Power grid field operation safety risk control inspection key point pushing method, device and equipment

Abstract

The invention discloses a power grid field operation safety risk control inspection key point pushing method, device and equipment. The method comprises the steps that a to-be-tested power grid field operation text is input into an entity recognition model, a target equipment entity in the field operation text is recognized, and the entity recognition model is obtained through training according to historical data of power grid field operation; matching security risk management and control inspection key points associated with the target equipment entity from a preset security regulation database according to the target equipment entity; determining violation management data associated with the security risk management and control inspection key points according to a preset historical violation database; and sorting and pushing the security risk control inspection key points to corresponding terminals according to the violation management data. According to the method, the security risk management and control inspection key points and the violation management data associated with the field operation are intelligently and specifically pushed to related personnel according to the field operation text, the validity of security management and control and inspection is ensured, and the efficiency of security risk management and control is improved.

Description

Technical Field

[0001] The present application relates to the technical field of safe power grid operation, and in particular to a method, device and equipment for pushing key points of safety risk management and control inspection of power grid on-site operations. Background Art

[0002] The construction sites of power grid operations are numerous and wide, with high risk factors. It is necessary to continuously improve the level of production safety and prevent safety accidents. At present, the safety risk control of on-site operations relies on the on-site operators to consciously implement the requirements of relevant power safety work regulations on the one hand, and on the safety auditors to analyze the work behavior of on-site operators and determine whether they violate the safety work regulations on the other hand. For different work sites and different work contents, there are a large number of control and inspection points in the safety work regulations. Manually determining the corresponding control and inspection points of different work contents in the safety work regulations is not only inefficient, but also cannot guarantee that all the control and inspection points corresponding to the work content can be obtained. Summary of the invention

[0003] In view of this, the present application provides a method, device and equipment for pushing the key points of safety risk management and inspection of power grid on-site operations. According to the on-site operation text, the key points of safety risk management and inspection and violation management data related to the on-site operation are intelligently and targetedly pushed to relevant personnel, so as to ensure the effectiveness of safety control and inspection and improve the efficiency of safety risk management.

[0004] According to one aspect of the present application, a method for pushing key points of safety risk management and control audits for power grid field operations is provided, comprising: inputting a power grid field operation text to be tested into an entity recognition model, identifying a target device entity in the field operation text, wherein the entity recognition model is trained based on historical data of power grid field operations; matching key points of safety risk management and control audits associated with the target device entity from a preset safety regulations database based on the target device entity; determining violation management data associated with the key points of safety risk management and control audits based on a preset historical violation database; and pushing the key points of safety risk management and control audits to a terminal corresponding to the field operation text based on the violation management data.

[0005] According to another aspect of the present application, a device for pushing key points of safety risk management and control audits for power grid field operations is provided, comprising: an identification module, used to input a power grid field operation text to be tested into an entity recognition model, and identify a target device entity in the field operation text, wherein the entity recognition model is trained based on historical data of power grid field operations; a matching module, used to match key points of safety risk management and control audits associated with the target device entity from a preset safety regulations database based on the target device entity; and, based on a preset historical violation database, determining violation management data associated with the key points of safety risk management and control audits; and a pushing module, used to push the key points of safety risk management and control audits to a terminal corresponding to the field operation text based on the violation management data.

[0006] According to another aspect of the present application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, and when the processor executes the program, the steps of the method for pushing the key points of the safety risk management and control audit of power grid field operations are implemented.

[0007] By means of the above technical scheme, the present application provides a method, device and equipment for pushing key points of safety risk control and inspection of power grid field operations, which identifies the target equipment entity in the field operation text through a pre-trained entity recognition model to accurately extract the target power equipment involved in the field operation text, and matches the safety risk control and inspection key points associated with the target equipment entity from the preset safety regulations database, and determines the violation management data associated with the safety risk control and inspection key points according to the preset historical violation database, so that the safety risk control and inspection key points and violation management data related to the target power equipment involved in the field operation text are accurately extracted from the massive control and inspection key points contained in a large amount of historical violation data and safety work regulations, avoiding interference of other irrelevant data to relevant personnel, and then distinguishing the priority of the safety risk control and inspection key points according to the violation management data, pushing the safety risk control and inspection key points to the corresponding terminal, providing accurate reference for relevant personnel, so that the operating personnel can effectively regulate the operating behavior during the operation according to the pushed data, and enable the auditors to quickly identify violations more targetedly according to the pushed data, thereby strengthening the safety risk control of field operations and ensuring safe on-site operations.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0010] Figure 1 A flow chart showing a method for pushing key points of safety risk management and control audit of power grid field operations provided by an embodiment of the present application is shown;

[0011] Figure 2 A structural block diagram of a device for pushing key points of safety risk management and control audit of power grid field operations provided by an embodiment of the present application is shown;

[0012] Figure 3 A diagram showing the training process for obtaining an entity recognition model provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0013] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0014] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.

[0015] Those skilled in the art will appreciate that, unless expressly stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, or there may be intermediate elements. In addition, the "connection" or "connection" used herein may include wireless connection or wireless fusion. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0016] Now, exemplary embodiments according to the present application will be described in more detail with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in a variety of different forms and should not be interpreted as being limited to the embodiments set forth herein. It should be understood that these embodiments are provided to make the disclosure of the present application thorough and complete, and to fully convey the concepts of these exemplary embodiments to those of ordinary skill in the art.

[0017] The method for pushing the key points of the safety risk management and control of power grid field operations provided in the embodiments of the present application can be applied to the terminal, can also be applied to the server, and can also be software running in the terminal or the server. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or a server cluster or distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application of the method for pushing the key points of the safety risk management and control of power grid field operations, etc., but is not limited to the above forms.

[0018] In the production site of the power system, it is a complex and crucial task to control and audit the safety risks of on-site operations. Once there are problems in control and audit, it will seriously affect the safety of on-site operations and may even have a major adverse impact on the power system. For the specific project construction process, the key points of on-site operation safety risk control and audit are different descriptions of the requirements for operation safety risk control by personnel in different positions. The actual content is basically the same, so the two can be collectively referred to as control and audit points, and the specific content of the control and audit points is determined according to the relevant State Grid Corporation of China's power safety work regulations (hereinafter referred to as safety regulations). As described in the background technology, the current safety risk control of on-site operations mainly relies on manual work. On the one hand, it relies on the on-site operators to consciously implement the requirements of relevant safety regulations, and on the other hand, it relies on safety auditors to analyze the operating behavior of the operators and determine whether they violate the safety regulations. For different operation contents in different operation scenarios, operators and auditors need to find the required control and audit points in the massive control and audit points included in the relevant safety regulations. The workload is huge, and it cannot be guaranteed that all the required data can be found, which seriously affects the effectiveness of safety risk control and audit. In view of this, the present application provides a method, device and equipment for pushing the key points of safety risk control and auditing of power grid on-site operations. According to the text content of the work plan, the key points of safety risk control and auditing and violation management data related to the on-site operation are pushed to the operators and auditors in advance, which helps the operators to implement the key points of safety risk control and auditing during the operation process, and effectively helps the auditors to quickly identify violations in a more targeted manner, which is of great significance to strengthening the safety risk control at the operation site and ensuring the safety of on-site operations.

[0019] In this embodiment, a method for pushing key points of safety risk management and control inspection of power grid field operations is provided, such as Figure 1 As shown, the method includes:

[0020] Step 101: input the power grid field operation text to be tested into an entity recognition model to identify the target device entity in the field operation text.

[0021] Among them, the entity recognition model is trained based on the historical data of power grid field operations.

[0022] In this embodiment, the target equipment entities in the field operation text are identified through a pre-trained entity recognition model to extract all target power equipment involved in the field operation text, providing accurate objects for subsequent safety risk management and auditing.

[0023] Among them, power equipment refers to equipment in the power field, such as transformers, circuit breakers, disconnectors, etc., as well as other components directly related to the equipment.

[0024] In one embodiment, the method for pushing key points of safety risk management and control audits for power grid field operations also includes: determining a first training sample set with location labels and a second training sample set without location labels based on historical data of power grid field operations; training a preset entity recognition model based on the first training sample set to obtain a first recognition model; inputting the second training sample set into the first recognition model for labeling to determine a third training sample set carrying a prediction label; performing noise enhancement processing on the third training sample set; training a preset entity recognition model based on the first training sample set and the third training sample set after noise enhancement processing to obtain a second recognition model; and determining the entity recognition model based on the recognition accuracy of the second recognition model for the first training sample set.

[0025] Specifically, the entity recognition model is determined based on the preset entity recognition model that has been trained. The preset entity recognition model is built based on a neural network model and is used to identify the power equipment entities involved in the input text. However, due to the particularity of the historical data of power operations, it is difficult to obtain a large number of labeled samples. If there are insufficient labeled samples during the training of the preset entity recognition model built based on the neural network, the preset entity recognition model will not be able to accurately identify the equipment entities in the input text.

[0026] In this embodiment, there is no need to label all historical data of power grid field operations. Only a small amount of historical data is marked as a first training sample set with location labels. The preset entity recognition model is trained based on the first training sample set, and the preset entity recognition model is used to predict the labels of the second training sample set without location labels. The third training sample set carrying the predicted labels is also used as training data for the preset model, thereby expanding the scale of training data, thereby solving the problem of insufficient labeled samples due to the difficulty in obtaining a large number of labeled samples in power operation scenarios, and ensuring the accuracy of the preset entity recognition model in recognizing the input text.

[0027] Specifically, this embodiment trains the preset entity recognition model based on the small sample learning method of Noisy Student. Figure 3 As shown, a small number of labeled samples are used to train the preset entity recognition model, and then the preset entity recognition model is used to predict the labels of a large number of unlabeled samples, and noise is added to the samples with predicted labels, so that the preset entity recognition model is trained together with the labeled samples and the samples with predicted labels after adding noise until the performance of the preset entity recognition model meets the requirements.

[0028] Exemplarily, first, a small amount of historical data of grid field operations is labeled according to a preset ratio as a first training sample set with location labels, and the remaining historical data is determined as a second training sample set without location labels, for example, 1% or less of the historical data is labeled. Then, a preset entity recognition model is trained according to the first training sample set to obtain a first recognition model, and the first recognition model is used as a teacher model.

[0029] Furthermore, the first recognition model is used to annotate the device entities in the second training sample set, generate prediction labels for the second training sample set, and determine a third training sample set carrying the prediction labels. Thus, labeled training data is determined based on the third training sample set and the first training sample set, the scale of training data is expanded, and the new preset entity recognition model is trained.

[0030] It is worth mentioning that after obtaining the third training sample set, this embodiment introduces strong noise to the third training sample set through methods such as dropout and random data enhancement, and jointly trains a new preset entity recognition model based on the first training sample set and the third training sample set after noise enhancement processing, so that the preset entity recognition model can better cope with various noises and uncertainties existing in the on-site operation text in the power operation scenario, thereby determining the second recognition model, and using the second recognition model as the student model.

[0031] Furthermore, the first training sample set with location tags is also used as a test set to obtain the recognition accuracy of the second recognition model to verify the recognition performance of the second recognition model for the device entity. Thus, the entity recognition model is determined according to the recognition accuracy of the second recognition model for the first training sample set.

[0032] Furthermore, as a refinement and extension of the specific implementation methods of the above-mentioned embodiments, in order to fully illustrate the specific implementation process of this embodiment, the entity recognition model is determined according to the recognition accuracy of the second recognition model for the first training sample set, specifically including: if the recognition accuracy is less than the second preset threshold, the first recognition model is updated according to the second recognition model until the recognition accuracy is greater than or equal to the second preset threshold; if the recognition accuracy is greater than or equal to the second preset threshold, the second recognition model with a recognition accuracy greater than or equal to the second preset threshold is determined as the entity recognition model.

[0033] In this embodiment, through multiple iterations and self-training of the preset entity recognition model, the recognition performance of the preset entity recognition model for the device entity is optimized, and the recognition accuracy of the preset entity recognition model for the device entity is gradually improved.

[0034] Exemplarily, if the recognition accuracy is less than a second preset threshold, the first recognition model is updated according to the second recognition model serving as a student model, so that the second recognition model serves as a teacher model, regenerates prediction labels for the second training sample set, and trains a new preset entity recognition model until the recognition accuracy of the second recognition model is greater than or equal to the second preset threshold, and determines the entity recognition model based on the second recognition model whose recognition accuracy is greater than or equal to the second preset threshold, so that the performance of the entity recognition model in identifying device entities meets the requirements.

[0035] The second preset threshold is determined according to the requirements of the actual power operation scenario for the recognition performance of the entity recognition model. For example, the second preset threshold is set to 90% so that the recognition performance of the entity recognition model for the device entity reaches a higher level.

[0036] Further, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, the step of training the preset entity recognition model specifically includes: constructing structure data in the form of triples according to the preset demand information for extracting power equipment and the training samples input to the preset entity recognition model, wherein the output data in the structure data is a set of labeled device entities pre-identified in the input data; based on the BERT model, determining the feature vector matrix of the training samples input to the preset entity recognition model; determining the entity start position of the training samples input to the preset entity recognition model according to the fully connected layer of the BERT model and the feature vector matrix. The method comprises the following steps: first, a first component of the loss function corresponding to the entity start position label, and a second component of the loss function corresponding to the entity end position label, and a third component of the loss function corresponding to the predicted device entity are determined according to the output data in the structural data; the first component value of the loss function, the second component value of the loss function, and the third component value of the loss function are respectively calculated by the cross entropy function, and the calculated first component value of the loss function, the second component value of the loss function, and the third component value of the loss function are weighted to determine the target loss function until the target loss function converges.

[0037] In this embodiment, a preset entity recognition model is established based on the BERT model, and structural data in the form of triples is constructed according to the preset demand for extracting power equipment and the training samples input into the preset entity recognition model, wherein the output data is a set of labeled equipment entities pre-identified in the input data, so that the preset entity recognition model can more fully consider the contextual information in the training samples according to the preset demand for extracting power equipment during the training process, and improve the preset entity recognition model's ability to understand the contextual context in the training text, thereby ensuring the preset entity recognition model's recognition effect on equipment entities in different on-site operation texts to adapt to complex power grid operation scenarios.

[0038] Exemplarily, the demand for power equipment is extracted in advance using natural language to specifically describe the demand, and the demand information Q is determined in the form of question text. The demand information Q is expressed as: Q = {q1, q2, ..., q m}, Q is a string, and m is the length of the string.

[0039] Furthermore, according to the demand information Q, the input data X in the training sample, and the output data y in the training sample, a triplet of structured data (Q, y, X) is constructed. The triplet of structured data is shown in Table 1.

[0040] Among them, the special characters [CLS] and [SEP] of the BERT model are used to connect the demand information Q with the input data X in the training sample to construct a standard structure. The standard structure is expressed as

[0041] {[CLS],q1,q2,…,q m ,[SEP],x1,x2,…,x n}, X={x1,x2,…,x n} is a string, and n is the length of the string. The output data y is a set of annotated device entities that have been pre-identified in the input data X, specifically including the annotated start position label x of each annotated device entity in the input data X. i Begin and mark the end position label x i End , y={(x i Begin ,x i End )|i=1,2,…,K}, K is the number of labeled device entities contained in the input data X. It is worth mentioning that each labeled device entity can be converted into a substring of the input data X.

[0042] Table 1

[0043]

[0044] Furthermore, during the training process, the entity recognition model is preset to establish two labels for each character in the input data X, which are respectively used to determine whether it is the start position or the end position of the device entity.

[0045] Specifically, for an input data X of length n in the structure data, the matrix represented by its feature vector is obtained through the BERT model Where d is the number of hidden layers. Next, based on the fully connected layer and feature vector matrix of the BERT model, the entity start position label and entity end position label of the input data X are determined.

[0046] For example, the probability matrix P of the entity start position label and the entity end position label in the input data X is obtained by the BERT model Begin and P End Expressed as: P Begin =softmax(M X ·W Begin T ),

[0047] P End =softmax(M X ·W End T ).in, There are two learnable fully connected layers, which are used to determine the starting position and the ending position respectively.

[0048] Furthermore, according to the probability matrix P Begin and P End Determine the starting position labels of all entities in input data X End position tag with entity For example,

[0049] in, P Begin The i-th row value of corresponds to the i-th character in the input data X. Indicates that the i-th character in the input data X is judged as the entity start position by the preset entity recognition model. P End The j-th row value of corresponds to the j-th character in the input data X. Indicates that the j-th character in the input data X is determined as the end position of the entity by the preset entity recognition model.

[0050] Furthermore, the start position tag is calculated and the ending position tag Matching forms the probability value corresponding to the predicted equipment entity that meets the preset power extraction equipment demand Thus, multiple predicted device entities existing in an input data X are determined.

[0051] in, is a learnable fully connected layer, and the function concat(·) connects two characters.

[0052] Furthermore, according to the entity start position label, entity end position label, predicted device entity and output data in the structure data, the structure of the target loss function of the preset entity recognition model is determined, each component of the target loss function is calculated by the cross entropy function, and each component is weighted to obtain the target loss function, and training is performed until the target loss function converges.

[0053] Exemplarily, the label starting position x is determined based on the output data y in the training sample. i Begin and mark the end position label x i End The corresponding probability matrix Y Begin With Y End , and the probability matrix Y corresponding to the labeled device entity Begin,End . Further, according to the probability matrix P corresponding to the start position label Begin and label the start position of the label x i Begin The corresponding probability matrix Y Begin Determine the first component of the loss function, based on the end position label The corresponding probability matrix P End End position label with annotation x i End The corresponding probability matrix Y End Determine the second component of the loss function, based on the probability value corresponding to the predicted device entity Probability matrix Y corresponding to the annotated device entity Begin,End Determine the third component of the loss function. Then determine the target loss function based on the first component of the loss function, the second component of the loss function and the third component of the loss function.

[0054] Specifically, CE is used to represent the cross entropy function, which is used to calculate the values ​​of each component of the loss function. Therefore, the value of the first component of the loss function is expressed as L Begin =CE(P Begin ,Y Begin ), the second component value of the loss function is expressed as L End =CE(P End ,Y End ), the third component of the loss function is expressed as L Span =CE(P Begin ,End,Y Begin,End ), the final objective loss function is expressed as L = αL Begin +βL End +γL Span , α, β, γ∈[0,1] are all preset hyperparameters used to control the proportion of each loss function component.

[0055] In this embodiment, by constructing structural data including preset requirements for extracting power equipment, the entity recognition task (Named Entity Recognition, NER) of the BERT model is converted into a reading comprehension task (Machine Reading Comprehension, MRC), so that the trained entity recognition model can better understand the specific context of the power operation scenario in the on-site operation text according to the preset requirements for extracting power equipment, thereby enabling the entity recognition model to better adapt to complex power grid operation scenarios.

[0056] Step 102, matching the security risk management and control audit points associated with the target device entity from a preset security regulations database according to the target device entity.

[0057] In this embodiment, all control and inspection points related to the on-site operation are quickly located in the preset safety regulations database through the target equipment entity in the on-site operation text, providing more comprehensive control and inspection points, avoiding errors that may occur during manual search and understanding, and improving the efficiency of safety risk control and inspection.

[0058] It should be noted that this embodiment pre-sorts all the control and inspection points included in the relevant safety regulations, extracts the source power equipment involved in the control and inspection points in the safety regulations, determines the source equipment entity, sorts out the control and inspection points according to the source equipment entity, and establishes a safety regulations database. Among them, the preset safety regulations database includes source equipment entities and control and inspection points, and the source equipment entities and the control and inspection points have a corresponding relationship.

[0059] Further, as a refinement and extension of the specific implementation methods of the above-mentioned embodiment, in order to fully illustrate the specific implementation process of this embodiment, step 102, that is, the step of matching the security risk management and audit points associated with the target device entity from the preset security regulations database according to the target device entity, specifically includes: matching the target device entity with the source device entity in the preset security regulations database, and determining the matching confidence between the target device entity and the source device entity; if the matching confidence is greater than or equal to the first preset threshold, then determining the control and audit points corresponding to the source device entity that matches the target device entity as the security risk management and audit points associated with the target device entity.

[0060] In this embodiment, the target device entity in the on-site operation text is matched with the source device entity in the preset safety regulations database, and the matching degree between the target device entity and the source device entity is quantified by the matching confidence level to prevent false matching, thereby accurately locating the source device entity corresponding to the target device entity in the preset safety regulations database, and then accurately extracting the safety risk control and inspection points associated with the on-site operation in the relevant safety regulations requirements based on the control and inspection points corresponding to the source device entity, thereby improving the accuracy and comprehensiveness of the information.

[0061] In one embodiment, the method for pushing key points of safety risk management and control inspection for power grid field operations also includes: performing text enhancement processing on the source device entity in the preset safety regulations database to obtain a fuzzy device entity; associating the fuzzy device entity with the key points of management and control inspection corresponding to the source device entity, and updating the source device entity in the preset safety regulations database according to the fuzzy device entity.

[0062] In this embodiment, a fuzzy device entity is obtained by performing text enhancement processing on the source device entity, so that the fuzzy device entity can cover more equipment features and descriptions in the power operation scenario to adapt to the diversified expression of power equipment in the power operation scenario, thereby associating the fuzzy device entity with the management and inspection points corresponding to the source device entity, and updating the source device entity in the preset safety regulations database according to the fuzzy device entity, so that the preset safety regulations database can adapt to the newly emerging equipment description methods, so that the preset safety regulations database can keep pace with the times and improve its practicality in different power operation scenarios.

[0063] Exemplarily, the source equipment entity is obtained from the preset safety regulations database, and a text enhancement method such as back translation is used to generate a fuzzy equipment entity that is different from the source equipment entity description but has a similar meaning, and the fuzzy equipment entity is associated with the control and inspection points corresponding to the source equipment entity, and the source equipment entity in the preset safety regulations database is updated according to the fuzzy equipment entity. Further, a fuzzy matching model is established based on the BERT model. Specifically, the word embedding method of the BERT model is used to represent the historical equipment entity in the historical data of the power field operation in the form of a sentence vector, and the special character [CLS] is used to represent it as a 768-dimensional high-dimensional feature vector as input, and the updated source equipment entity is used as a label to train the BERT model so that synonyms are as close as possible in the feature space, so that synonyms with different expressions are associated. Through the above method, each target equipment entity identified from the field operation text can be classified into a source equipment entity in a preset safety regulations database through the fuzzy matching model. Then, the control and inspection points corresponding to the source equipment entity matching the target equipment entity are determined as the safety risk control and inspection points associated with the target equipment entity.

[0064] Furthermore, in order to ensure the matching effect between the target device entity and the source device entity, the fuzzy matching model outputs the result only when the matching confidence between the source device entity matching the target device entity and the target device entity is greater than or equal to the first preset threshold. That is, the fuzzy matching model outputs the source device entity matching the target device entity only when the matching confidence is greater than or equal to the first preset threshold. This prevents the model from mistakenly classifying the target device entity as a certain source device entity, resulting in the occurrence of a wrong match. For example, when the target power equipment in the field operation text is not included in the relevant safety regulations, the blind matching of the fuzzy matching model is prevented.

[0065] For example, Among them, S(x) is the probability distribution confidence (matching confidence) of the target device entity input into the fuzzy matching model, and ε is the first preset threshold, which is determined according to the precise matching requirements in the power operation scenario. For example, ε is set to 0.9.

[0066] Step 103, determining the violation management data associated with the safety risk control audit points based on the preset historical violation database.

[0067] It should be noted that, similar to the preset safety regulations database, this embodiment pre-collects and sorts out the historical violation data of power field operations, sorts out the historical violation data according to the control and inspection points in the preset safety regulations database, determines the number of violations and risk levels corresponding to each control and inspection point, and determines the number of violations and risk levels as violation management data, thereby establishing a violation database. Among them, the preset historical violation database includes source equipment entities and control and inspection points. And the control and inspection points have a corresponding relationship with the violation management data.

[0068] In this embodiment, by determining the violation management data associated with the key points of on-site operation safety risk control and inspection, the potential risks in the operation process are clarified for the operators to avoid the occurrence of violations. At the same time, the risk level corresponding to each operation behavior in the operation process is clarified for the inspectors, making the inspection work more targeted, and the inspection resources can be reasonably allocated according to the risk level, focusing on high-risk operation links to improve the inspection efficiency.

[0069] Exemplarily, the violation management data associated with the control and inspection points corresponding to the safety risk control and inspection points in the preset historical violation database are determined as the violation management data associated with the safety risk control and inspection points.

[0070] Step 104, push the safety risk control inspection points to the corresponding terminal based on the violation management data.

[0071] In this embodiment, the priorities of the safety risk control and inspection points are distinguished according to the violation management data, ensuring that the pushed safety risk control and inspection points are the control and inspection points that relevant personnel most want to obtain, thereby ensuring the effectiveness of the pushed content.

[0072] Further, as a refinement and extension of the specific implementation methods of the above-mentioned embodiment, in order to fully illustrate the specific implementation process of this embodiment, step 104, that is, the step of pushing the safety risk control inspection points to the corresponding terminal according to the violation management data, specifically includes: classifying the safety risk control inspection points according to preset rules to determine the classification results; sorting the safety risk control inspection points according to the violation management data and the classification results to determine the data order; dividing the data order into a first sub-sequence and a second sub-sequence according to a preset bit sequence; pushing the safety risk control inspection points in the first sub-sequence to the terminal corresponding to the on-site operation text in sequence; folding the safety risk control inspection points in the second sub-sequence into a push data group, and pushing the first push data group to the terminal corresponding to the on-site operation text.

[0073] In this embodiment, the safety risk control inspection points are classified and sorted according to the violation management data to distinguish different safety risk control inspection points, obtain the priority of the safety risk control inspection points, prevent the safety risk control inspection points from being too large in number and too much in content from affecting the push effect, and ensure the effectiveness of the push.

[0074] Specifically, the control and inspection points in the preset safety regulations database can be divided into two categories: general control and inspection points and special control and inspection points. General control and inspection points generally belong to management work, which represents a series of common procedural operations in production safety control work, such as ensuring that construction personnel are on duty with certificates, ensuring that the working time is consistent with the work ticket, and the number of workers meets the requirements. Special control and inspection points are generally related to the content of the operation, which represents the handling method for a specific operation in production safety control work. For example, power cables have different control and inspection points for different operations such as cable head production and cable withstand voltage testing.

[0075] Exemplarily, according to the above preset rules, the security risk control audit points associated with the target device entity are divided into special security risk control audit points and general security risk control audit points. Among them, the security risk control audit points that meet the preset rules are regarded as special security risk control audit points, and the security risk control audit points that do not meet the preset rules are regarded as general security risk control audit points.

[0076] Next, the special safety risk control and inspection points and the general safety risk control and inspection points are sorted from high to low according to the number of violations and the risk level in the violation management data, and the first data sequence and the second data sequence are determined. Further, the first data sequence is divided according to the preset position sequence, and the first sub-sequence and the second sub-sequence of the first data sequence are determined. Next, the safety risk control and inspection points in the first sub-sequence of the first data sequence are pushed to the terminal corresponding to the on-site operation text in sequence to display and push the special control and inspection points with high risk and high frequency. And the safety risk control and inspection points in the second sub-sequence of the first data sequence are folded into the first push data group and pushed to the terminal corresponding to the on-site operation text to fold and push the remaining low-risk and low-frequency special control and inspection points. Similarly, the first sub-sequence and the second sub-sequence of the second data sequence are determined according to the preset position sequence, and the push method of the general control and inspection points with different risks and different frequencies is distinguished according to the above method, which will not be repeated here.

[0077] It is worth mentioning that all relevant personnel of the power grid can wear mobile communication devices, that is, the present application can correspond to multiple mobile terminals. At this time, the method of the present application is equivalent to a server, which can simultaneously analyze the on-site operation texts sent by different mobile terminals, and send the safety risk control and inspection points to the corresponding mobile terminals, thereby meeting the complex situation of multiple operators and inspectors working at the same time in the power operation scene, improving the efficiency of safety risk control and inspection, and ensuring the safety of relevant personnel in complex power operation scenes. Furthermore, in addition to inputting the on-site operation text in the mobile terminal, the present embodiment can also configure a manual operation selection module for the mobile terminal. The manual operation selection module includes all the control and inspection points in the preset safety regulations database. If the user selects one of the control and inspection points, the manual operation selection module searches for the source device entity associated with it according to the selected control and inspection point, and uses the control and inspection points corresponding to the source device entity as the safety risk control and inspection points, thereby increasing the control and inspection methods.

[0078] In one embodiment, the method for pushing the key points of safety risk control and inspection of power grid on-site operations also includes: if the entity recognition model fails to identify the target device entity, or fails to obtain the key points of safety risk control and inspection, then the key points of control and inspection that do not comply with the preset rules in the preset safety regulations database are pushed to the terminal corresponding to the on-site operation text.

[0079] In this embodiment, corresponding push data is determined for special situations, and relevant personnel are prompted to perform corresponding inspections to meet complex power grid operation scenarios.

[0080] For example, if the on-site operation text does not contain power equipment, or the power equipment contained does not belong to the control and inspection objects concerned by the relevant safety regulations, that is, the entity recognition model fails to identify the target equipment entity, or fails to obtain the safety risk control and inspection points, then the control and inspection points that do not meet the preset rules in the preset safety regulations database, that is, the general control and inspection points, are determined as the push content. If the relevant personnel receive the push content with only the general control and inspection points, they can check the work plan to prevent deviations in the operation process.

[0081] The present application provides a method, device and equipment for pushing the key points of safety risk control and inspection of power grid field operations, which effectively pushes the corresponding safety risk control and inspection key points in the relevant safety regulations for the power equipment involved in the field operation text, and is easy to use. Compared with the current situation of relying on manual experience summary, the intelligent push method proposed in this application improves work efficiency, reduces the skill requirements for relevant personnel, and can effectively help ensure the safety level of field operations.

[0082] In one embodiment, all the power field operations obtained in a certain province in a certain quarter are used as an application example. The environment used in the application example is: a Core i9-10900K processor with a main frequency of 3.70GHz, a graphics card of NVIDIA GeForce RTX 3090, a video memory size of 24GB, and a memory capacity of 64GB. In the software environment used in this article, the Python version is 3.8.10, and the deep learning framework is PyTorch, and its version is 1.11.0. The specific application steps are as follows:

[0083] 1) Construct a historical data set of power field operations. Collect 38,293 pieces of field operation text data from the province for one month to establish an entity recognition model. In order to verify that this application greatly reduces the requirements for labeled training set data, 200 pieces of data were manually labeled in the 38,293 data sets as the first labeled training sample set. At the same time, in order to verify the effectiveness of the entity recognition model, 4,270 pieces of data were labeled as a test set, and the remaining 33,823 pieces of data constituted the second unlabeled training sample set.

[0084] 2) Use the method in the above embodiment to use a small amount of labeled first training sample set and a large amount of unlabeled second training sample set as training data to train the preset entity recognition model. The indicators used to analyze the entity recognition performance are precision, recall and F1 value. Among them, the F1 value is determined according to the precision and recall. In the training process of the preset entity recognition model, each training is completed after 2 complete epochs, and the iterative process based on the first recognition model and the second recognition model is carried out for 2 rounds. The device entity recognition results are shown in Table 2. Table 2 shows that if only a small amount of labeled data is used for training, the obtained model will be difficult to achieve the basic function of entity recognition; after 2 rounds of iterations of the method of the present application, the F1 value of device entity recognition has reached 0.9988, indicating that the present application can effectively extract device entities in the sample set.

[0085] Table 2

[0086] Iteration round Precision % Recall Rate % F1 value 0 3.95 3.05 0.0344 1 97.48 99.96 0.9870 2 99.78 99.99 0.9988

[0087] 3) When the input field operation text to be tested is a positive example, that is, the input field operation text does contain power equipment, a complete intelligent push result of the key points of safety risk control and inspection will be obtained. Take "10kV*** line * pole removal of branch cable" as a positive example. The target equipment entity "cable" is identified through the entity recognition model, and the target equipment entity "cable" is correctly matched with the source equipment entity "power cable" in the preset safety regulations database. The number of control and inspection key points associated with "power cable" obtained through the preset safety regulations database is 28; the number of violations related to the control and inspection key points associated with "power cable" recorded in the preset historical violation database is 40, of which 14 are special and the rest are general. Thus, the push results as shown in Table 3 are obtained.

[0088] Table 3

[0089]

[0090]

[0091] 4) When the input field operation text to be tested is a counterexample, that is, the input field operation text does not contain power equipment, or the included power equipment does not belong to the equipment objects concerned by the relevant safety regulations, the general control and inspection points will be pushed to avoid the erroneous push of special control and inspection points. Take "emergency repair work: emergency repair of surveillance cameras" as a counterexample. The "camera" is identified through the entity recognition model, but the matching confidence between the "camera" and the source device entity is less than 0.9, which means that the target device entity cannot be associated with any source device entity in the preset safety regulations database. Therefore, in this example, only general control and inspection points can be pushed.

[0092] It should be noted that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0093] Furthermore, if Figure 2 As shown, as a specific implementation of the above-mentioned method for pushing the key points of the safety risk management and control audit of power grid field operations, an embodiment of the present application provides a device 200 for pushing the key points of the safety risk management and control audit of power grid field operations. The device 200 for pushing the key points of the safety risk management and control audit of power grid field operations includes: an identification module 201, a matching module 202 and a pushing module 203.

[0094] The recognition module 201 is used to input the power grid field operation text to be tested into the entity recognition model to recognize the target device entity in the field operation text, wherein the entity recognition model is trained based on the historical data of the power grid field operation;

[0095] A matching module 202 is used to match the security risk management and control audit points associated with the target device entity from a preset security regulations database according to the target device entity; and

[0096] Determine the violation management data associated with the key points of safety risk control and inspection based on the preset historical violation database;

[0097] The push module 203 is used to push the safety risk control inspection points to the terminal corresponding to the on-site operation text according to the violation management data.

[0098] In one embodiment, the matching module 202 is specifically used to match the target device entity with the source device entity in the preset security regulations database, and determine the matching confidence between the target device entity and the source device entity; if the matching confidence is greater than or equal to the first preset threshold, the control and inspection points corresponding to the source device entity that matches the target device entity are determined as the security risk control and inspection points associated with the target device entity.

[0099] The push device 200 for the key points of the safety risk control and inspection of power grid field operations also includes:

[0100] The enhancement module is used to perform text enhancement processing on the source device entity in the preset safety regulations database to obtain a fuzzy device entity; associate the fuzzy device entity with the management and control audit points corresponding to the source device entity, and update the source device entity in the preset safety regulations database according to the fuzzy device entity.

[0101] In one embodiment, the push module 203 is specifically used to classify the safety risk management and inspection points according to preset rules to determine the classification results; sort the safety risk management and inspection points according to the violation management data and the classification results to determine the data order; divide the data order into a first sub-sequence and a second sub-sequence according to a preset sequence; push the safety risk management and inspection points in the first sub-sequence to the terminal corresponding to the on-site operation text in sequence; fold the safety risk management and inspection points in the second sub-sequence into a push data group, and push the first push data group to the terminal corresponding to the on-site operation text.

[0102] The push device 200 for the key points of the safety risk control and inspection of power grid field operations also includes:

[0103] The judgment module is used to push the control and inspection points that do not comply with the preset rules in the preset safety regulations database to the terminal corresponding to the on-site operation text if the entity recognition model fails to identify the target device entity or fails to obtain the safety risk control and inspection points.

[0104] The training module is used to determine a first training sample set with location labels and a second training sample set without location labels based on historical data of power grid field operations; train a preset entity recognition model based on the first training sample set to obtain a first recognition model; input the second training sample set into the first recognition model for labeling to determine a third training sample set carrying a prediction label; perform noise enhancement processing on the third training sample set; train a preset entity recognition model based on the first training sample set and the third training sample set after noise enhancement processing to obtain a second recognition model; and determine the entity recognition model based on the recognition accuracy rate of the second recognition model for the first training sample set.

[0105] In one embodiment, the training module is specifically used to, if the recognition accuracy is less than a second preset threshold, update the first recognition model according to the second recognition model until the recognition accuracy is greater than or equal to the second preset threshold; if the recognition accuracy is greater than or equal to the second preset threshold, determine the second recognition model with a recognition accuracy greater than or equal to the second preset threshold as the entity recognition model.

[0106] In one embodiment, a training module is specifically used to construct structural data in the form of triples according to preset demand information of extracted power equipment and training samples input into a preset entity recognition model, wherein the output data in the structural data is a set of labeled equipment entities pre-identified in the input data; based on the BERT model, determine the feature vector matrix of the training samples input into the preset entity recognition model; determine the entity start position label and the entity end position label of the training samples input into the preset entity recognition model according to the fully connected layer and the feature vector matrix of the BERT model; match the predicted equipment entity that conforms to the structural data according to the entity start position label and the entity end position label; determine the first component of the loss function corresponding to the entity start position label, the second component of the loss function corresponding to the entity end position label, and the third component of the loss function corresponding to the predicted equipment entity according to the output data in the structural data; calculate the first component value of the loss function, the second component value of the loss function, and the third component value of the loss function respectively through the cross entropy function, and perform weighted processing on the calculated first component value of the loss function, the second component value of the loss function, and the third component value of the loss function to determine the target loss function until the target loss function converges.

[0107] For the specific definition of the push device for the key points of the safety risk management and control of power grid field operations, please refer to the definition of the push method for the key points of the safety risk management and control of power grid field operations in the above text, which will not be repeated here. Each module in the above-mentioned push device for the key points of the safety risk management and control of power grid field operations can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0108] Based on the above Figure 1 The method shown, and Figure 2 In order to achieve the above-mentioned purpose, the embodiment of the present application further provides a computer device, which can be a personal computer, a server, a network device, etc. The computer device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to achieve the above-mentioned Figure 1 The method for pushing the key points of the power grid on-site operation safety risk management and control audit is shown.

[0109] Optionally, the computer device may also include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface may include a display screen (Display), an input unit such as a keyboard (Keyboard), etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a marked wired interface, a wireless interface (such as a Bluetooth interface, a WI-FI interface), etc.

[0110] Those skilled in the art will appreciate that the computer device structure provided in this embodiment does not limit the computer device, and may include more or fewer components, or a combination of certain components, or different component arrangements.

[0111] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages and saves the hardware and software resources of the computer device, and supports the operation of information processing programs and other software and / or programs. The network communication module is used to realize communication between the components inside the storage medium, and communication with other hardware and software in the physical device.

[0112] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform, or the embodiments of the present application can be implemented by hardware.

[0113] Those skilled in the art will appreciate that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily necessary for implementing the present application. Those skilled in the art will appreciate that the modules in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the description of the implementation scenario, or can be changed accordingly and located in one or more devices different from the present implementation scenario. The modules of the above-mentioned implementation scenario can be combined into one module, or can be further split into multiple submodules.

[0114] The above serial numbers of this application are only for description and do not represent the advantages and disadvantages of the implementation scenarios. The above disclosure is only a few specific implementation scenarios of this application, but this application is not limited to them, and any changes that can be thought of by technicians in this field should fall within the scope of protection of this application.

Claims

1. A method for pushing key points of safety risk management and control of power grid field operations, characterized in that: The method comprises: Inputting the power grid field operation text to be tested into an entity recognition model to identify the target device entity in the field operation text, wherein the entity recognition model is trained based on historical data of the power grid field operation; According to the target device entity, the security risk control and inspection key points associated with the target device entity are matched from a preset security regulations database; Determine the violation management data associated with the safety risk control audit points according to the preset historical violation database; The safety risk control inspection points are pushed to the terminal corresponding to the on-site operation text according to the violation management data.

2. The method for pushing key points of power grid on-site operation safety risk management and control audit according to claim 1 is characterized in that: The step of matching the security risk control and inspection key points associated with the target device entity from a preset security regulation database according to the target device entity includes: Matching the target device entity with a source device entity in the preset security regulations database, and determining a matching confidence level between the target device entity and the source device entity; If the matching confidence is greater than or equal to a first preset threshold, the control and inspection points corresponding to the source device entity that matches the target device entity are determined as the security risk control and inspection points associated with the target device entity.

3. The method for pushing key points of power grid on-site operation safety risk management and control audit according to claim 2 is characterized in that: The method further comprises: Performing text enhancement processing on the source device entity in the preset safety regulations database to obtain a fuzzy device entity; The fuzzy device entity is associated with the control and inspection points corresponding to the source device entity, and the source device entity in the preset safety regulations database is updated according to the fuzzy device entity.

4. The method for pushing key points of safety risk management and control of power grid on-site operations according to claim 1 is characterized in that: The method of pushing the safety risk control inspection key points to the terminal corresponding to the on-site operation text according to the violation management data includes: Classify the safety risk management and control audit points according to preset rules and determine the classification results; Sorting the safety risk control audit points according to the violation management data and the classification results to determine the data sequence; Dividing the data sequence into a first sub-sequence and a second sub-sequence according to a preset bit sequence; Pushing the safety risk management and control audit points in the first sub-sequence to the terminal corresponding to the on-site operation text in sequence; The safety risk management and control audit points in the second sub-sequence are folded into a push data group, and the first push data group is pushed to the terminal corresponding to the on-site operation text.

5. The method for pushing key points of power grid on-site operation safety risk management and control audit according to claim 4 is characterized in that: The method further comprises: If the entity recognition model fails to identify the target device entity, or fails to obtain the security risk management and inspection points, the management and inspection points in the preset safety regulations database that do not comply with the preset rules will be pushed to the terminal corresponding to the on-site operation text.

6. The method for pushing out the key points of safety risk management and control of power grid field operations according to claim 1 is characterized in that: The method further comprises: Determining a first training sample set with location labels and a second training sample set without location labels based on the historical data of the field operation of the power grid; Training a preset entity recognition model according to the first training sample set to obtain a first recognition model; Inputting the second training sample set into the first recognition model for labeling, and determining a third training sample set carrying a prediction label; Performing noise enhancement processing on the third training sample set; Training the preset entity recognition model according to the first training sample set and the third training sample set after noise enhancement processing to obtain a second recognition model; The entity recognition model is determined according to the recognition accuracy of the second recognition model on the first training sample set.

7. The method for pushing key points of safety risk management and control of power grid on-site operations according to claim 6 is characterized in that: The determining the entity recognition model according to the recognition accuracy of the second recognition model for the first training sample set includes: If the recognition accuracy is less than a second preset threshold, updating the first recognition model according to the second recognition model until the recognition accuracy is greater than or equal to the second preset threshold; If the recognition accuracy is greater than or equal to the second preset threshold, a second recognition model whose recognition accuracy is greater than or equal to the second preset threshold is determined as the entity recognition model.

8. The method for pushing key points of power grid on-site operation safety risk management and control audit according to claim 6 is characterized in that: The training of the preset entity recognition model includes: According to the preset demand information of the extracted power equipment and the training samples input into the preset entity recognition model, constructing the structure data in the form of triples, wherein the output data in the structure data is a set of labeled equipment entities pre-identified in the input data; Based on the BERT model, determining a feature vector matrix of the training sample input into the preset entity recognition model; Determine, according to the fully connected layer of the BERT model and the feature vector matrix, an entity start position label and an entity end position label of a training sample input to the preset entity recognition model; Matching a predicted device entity that conforms to the structure data according to the entity start position tag and the entity end position tag; Determine, according to the output data in the structure data, a first component of the loss function corresponding to the entity start position label, a second component of the loss function corresponding to the entity end position label, and a third component of the loss function corresponding to the prediction device entity; The first component value of the loss function, the second component value of the loss function and the third component value of the loss function are respectively calculated by the cross entropy function, and the calculated first component value of the loss function, the second component value of the loss function and the third component value of the loss function are weighted to determine the target loss function until the target loss function converges.

9. A device for pushing key points of safety risk control and inspection of power grid on-site operations, characterized in that: The device comprises: A recognition module, used for inputting the power grid field operation text to be tested into an entity recognition model, and identifying the target device entity in the field operation text, wherein the entity recognition model is trained based on the historical data of the power grid field operation; A matching module, configured to match the security risk management and control audit points associated with the target device entity from a preset security regulations database according to the target device entity; and to determine the violation management data associated with the security risk management and control audit points according to a preset historical violation database; A push module is used to push the safety risk control inspection points to the terminal corresponding to the on-site operation text according to the violation management data.

10. A computer device comprising a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, characterized in that: When the processor executes the program, it implements the method for pushing the key points of power grid field operation safety risk management and control audit as described in any one of claims 1 to 8.

Images