, where A b =1 or A b =0, when A b =1, it means that the abnormal entity corresponding to the abnormal case numbered b is the same as the selected entity. b =0, it means that the abnormal entity corresponding to the abnormal case numbered b is different from the selected entity. β1 and β2 both represent proportional coefficients and β1+β2=1. b Indicates the occurrence time of the abnormal case numbered b;

[0030] For maxK b The corresponding number is searched, and the abnormal case matching the searched number is taken as the target abnormal case.

[0031] Further, the S40 includes: when it is analyzed that the selected entity has an abnormal trend, obtaining the actual abnormal handling plan of the power plant unit by the operation and maintenance personnel, obtaining the operation data of the selected entity under the abnormal situation, entering the obtained actual abnormal handling plan and operation data into the target abnormal case, and replacing the target abnormal case with the entered target abnormal case in the abnormal case map, so as to update the abnormal case map;

[0032] When it is analyzed that the selected entity is working abnormally but has no abnormal trend, the abnormal operation data and abnormal operation status of the selected entity are entered into the power plant data logic map. The operation and maintenance personnel will find out the abnormal cause of the selected entity, and enter the abnormal cause, abnormal operation data and processing plan of the selected entity into the abnormal case map to update the power plant data logic map.

[0033] An intelligent analysis system for power equipment anomalies based on the Internet of Things, the system comprising a graph construction module, an abnormal situation analysis module, a target abnormal case determination module and a graph update module;

[0034] The graph construction module is used to construct a power plant data logic graph and an abnormal case graph;

[0035] The abnormal situation analysis module is used to analyze the abnormal situation of the real-time operation data of the power plant;

[0036] The target abnormal case determination unit is used to determine the target abnormal case in the abnormal case map;

[0037] The map updating module is used to update the abnormal case map and the power plant data logic map;

[0038] Further, the graph construction module includes a data cleaning unit, a definition unit, a semi-structured data conversion unit, a power plant data logic graph construction unit and an abnormal case graph construction unit;

[0039] The data cleaning unit cleans the normal historical operation data and the abnormal historical operation data of the power plant unit to obtain the normal historical operation data set and the abnormal historical operation data set of the power plant unit;

[0040] The definition unit defines entities, attributes, and relationships based on the normal historical operation data set of the power plant units;

[0041] The semi-structured data conversion unit uses natural language processing technology to extract named entities, relationships between named entities, and attributes of named entities in various types of semi-structured data, and converts the semi-structured data of the power plant unit into structured data in combination with the entities, relationships, and attributes defined by the definition unit;

[0042] The power plant data logic map construction unit constructs the power plant data logic map by performing knowledge fusion processing on the normal historical operation data set of the power plant unit and the standard procedure data set, and based on the knowledge fusion result;

[0043] The abnormal case map construction unit constructs the abnormal case map by performing knowledge fusion processing on the abnormal historical operation data set of the power plant unit and the emergency plan, technical measures, and accident report data set.

[0044] Further, the abnormal situation analysis module includes an abnormal data judgment unit and an abnormal trend analysis unit;

[0045] The abnormal data judgment unit determines the real-time operation data of each entity according to the acquired real-time operation data of the power plant, acquires the standard operation data range of each entity according to the power plant data logic map, judges whether the real-time operation data of the selected entity is within the standard operation data range of the selected entity, and judges whether the acquired real-time operation data is abnormal data according to the judgment result;

[0046] When the abnormal trend analysis unit determines that the acquired real-time operation data is normal data, it acquires the historical operation data of the selected entity within any period of time, combines the historical operation data acquired within any period of time with the associated entities of the selected entity, and analyzes whether the selected entity has an abnormal trend.

[0047] Further, the target abnormal case determination module includes a target abnormal case determination unit and a maintenance processing unit;

[0048] The target abnormal case determination unit determines the target abnormal case in the abnormal case map according to the historical operation data of the selected entity when the selected entity has an abnormal trend or the selected entity works abnormally;

[0049] The maintenance processing unit uses the abnormality processing plan recorded in the target abnormality case as the maintenance processing plan for the power plant unit by the operation and maintenance personnel.

[0050] Further, the map update module includes an abnormal case map update unit and a power plant data logic map update unit;

[0051] When the abnormal case map updating unit analyzes that the selected entity has an abnormal trend, it obtains the actual abnormal handling plan of the power plant unit by the operation and maintenance personnel, obtains the operating data of the selected entity under abnormal conditions, enters the obtained actual abnormal handling plan and operating data into the target abnormal case, and replaces the target abnormal case with the entered target abnormal case in the abnormal case map;

[0052] When the power plant data logic map update unit analyzes that the selected entity is operating abnormally but has no abnormal trend, the abnormal operation data and abnormal operation status of the selected entity are entered into the power plant data logic map. The operation and maintenance personnel find out the cause of the abnormality of the selected entity and enter the cause of the abnormality, abnormal operation data and processing plan of the selected entity into the abnormal case map.

[0053] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0054] 1. The present invention selects the logical relationship between the entity and each associated entity. When the historical operation data of the selected entity is determined, the theoretical historical operation data of each associated entity is calculated. According to the changes between the theoretical historical operation data of each associated entity and the actual historical operation data of each associated entity, the abnormal coefficient of the historical operation data of the selected entity is predicted, which is conducive to analyzing the operation status of the selected entity through the operation status of the associated entity, thereby improving the analysis accuracy of the abnormal trend of the selected entity.

[0055] 2. The present invention converts semi-structured data into structured data by identifying entities, relationships, and attributes from structured data. Based on the relationship between structured data, normal historical operation data sets of power plant units, and abnormal historical operation data sets of power plant units, a power plant data logic map and an abnormal case map are constructed. In this process, multivariate data can be efficiently integrated together, not limited to structured, single data sources, which is conducive to improving the efficiency of abnormal analysis of power equipment.

[0056] 3. The present invention can quickly judge abnormal operation data through the power plant data logic map, and can effectively identify abnormal power equipment before the power equipment becomes abnormal by analyzing whether the operation data has an abnormal trend. When the power equipment is abnormal, the target abnormal case is searched according to the matching degree between the abnormal operation data of the power equipment and the abnormal operation data recorded in each abnormal case, so that maintenance personnel can quickly determine the abnormal handling plan, further improving the abnormal analysis effect of the system.

[0057] 4. The present invention utilizes the knowledge graph to provide decision support features to help operation and maintenance personnel quickly respond to abnormal events. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0059] Figure 1 It is a schematic diagram of the working process of an intelligent analysis system and method for abnormality of power equipment based on the Internet of Things of the present invention;

[0060] Figure 2 It is a structural schematic diagram of the working principle of an intelligent analysis system and method for abnormalities of power equipment based on the Internet of Things of the present invention. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] See also Figure 1 and Figure 2 The present invention provides a technical solution: an intelligent analysis method for abnormal power equipment based on the Internet of Things, the method comprising:

[0063] S10: Perform data cleaning on the historical operation data of the power plant units to obtain the normal historical operation data set and the abnormal historical operation data set of the power plant units. Combined with the control logic of the power plant units, define the entities, attributes, and relationships in the normal historical operation data set of the power plant units. Based on the defined entities, attributes, and relationships, convert the semi-structured data obtained from the power plant units into a structured data set, perform knowledge fusion processing on each structured data set, and construct the power plant data logic map and abnormal case map;

[0064] S10 includes:

[0065] S101: Perform data cleaning on normal historical operation data and abnormal historical operation data of power plant units. Perform data cleaning on the historical operation data of power plant units to eliminate noise and redundant data in the historical operation data, so as to extract valid data from the historical operation data. Data cleaning includes data collection, data preprocessing, data conversion and data integration. The specific implementation method is the existing technology. A normal historical operation data set and an abnormal historical operation data set of the power plant units are obtained. In the normal historical operation data set of the power plant units, the historical operation data sets belonging to the same data type are defined as an entity. The entity name is the name of the corresponding data type. The data type includes bus voltage, main steam temperature, drum water level, etc.

[0066] S102: According to the control logic of the power plant unit, the relationship between the entities is defined. The relationship between two entities is the control logic between the two entities. According to the data feature set of each entity, the data features of the entity are determined by using the PCA linear feature extraction method. PCA is a linear equation for unsupervised feature extraction with a theoretical basis. The data feature set is the output data set of the entity projected onto the selected feature vector. The attributes of each entity are defined. The attribute of the entity is the data feature set of the entity.

[0067] S103: Acquire semi-structured data in the power plant units. The semi-structured data includes the power plant's procedures and specifications, emergency plans, technical measures and accident reports. Use natural language processing technology to extract named entities, the relationships between named entities and the attributes of named entities in various types of semi-structured data. Natural language processing technology refers to allowing a computer to accept user input in the form of natural language, and perform a series of operations such as processing and calculation internally through algorithms defined by humans to simulate human understanding of natural language and return the results expected by the user. Based on the extracted named entities and the relationships between named entities, match the named entities with entities defined by the normal historical operation data set. If the match is successful (if the name of the named entity is the same as the entity, and the relationship between the named entity and other named entities is the same as the relationship between the entity and other named entities, then the named entity matches the entity successfully), then convert the corresponding type of semi-structured data into structured data. According to, structured data = {named entity, named entity relationship, named entity attribute}, the collection of various types of structured data obtained is called a structured data set of the corresponding type, for example, semi-structured data is: the bus voltage setting range of the power plant is [201kv, 232kv], the incoming line voltage setting range of the power plant is [500kv, 800kv], bus voltage = incoming line voltage / transformation ratio, the type of semi-structured data is regulations and specifications, it can be seen that the bus voltage and incoming line voltage are named entities, [201kv, 232kv], [500kv, 800kv] are the attributes of the named entity bus voltage and the named entity incoming line voltage respectively, the relationship between the named entity bus voltage and the named entity incoming line voltage is bus voltage = incoming line voltage / transformation ratio, then after converting the semi-structured data of the type of regulations and specifications into structured data, the structured data = {bus voltage, bus voltage = incoming line voltage / transformation ratio, [201kv, 232kv]};

[0068] S104: Perform knowledge fusion processing on the normal historical operation data set of the power plant unit and the standard and regulation data set, and achieve entity alignment, attribute alignment and relationship alignment in the process of knowledge fusion processing. Based on the knowledge fusion result, construct the power plant data logic map. The specific construction method is: take the normal historical operation data after cleaning, the control logic of the power plant unit and the regulations and specifications of the power plant as input, use natural language processing technology to process the input power plant regulations and specifications, extract the named entities of the power plant regulations and specifications, the relationship between each named entity and the attributes of the named entity, and match the named entity with the entity defined by the normal historical operation data set based on the extracted named entity and the relationship between each named entity. If the match is successful, one of the named entity and the definition entity is retained. If the match is unsuccessful, both the named entity and the definition entity are retained to achieve entity alignment. Based on the matching result, obtain the regulation and specification data set, classify the normal historical operation data set of the power plant unit according to the definition entity, and obtain several normal historical operation data subsets. According to the named entity, The specification and procedure data set of the power plant unit is classified and processed to obtain several specification and procedure data subsets, and a specification and procedure data subset is randomly selected. If the named entity of the selected specification and procedure data subset is bus voltage, the selected specification and procedure data subset is merged with the normal operation data subset whose entity is defined as bus voltage to achieve attribute alignment. According to the entity alignment, the entities in the power plant data logic map are determined. According to the attribute alignment, the attributes of each entity in the power plant data logic map are determined, and the control logic of the input power plant unit and the relationship between each extracted named entity are aligned. If the relationship between the two entities analyzed by the control logic of the power plant unit is: bus voltage = incoming line voltage / transformation ratio, and the relationship between the two extracted named entities is: incoming line voltage = bus voltage × transformation ratio, then the relationship alignment result is: bus voltage = incoming line voltage / transformation ratio. Based on the relationship alignment, the relationship between each entity in the power plant data logic map is determined, and the power plant data logic map is constructed according to the format of entity attribute-entity-entity relationship-associated entity;

[0069] Through the entities defined in the normal historical operation data set, the relationship between the entities and the attributes of the entities are defined in the abnormal historical operation data set, and the abnormal historical operation data set of the power plant unit and the emergency plan data set, technical measures data set, and accident report data set are subjected to knowledge fusion processing. In the process of knowledge fusion processing, entity alignment, attribute alignment, and relationship alignment are achieved. Based on the knowledge fusion result, an abnormal case map is constructed. The construction method of the abnormal case map is the same as the construction method of the power plant data logic map. Knowledge fusion refers to the process of integrating knowledge in multiple knowledge bases to form a knowledge base. According to the method of converting semi-structured data into structured data in S103, the emergency plan, technical measures, and accident report are processed to obtain the emergency plan data set, technical measures data set, and accident report data set;

[0070] S20: acquiring the real-time operation data of the power plant, and analyzing the abnormal situation of the real-time operation data of the power plant according to the constructed power plant data logic map;

[0071] The S20 includes:

[0072] S201: According to the acquired real-time operation data of the power plant, the real-time operation data of each entity is determined, and according to the power plant data logic map, the standard operation data range of each entity is obtained, and an entity is randomly selected to determine whether the real-time operation data of the selected entity is within the standard operation data range of the selected entity. If yes, it means that the selected entity is working normally, that is, the determined real-time operation data is normal data; if no, it means that the selected entity is working abnormally, that is, the determined real-time operation data is abnormal data;

[0073] S202: When the selected entity is analyzed to be working normally, the historical operation data of the selected entity in the time period [tT, t] is obtained, and the historical operation data obtained in the time period [tT, t] is combined with the associated entities of the selected entity, where t represents the time when the selected entity is analyzed to be working normally, T represents the time interval value and T>0, and whether the selected entity has an abnormal trend is analyzed. The specific analysis method is:

[0074] According to the power plant data logic map, the associated entities of the selected entity are determined, and each associated entity is numbered. The numbering result is: i = 1, 2, ..., z; z represents the total number of associated entities of the selected entity. The associated entity refers to the entity that has a logical relationship with the selected entity, that is, there is a control logic between the associated entity and the selected entity (the control logic is the relationship between the associated entity and the selected entity). For example: the relationship between the entity bus voltage and the entity incoming line voltage is "bus voltage = incoming line voltage / transformation ratio". It can be seen that when the selected entity is the entity bus voltage, the entity incoming line voltage is the associated entity of the entity bus voltage. According to the order of the historical operation data collection time, the historical operation data of the selected entity is put into the set M, M = {w1, w2, ..., wn}, and the historical operation data of the associated entity numbered i is put into the set N i In, N i ={r1i,r2i,…,rni}, according to the logical relationship between the selected entity and each associated entity, the abnormal coefficient of each historical operation data of the selected entity is predicted. The specific prediction formula is:

[0075] F wp =(1 / z)*Σ[|Q wpi -rpi| / Q wpi ];

[0076] Wherein, p = 1, 2, ..., n, which indicates that the collection time points of each historical operation data are numbered in chronological order, n indicates the total number, wp indicates the historical operation data of the selected entity collected at the collection time point numbered p, rpi indicates the historical operation data of the associated entity numbered i collected at the collection time point numbered p, Q wpi It means that when the historical operation data of the selected entity is wp, according to the logical relationship between the selected entity and the associated entity numbered i, the theoretical historical operation data corresponding to the associated entity numbered i is obtained, F wp represents the abnormal coefficient corresponding to the historical operation data wp of the selected entity. The subscript of Σ is i=1 and the superscript is z. Since the selected entity has been working normally, Q wpi >0;

[0077] The coordinates (p, F wp ) is represented in a plane rectangular coordinate system, and each coordinate point represented in the plane rectangular coordinate system is sequentially connected by a smooth curve. If the connection curve shows an overall upward trend or the connection curve is parallel to the x-axis as a whole, it indicates that the selected entity has an abnormal trend. If the connection curve shows an overall downward trend, it indicates that the selected entity does not have an abnormal trend.

[0078] S30: determining a target abnormal case in the abnormal case map, and the operation and maintenance personnel perform maintenance processing on the power plant unit according to the abnormal processing plan in the target abnormal case;

[0079] When the selected entity has an abnormal trend or the selected entity works abnormally, the target abnormal case is determined in the abnormal case map according to the historical operation data of the selected entity. The specific determination method is as follows:

[0080] When 0≤F wp ≤U, the historical operation data wp of the selected entity is deleted from the set M to obtain the set M1, M1={x1,x2,…,xv}, where 0<U≤0.5, and the eigenvalues ​​of the set M1 are predicted. The specific prediction formula is:

[0081]

[0082] Wherein, c=1,2,…,v, represents the number corresponding to each storage location in the set M1, v represents the total number of storage locations in the set M1, xc represents the historical operation data stored in the c-th storage location in the set M1, and S represents the characteristic value of the set M1;

[0083] In the abnormal case graph, the abnormal operation data in each abnormal case is obtained, and the obtained abnormal operation data is put into the set G b Based on the above method of calculating the eigenvalues ​​of set M1, b The eigenvalue E b Calculation is performed, b = 1, 2, ..., B, which represents the number corresponding to each abnormal case, and B represents the total number of abnormal cases in the abnormal case map;

[0084] According to K b =A b *{β1*[(SG b ) / S]+β2*[(T b -t) / t]} predicts the matching degree between the abnormal cause of the abnormal case numbered b and the abnormal cause of the selected entity, where A b =1 or A b =0, when A b =1, it means that the abnormal entity corresponding to the abnormal case numbered b is the same as the selected entity. b =0, it means that the abnormal entity corresponding to the abnormal case numbered b is different from the selected entity. β1 and β2 both represent proportional coefficients and β1+β2=1. b Indicates the occurrence time of the abnormal case numbered b;

[0085] For maxK bThe corresponding number is searched, and the abnormal case matching the searched number is taken as the target abnormal case;

[0086] S40: obtaining the actual abnormality handling scheme of the power plant unit by the operation and maintenance personnel, integrating the target abnormality cases in combination with the operation data of the selected entity under abnormal conditions, and updating the abnormality case map and the power plant data logic map based on the integrated processing results;

[0087] S40 includes: when it is analyzed that the selected entity has an abnormal trend, obtaining the actual abnormal processing plan of the operation and maintenance personnel for the power plant unit, obtaining the operation data of the selected entity under the abnormal situation, entering the obtained actual abnormal processing plan and operation data into the target abnormal case, and replacing the target abnormal case with the entered target abnormal case in the abnormal case map;

[0088] When it is analyzed that the selected entity is working abnormally and has no abnormal trend, the abnormal operation data and abnormal operation status of the selected entity are entered into the power plant data logic map. The operation and maintenance personnel will find out the cause of the abnormality of the selected entity and enter the abnormal cause, abnormal operation data and processing plan of the selected entity into the abnormal case map.

[0089] An intelligent analysis system for power equipment anomalies based on the Internet of Things, the system includes a graph construction module, an abnormal situation analysis module, a target abnormal case determination module and a graph update module;

[0090] The graph construction module is used to construct the power plant data logic graph and abnormal case graph;

[0091] The graph construction module includes a data cleaning unit, a definition unit, a semi-structured data conversion unit, a power plant data logic graph construction unit, and an abnormal case graph construction unit;

[0092] The data cleaning unit cleans the normal historical operation data and the abnormal historical operation data of the power plant unit to obtain the normal historical operation data set and the abnormal historical operation data set of the power plant unit;

[0093] The definition unit defines entities, attributes, and relationships based on the normal historical operation data set of the power plant units;

[0094] The semi-structured data conversion unit uses natural language processing technology to extract named entities, relationships between named entities, and attributes of named entities in various types of semi-structured data, and converts the semi-structured data of power plant units into structured data by combining the entities, relationships, and attributes defined by the definition unit;

[0095] The power plant data logic map construction unit constructs the power plant data logic map by performing knowledge fusion processing on the normal historical operation data set of the power plant units and the standard procedure data set, and based on the knowledge fusion result;

[0096] The abnormal case map construction unit constructs the abnormal case map by fusing the abnormal historical operation data set of the power plant unit with the emergency plan, technical measures, and accident report data set;

[0097] The abnormal situation analysis module is used to analyze the abnormal situation of the real-time operation data of the power plant;

[0098] The abnormal situation analysis module includes an abnormal data judgment unit and an abnormal trend analysis unit;

[0099] The abnormal data judgment unit determines the real-time operation data of each entity according to the acquired real-time operation data of the power plant, obtains the standard operation data range of each entity according to the power plant data logic map, judges whether the real-time operation data of the selected entity is within the standard operation data range of the selected entity, and judges whether the acquired real-time operation data is abnormal data according to the judgment result;

[0100] When the abnormal trend analysis unit determines that the acquired real-time operation data is normal data, it acquires the historical operation data of the selected entity within any period of time, combines the historical operation data acquired within any period of time with the associated entities of the selected entity, and analyzes whether the selected entity has an abnormal trend;

[0101] The target abnormal case determination unit is used to determine the target abnormal case in the abnormal case map;

[0102] The target abnormal case determination module includes a target abnormal case determination unit and a maintenance processing unit;

[0103] The target abnormal case determination unit determines the target abnormal case in the abnormal case map according to the historical operation data of the selected entity when the selected entity has an abnormal trend or the selected entity works abnormally;

[0104] The maintenance processing unit uses the abnormality processing plan recorded in the target abnormality case as the maintenance processing plan for the power plant unit by the operation and maintenance personnel;

[0105] The map update module is used to update the abnormal case map and the power plant data logic map;

[0106] The map update module includes an abnormal case map update unit and a power plant data logic map update unit;

[0107] When the abnormal case map updating unit analyzes that the selected entity has an abnormal trend, it obtains the actual abnormal handling plan of the power plant unit by the operation and maintenance personnel, obtains the operating data of the selected entity under abnormal conditions, enters the obtained actual abnormal handling plan and operating data into the target abnormal case, and replaces the target abnormal case with the entered target abnormal case in the abnormal case map;

[0108] When the power plant data logic map update unit analyzes that the selected entity is working abnormally but has no abnormal trend, it enters the abnormal operation data and abnormal operation status of the selected entity into the power plant data logic map. The operation and maintenance personnel find out the cause of the abnormality of the selected entity and enter the abnormal cause, abnormal operation data and processing plan of the selected entity into the abnormal case map.

[0109] Example 1: Assume that the selected entity has associated entity 1 and associated entity 2, and assume that the historical operation data w1 of the selected entity collected at the collection time point numbered 1 is 20, and the historical operation data collected by associated entity 1 and associated entity 2 at the collection time point numbered 1 are r11=40 and r12=24 respectively. Then, when the historical operation data of the selected entity is 20, according to the logical relationship between the selected entity and associated entity 1, the theoretical historical operation data corresponding to associated entity 1 is Q w11 =39. According to the logical relationship between the selected entity and associated entity 2, the theoretical historical operation data corresponding to associated entity 2 is Q w12 =25, then the abnormal coefficient of the selected entity historical operation data w1 is:

[0110] F w1 =(1 / z)*Σ[|Q w1i -r1i| / Q w1i ] = (1 / 2) * {[|39-40| / 39] + [|25-24| / 25]} = 0.03;

[0111] Then the abnormal coefficient of the selected entity historical operation data w1 is 0.03.

[0112] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0113] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent analysis method for abnormality of power equipment based on the Internet of Things, characterized by: The method comprises: S10: Perform data cleaning on the historical operation data of the power plant units to obtain the normal historical operation data set and the abnormal historical operation data set of the power plant units. Combined with the control logic of the power plant units, define the entities, attributes, and relationships in the normal historical operation data set of the power plant units. Based on the defined entities, attributes, and relationships, convert the semi-structured data obtained from the power plant units into a structured data set, perform knowledge fusion processing on each structured data set, and construct the power plant data logic map and abnormal case map; S20: acquiring the real-time operation data of the power plant, and analyzing the abnormal situation of the real-time operation data of the power plant according to the constructed power plant data logic map; S30: determining a target abnormal case in the abnormal case map, and the operation and maintenance personnel perform maintenance processing on the power plant unit according to the abnormal processing plan in the target abnormal case; S40: The actual abnormality handling scheme of the power plant units by the operation and maintenance personnel is obtained, and the target abnormality cases are integrated in combination with the operation data of the selected entities under abnormal conditions. Based on the integrated processing results, the abnormal case map and the power plant data logic map are updated.

2. According to the method of intelligent analysis of power equipment abnormality based on the Internet of Things according to claim 1, it is characterized by: The S10 includes: S101: performing data cleaning on normal historical operation data and abnormal historical operation data of power plant units to obtain normal historical operation data sets and abnormal historical operation data sets of power plant units. In the normal historical operation data sets of power plant units, historical operation data sets belonging to the same data type are defined as an entity, and the entity name is the name of the corresponding data type; S102: According to the control logic of the power plant unit, the relationship between the entities is defined. The relationship between two entities is the control logic existing between the two entities. According to the data feature set of each entity, the attribute of each entity is defined. The attribute of the entity is the data feature set of the entity. S103: Acquire semi-structured data in the power plant units, the semi-structured data includes the power plant's regulations and specifications, emergency plans, technical measures and accident reports, and use natural language processing technology to extract named entities, relationships between named entities and attributes of named entities in various types of semi-structured data. Based on the extracted named entities and the relationships between named entities, match the named entities with entities defined by the normal historical operation data set. If the match is successful, convert the corresponding type of semi-structured data into structured data, structured data = {named entity, named entity relationship, named entity attribute}, and the collection of various types of structured data acquired is called a structured data set of the corresponding type; S104: Perform knowledge fusion processing on the normal historical operation data set of the power plant units and the standard and procedure data set, achieve entity alignment, attribute alignment and relationship alignment in the knowledge fusion process, and construct a logical graph of the power plant data based on the knowledge fusion result; Through the entities defined in the normal historical operation data set, the relationships between entities and the attributes of each entity are defined in the abnormal historical operation data set, and the abnormal historical operation data set of the power plant unit and the emergency plan, technical measures, and accident report data set are subjected to knowledge fusion processing. In the process of knowledge fusion processing, entity alignment, attribute alignment, and relationship alignment are achieved. Based on the knowledge fusion results, an abnormal case map is constructed.

3. According to claim 2, a method for intelligent analysis of power equipment abnormalities based on the Internet of Things is characterized by: The S20 includes: S201: According to the acquired real-time operation data of the power plant, the real-time operation data of each entity is determined, and according to the power plant data logic map, the standard operation data range of each entity is acquired, and an entity is randomly selected to determine whether the real-time operation data of the selected entity is within the standard operation data range of the selected entity. If so, it indicates that the selected entity is working normally, and if not, it indicates that the selected entity is working abnormally; S202: When the selected entity is analyzed to be working normally, the historical operation data of the selected entity in the time period [tT, t] is obtained, and the historical operation data obtained in the time period [tT, t] is combined with the associated entities of the selected entity, where t represents the time when the selected entity is analyzed to be working normally, T represents the time interval value and T>0, and whether the selected entity has an abnormal trend is analyzed. The specific analysis method is: According to the power plant data logic map, the associated entities of the selected entity are determined, and each associated entity is numbered. The numbering result is: i = 1, 2, ..., z; z represents the total number of associated entities of the selected entity. The historical operation data of the selected entity is put into the set M in the order of the historical operation data collection time. M = {w1, w2, ..., wn}, put the historical operation data of the associated entity numbered i into the set N i In, N i ={r1i,r2i,…,rni}, according to the logical relationship between the selected entity and each associated entity, the abnormal coefficient of each historical operation data of the selected entity is predicted. The specific prediction formula is: F wp =(1 / z)*Σ[|Q wpi -rpi| / Q wpi ]; Wherein, p = 1, 2, ..., n, which indicates that the collection time points of each historical operation data are numbered in chronological order, n indicates the total number, wp indicates the historical operation data of the selected entity collected at the collection time point numbered p, rpi indicates the historical operation data of the associated entity numbered i collected at the collection time point numbered p, Q wpi It means that when the historical operation data of the selected entity is wp, according to the logical relationship between the selected entity and the associated entity numbered i, the theoretical historical operation data corresponding to the associated entity numbered i is obtained, F wp It represents the abnormal coefficient corresponding to the historical operation data wp of the selected entity, the subscript of Σ is i=1 and the superscript is z; The coordinates (p, F wp ) is represented in a plane rectangular coordinate system, and each coordinate point represented in the plane rectangular coordinate system is connected in sequence using a smooth curve. If the connection curve shows an overall upward trend or the connection curve is parallel to the x-axis as a whole, it means that the selected entity has an abnormal trend. If the connection curve shows an overall downward trend, it means that the selected entity does not have an abnormal trend.

4. According to claim 3, the method for intelligent analysis of power equipment abnormalities based on the Internet of Things is characterized by: When the selected entity has an abnormal trend or the selected entity works abnormally, the target abnormal case is determined in the abnormal case map according to the historical operation data of the selected entity. The specific determination method is as follows: When 0≤F wp ≤U, the historical operation data wp of the selected entity is deleted from the set M to obtain the set M1, M1={x1,x2,…,xv}, where 0<U≤0.5, and the eigenvalues ​​of the set M1 are predicted. The specific prediction formula is: Wherein, c=1,2,…,v, represents the number corresponding to each storage location in the set M1, v represents the total number of storage locations in the set M1, xc represents the historical operation data stored in the c-th storage location in the set M1, and S represents the characteristic value of the set M1; In the abnormal case graph, the abnormal operation data in each abnormal case is obtained, and the obtained abnormal operation data is put into the set G b Based on the above method of calculating the eigenvalues ​​of set M1, b The eigenvalue E b Calculation is performed, b = 1, 2, ..., B, which represents the number corresponding to each abnormal case, and B represents the total number of abnormal cases in the abnormal case map; According to K b =A b *{β1*[(SG b ) / S]+β2*[(T b -t) / t]} predicts the matching degree between the abnormal cause of the abnormal case numbered b and the abnormal cause of the selected entity, where A b =1 or A b =0, when A b =1, it means that the abnormal entity corresponding to the abnormal case numbered b is the same as the selected entity. b =0, it means that the abnormal entity corresponding to the abnormal case numbered b is different from the selected entity. β1 and β2 both represent proportional coefficients and β1+β2=1. b Indicates the occurrence time of the abnormal case numbered b; For maxK b The corresponding number is searched, and the abnormal case matching the searched number is taken as the target abnormal case.

5. According to claim 4, the method for intelligent analysis of power equipment abnormalities based on the Internet of Things is characterized by: The step S40 includes: when it is analyzed that the selected entity has an abnormal trend, obtaining the actual abnormal handling plan of the power plant unit by the operation and maintenance personnel, obtaining the operation data of the selected entity under abnormal conditions, entering the obtained actual abnormal handling plan and operation data into the target abnormal case, and replacing the target abnormal case with the entered target abnormal case in the abnormal case map; When it is analyzed that the selected entity is working abnormally and has no abnormal trend, the abnormal operation data and abnormal operation status of the selected entity are entered into the power plant data logic map. The operation and maintenance personnel will find out the cause of the abnormality of the selected entity and enter the abnormal cause, abnormal operation data and processing plan of the selected entity into the abnormal case map.

6. An IoT-based power equipment abnormality intelligent analysis system applied to the IoT-based power equipment abnormality intelligent analysis method according to any one of claims 1 to 5, characterized in that: The system includes a graph construction module, an abnormal situation analysis module, a target abnormal case determination module and a graph update module; The graph construction module is used to construct a power plant data logic graph and an abnormal case graph; The abnormal situation analysis module is used to analyze the abnormal situation of the real-time operation data of the power plant; The target abnormal case determination unit is used to determine the target abnormal case in the abnormal case map; The map updating module is used to update the abnormal case map and the power plant data logic map.

7. The intelligent analysis system for abnormality of power equipment based on the Internet of Things according to claim 6 is characterized by: The graph construction module includes a data cleaning unit, a definition unit, a semi-structured data conversion unit, a power plant data logic graph construction unit and an abnormal case graph construction unit; The data cleaning unit cleans the normal historical operation data and the abnormal historical operation data of the power plant unit to obtain the normal historical operation data set and the abnormal historical operation data set of the power plant unit; The definition unit defines entities, attributes, and relationships based on the normal historical operation data set of the power plant units; The semi-structured data conversion unit uses natural language processing technology to extract named entities, relationships between named entities, and attributes of named entities in various types of semi-structured data, and converts the semi-structured data of the power plant unit into structured data in combination with the entities, relationships, and attributes defined by the definition unit; The power plant data logic map construction unit constructs the power plant data logic map by performing knowledge fusion processing on the normal historical operation data set of the power plant unit and the standard procedure data set, and based on the knowledge fusion result; The abnormal case map construction unit constructs the abnormal case map by performing knowledge fusion processing on the abnormal historical operation data set of the power plant unit and the emergency plan, technical measures, and accident report data set.

8. The intelligent analysis system for abnormality of power equipment based on the Internet of Things according to claim 7 is characterized by: The abnormal situation analysis module includes an abnormal data judgment unit and an abnormal trend analysis unit; The abnormal data judgment unit determines the real-time operation data of each entity according to the acquired real-time operation data of the power plant, acquires the standard operation data range of each entity according to the power plant data logic map, judges whether the real-time operation data of the selected entity is within the standard operation data range of the selected entity, and judges whether the acquired real-time operation data is abnormal data according to the judgment result; When the abnormal trend analysis unit determines that the acquired real-time operation data is normal data, it acquires the historical operation data of the selected entity within any period of time, combines the historical operation data acquired within any period of time with the associated entities of the selected entity, and analyzes whether the selected entity has an abnormal trend.

9. The intelligent analysis system for abnormality of power equipment based on the Internet of Things according to claim 8 is characterized by: The target abnormal case determination module includes a target abnormal case determination unit and a maintenance processing unit; The target abnormal case determination unit determines the target abnormal case in the abnormal case map according to the historical operation data of the selected entity when the selected entity has an abnormal trend or the selected entity works abnormally; The maintenance processing unit uses the abnormality processing plan recorded in the target abnormality case as the maintenance processing plan for the power plant unit by the operation and maintenance personnel.

10. The intelligent analysis system for abnormality of power equipment based on the Internet of Things according to claim 9 is characterized in that: The map update module includes an abnormal case map update unit and a power plant data logic map update unit; When the abnormal case map updating unit analyzes that the selected entity has an abnormal trend, it obtains the actual abnormal handling plan of the power plant unit by the operation and maintenance personnel, obtains the operating data of the selected entity under abnormal conditions, enters the obtained actual abnormal handling plan and operating data into the target abnormal case, and replaces the target abnormal case with the entered target abnormal case in the abnormal case map; When the power plant data logic map update unit analyzes that the selected entity is operating abnormally but has no abnormal trend, the abnormal operation data and abnormal operation status of the selected entity are entered into the power plant data logic map. The operation and maintenance personnel find out the cause of the abnormality of the selected entity and enter the cause of the abnormality, abnormal operation data and processing plan of the selected entity into the abnormal case map.

Images