Work ticket-based operation risk early warning method, system and equipment

Abstract

The invention discloses a work risk early warning method, system and equipment based on a work ticket, and relates to the technical field of work risk early warning, the work ticket is acquired, word extraction is performed on the work ticket to obtain a corresponding keyword, the keyword is input into a preset violation information base, a corresponding target violation text is matched, and the work risk early warning is performed. And sequentially inputting the plurality of preset historical violation texts and the target violation text into a pre-trained text matching model to obtain a matching degree corresponding to each historical violation text, and performing risk early warning on the operation plan associated with the work ticket according to all the matching degrees. According to the invention, the technical problem that the risk early-warning evaluation efficiency is low due to the fact that the risk early-warning evaluation for the work ticket mostly depends on expert experience and a large amount of manpower and material resources need to be consumed when the risk early-warning evaluation is needed for a large number of work tickets in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of operation risk early warning, and in particular to an operation risk early warning method, system and equipment based on a work ticket. Background Art

[0002] Due to illegal operation, aging equipment, poor management and other reasons, various power accidents often occur in the power system, which seriously affect power production. In order to ensure the normal operation of the power system, certain rules and regulations are required in the actual power operation or operation process, such as the work ticket in the power operation, which is the written basis for the implementation of safety technical measures, organizational measures and the safety responsibilities of relevant personnel, and for maintenance. However, it is inevitable that there are still certain safety problems in the power system, such as fires caused by overload of power equipment and personal injury accidents caused by production personnel not operating in accordance with safety regulations.

[0003] At present, in order to reduce the probability of accidents in the power system, it is necessary to conduct risk warning assessment on work tickets, so as to make corresponding risk warning elimination measures. However, the existing risk warning assessment for work tickets mostly relies on expert experience. When a large number of work tickets require risk warning assessment, it takes a lot of manpower and material resources, resulting in low efficiency of risk warning assessment. Summary of the invention

[0004] The present invention provides a method, system and device for early warning of work risks based on work tickets, which solves the technical problem that the existing risk warning assessment for work tickets mostly relies on expert experience. When a large number of work tickets need risk warning assessment, a large amount of manpower and material resources are consumed, resulting in low efficiency of risk warning assessment.

[0005] A first aspect of the present invention provides a method for early warning of operation risks based on a work ticket, comprising:

[0006] Obtain a work ticket, extract words from the work ticket, and obtain corresponding keywords;

[0007] Input the keyword into a preset violation information database to match the corresponding target violation text;

[0008] Inputting a plurality of preset historical violation texts and the target violation text into a pre-trained text matching model in sequence, wherein the text matching model includes a sentence network and a detection network;

[0009] Performing text conversion on each of the preset historical traffic violation texts and the target traffic violation text in turn through the sentence network to obtain the target traffic violation features and the historical traffic violation features corresponding to each of the preset historical traffic violation texts;

[0010] The detection network is used to perform feature matching on the target violation feature and each of the historical violation features in turn to obtain a matching degree corresponding to each of the historical violation features, and a risk warning is issued for the work plan associated with the work ticket based on all the matching degrees.

[0011] Optionally, the step of obtaining a work ticket, extracting words from the work ticket, and obtaining corresponding keywords includes:

[0012] Based on the basic database of job plans, query the work orders according to the preset query conditions and obtain the corresponding work tickets;

[0013] Keywords in the work ticket are extracted according to a preset dictionary, wherein the keywords include unit and personnel names.

[0014] Optionally, the sentence network includes a first sentence module and a second sentence module, and the step of performing text conversion on each of the preset historical traffic violation texts and the target traffic violation text in sequence through the sentence network to obtain the target traffic violation features and the historical traffic violation features corresponding to each of the preset historical traffic violation texts includes:

[0015] Performing text conversion on the target traffic violation text by using the first sentence module to obtain target traffic violation features;

[0016] The preset historical traffic violation texts are sequentially converted by the second sentence module to obtain historical traffic violation features corresponding to each of the preset historical traffic violation texts.

[0017] Optionally, the detection network includes a first feature extraction module, a second feature extraction module and a detection module, and the specific processing process of the detection network is:

[0018] The first feature extraction module is used to fuse the input target text features and the historical text features to obtain a first fused feature, wherein the first feature extraction module includes two average pooling layers and a first feature extraction branch;

[0019] The target text feature and the historical text feature are subjected to feature fusion by the second feature extraction module to obtain a second fusion feature, wherein the second feature extraction module includes a vector network, a recurrent network, and a second feature extraction branch;

[0020] The detection module is used to perform feature matching on the first fusion feature and the second fusion feature to obtain a corresponding matching degree, wherein the detection module includes a feature fusion layer and an activation function layer connected in sequence.

[0021] Optionally, the step of fusing the input target text features and historical text features through the first feature extraction module to obtain a first fused feature includes:

[0022] The input target text feature is averaged through the average pooling layer to obtain the first text feature;

[0023] The input historical text features are processed by the average pooling layer to obtain the second text features;

[0024] A first feature extraction branch is used to perform feature fusion on the first text feature and the second text feature to obtain a first fusion feature, wherein the first feature extraction branch includes a feature fusion layer, a fully connected layer, a RELU activation layer, a fully connected layer, a RELU activation layer, a fully connected layer and a RELU activation layer connected in sequence.

[0025] Optionally, the step of fusing the target text feature and the historical text feature by the second feature extraction module to obtain a second fused feature includes:

[0026] Performing feature fusion on the target text features and the historical text features through a vector network, and outputting multiple fused feature vectors layer by layer;

[0027] Using an LSTM recurrent network to perform feature fusion and extraction operations on all the fused feature vectors, and outputting multiple third text features layer by layer;

[0028] The second feature extraction branch is used to perform feature fusion on all the third text features to obtain a second fused feature, wherein the second feature extraction branch includes a feature fusion layer, a fully connected layer and a RELU activation layer connected in sequence.

[0029] Optionally, the step of using the detection module to perform feature matching on the first fused feature and the second fused feature to obtain a corresponding matching degree includes:

[0030] Performing feature fusion on the first fused feature and the second fused feature through a feature fusion layer to obtain a third fused feature;

[0031] An activation function layer is used to perform matching analysis on the third fusion feature to obtain a corresponding matching degree.

[0032] Optionally, the step of providing risk warning for the work plan associated with the work ticket according to all the matching degrees includes:

[0033] Determining in turn whether each of the matching degrees is greater than a preset matching threshold;

[0034] If any of the matching degrees is greater than the matching threshold, a risk warning is issued for the work plan associated with the work ticket;

[0035] If each of the matching degrees is less than or equal to the matching threshold, an analysis result indicating that the operation plan associated with the work ticket is normal is generated.

[0036] A second aspect of the present invention provides a work ticket-based operation risk early warning system, comprising:

[0037] A collection module is used to obtain a work ticket, extract words from the work ticket, and obtain corresponding keywords;

[0038] A matching module, used for inputting the keyword into a preset violation information database to match the corresponding target violation text;

[0039] A comparison module, used for sequentially inputting a plurality of preset historical violation texts and the target violation text into a pre-trained text matching model, wherein the text matching model includes a sentence network and a detection network;

[0040] A conversion module, used to perform text conversion on each of the preset historical traffic violation texts and the target traffic violation text in turn through the sentence network to obtain the target traffic violation feature and the historical traffic violation features corresponding to each of the preset historical traffic violation texts;

[0041] The early warning module is used to use the detection network to perform feature matching on the target violation feature and each of the historical violation features in turn, obtain the matching degree corresponding to each of the historical violation features, and issue a risk warning for the work plan associated with the work ticket based on all the matching degrees.

[0042] A third aspect of the present invention provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the work ticket-based operation risk warning method as described in any one of the above items.

[0043] It can be seen from the above technical solutions that the present invention has the following advantages:

[0044] The present invention extracts keywords from the work ticket, inputs the keywords into a preset violation information database, matches the corresponding target violation text, and then uses a text matching model to compare the target violation text with the preset historical violation text one by one to obtain the corresponding matching degree, and performs risk warning on the work plan associated with the work ticket based on the matching degree. The work plan associated with the work ticket can be quickly evaluated for risk warning based on the historical violation text, overcoming the existing technical problem that risk warning evaluation for work tickets mostly relies on expert experience, and when a large number of work tickets need risk warning evaluation, a large amount of manpower and material resources are consumed, resulting in low efficiency of risk warning evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0046] Figure 1 A flowchart of a method for early warning of operation risks based on a work ticket provided in the first embodiment of the present invention;

[0047] Figure 2 A flowchart of a method for early warning of operation risks based on a work ticket provided in Embodiment 2 of the present invention;

[0048] Figure 3 A schematic diagram of the structure of a text matching model provided in Embodiment 2 of the present invention;

[0049] Figure 4 A structural block diagram of a work risk early warning system based on a work ticket provided in Embodiment 3 of the present invention;

[0050] Figure 5 This is a structural block diagram of a computer device provided in Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0051] The embodiments of the present invention provide a method, system and device for early warning of work risks based on work tickets, which are used to solve the technical problem that the existing risk warning assessment for work tickets mostly relies on expert experience. When a large number of work tickets need risk warning assessment, a large amount of manpower and material resources are consumed, resulting in low efficiency of risk warning assessment.

[0052] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below 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.

[0053] See also Figure 1 , Figure 1 A flowchart of the steps of a work ticket-based operation risk warning method provided in Example 1 of the present invention.

[0054] The present invention provides a method for early warning of operation risks based on a work ticket, comprising:

[0055] Step 101: Obtain a work ticket, extract words from the work ticket, and obtain corresponding keywords;

[0056] In an embodiment of the present invention, a corresponding plan sheet is queried from the basic database of the operation plan according to a preset query condition, and a work ticket associated with the plan sheet is obtained according to the plan sheet number of the plan sheet. Words are extracted from the work sheet through a preset dictionary to obtain corresponding keywords, wherein the keywords include personnel names.

[0057] Step 102: input the keyword into a preset traffic violation information database to match the corresponding target traffic violation text;

[0058] The violation information database refers to the database that records the historical violation records of each staff member.

[0059] In the embodiment of the present invention, inputting the personnel names into the preset traffic violation information can match the target traffic violation texts corresponding to the personnel names.

[0060] It should be noted that if the target violation text is not matched, a normal analysis result of the work plan associated with the work ticket is generated.

[0061] Step 103: sequentially input a plurality of preset historical violation texts and a target violation text into a pre-trained text matching model, wherein the text matching model includes a sentence network and a detection network;

[0062] In an embodiment of the present invention, a pre-trained text matching model is used to compare the target traffic violation text with a plurality of preset historical traffic violation texts one by one, wherein the text matching model includes a sentence network and a detection network.

[0063] It is worth mentioning that if there are multiple target violation texts, multiple preset historical violation texts and multiple target violation texts are sequentially input into the pre-trained text matching model. For example, the historical violation texts include the first historical text and the second historical text, and the target violation text includes the first violation text and the second violation text. The text matching model is used to sequentially match and analyze the first violation text, the first historical text, and the second historical text. The text matching model is used to sequentially match and analyze the second violation text, the first historical text, and the second historical text.

[0064] It is worth mentioning that the specific training process of the text matching model is: 1. Obtain multiple text training data, perform data preprocessing on the multiple text training data, and generate a text feature set. Among them, data preprocessing includes but is not limited to removing duplicate data entries, filling missing values, removing stop words and manual annotation; 2. Use the text feature set to input the preset initial matching model for training, and output the training matching data. 3. Calculate the training loss function value of the text feature set based on the training matching data. 4. When the training loss function value is greater than or equal to the preset standard loss function value, use the stochastic gradient descent method to adjust the model parameters of the preset initial matching model until the training loss function value is less than the preset standard loss function value. 5. When the training loss function value is less than the preset standard loss function value, a text matching model is generated.

[0065] It is worth mentioning that the stochastic gradient descent method takes the loss function of the model as the optimization target, and minimizes the loss function and improves the accuracy of the model by continuously updating the iterative model parameters according to the training data. The core idea of ​​the stochastic gradient descent algorithm is: in each iteration, a small batch of training samples (mini-batch) is randomly selected, the gradient of the loss function of the batch of samples with respect to the current model parameters is calculated, and the model parameters are updated in the opposite direction of the gradient. The process of calculating the gradient of the loss function g is as follows:

[0066]

[0067] in, is the model's predicted value for input x, and L is a function that measures the difference between the predicted value and the true value, such as mean square error or cross entropy loss. It is a small batch of data in the text feature set D.

[0068] Step 104: Perform text conversion on each preset historical traffic violation text and the target traffic violation text in turn through a sentence network to obtain the target traffic violation feature and the historical traffic violation features corresponding to each preset historical traffic violation text;

[0069] In the embodiment of the present invention, text conversion is performed on each preset historical traffic violation text and the target traffic violation text respectively through the embedding sentence network to obtain the target traffic violation features and the historical traffic violation features corresponding to each preset historical traffic violation text.

[0070] Step 105: Use the detection network to perform feature matching on the target violation feature and each historical violation feature in turn, obtain the matching degree corresponding to each historical violation feature, and issue a risk warning for the work plan associated with the work ticket based on all the matching degrees.

[0071] In an embodiment of the present invention, a detection network is used to perform feature matching on the target violation feature and each historical violation feature in turn to obtain the matching degree corresponding to each historical violation feature, select the maximum matching degree from all the matching degrees as the target matching degree, and judge whether the target matching degree is greater than a preset matching threshold. When the target matching degree is greater than the preset matching threshold, a risk warning is issued for the work plan associated with the work ticket.

[0072] In an embodiment of the present invention, the present invention extracts keywords from a work ticket, inputs the keywords into a preset violation information database, matches the corresponding target violation text, and then uses a text matching model to compare the target violation text with preset historical violation texts one by one to obtain a corresponding matching degree, and performs a risk warning on the work plan associated with the work ticket based on the matching degree. A rapid risk warning assessment can be performed on the work plan associated with the work ticket based on the historical violation text, overcoming the existing technical problem that risk warning assessments for work tickets mostly rely on expert experience, and when a large number of work tickets require risk warning assessments, a large amount of manpower and material resources are consumed, resulting in low efficiency of risk warning assessments.

[0073] See also Figure 2 , Figure 2 A flowchart of the steps of a work ticket-based operation risk warning method provided in Example 2 of the present invention.

[0074] The present invention provides a method for early warning of operation risks based on a work ticket, comprising:

[0075] Step 201: Based on the operation plan basic database, the work order is queried according to the preset query conditions to obtain the corresponding work ticket;

[0076] In an embodiment of the present invention, the controlled operation plan sheet is queried from the operation plan basic library according to the preset query conditions, and the associated work ticket is obtained according to the number of the operation plan sheet.

[0077] In another embodiment, the managed work plan sheet is queried from the work plan basic library according to preset query conditions, and the associated work ticket is obtained according to the key attributes of the work plan sheet, wherein the key attributes include but are not limited to the planned time, unit, work content, work supervisor, construction location, etc.

[0078] It should be noted that the query conditions are specifically as follows: non-control keywords of the plan sheets in the operation plan basic library are extracted according to the preset dictionary, wherein the non-control keywords include video control, operation category, major, and plan content, and the plan sheets with video control of "no", major of "biotechnology", operation category of "deficiency elimination" and plan content including "QX" are taken as non-control operation plan sheets, and the non-control operation plan sheets in the operation plan basic library are eliminated, and the control operation plan sheets are obtained as the operation plan sheets.

[0079] Step 202: extract keywords from the work ticket according to a preset dictionary, wherein the keywords include unit and personnel names.

[0080] In an embodiment of the present invention, keywords in the work ticket are extracted according to a preset dictionary, wherein the keywords include units, personnel names (work leaders and work team members).

[0081] Step 203: input the keyword into a preset traffic violation information database to match the corresponding target traffic violation text;

[0082] In the embodiment of the present invention, keywords are input into a preset violation information database to match target violation texts of various personnel.

[0083] It should be noted that the violation information database stores one year's violation records of each work leader and work team member.

[0084] Step 204: sequentially input a plurality of preset historical violation texts and a target violation text into a pre-trained text matching model, wherein the text matching model includes a sentence network and a detection network;

[0085] Historical violation text refers to the violation information extracted from historical accident reports.

[0086] In an embodiment of the present invention, multiple preset historical violation texts and target violation texts are sequentially input into a pre-trained text matching model, wherein the text matching model includes a sentence network and a detection network. If there are multiple target violation texts, the text matching model is used to compare each target violation text with each historical violation text one by one.

[0087] Step 205: Perform text conversion on each preset historical traffic violation text and the target traffic violation text in turn through a sentence network to obtain the target traffic violation feature and the historical traffic violation features corresponding to each preset historical traffic violation text;

[0088] Further, the sentence network includes a first sentence module and a second sentence module, and step 205 includes the following sub-steps:

[0089] S11, performing text conversion on the target traffic violation text through the first sentence module to obtain the target traffic violation features;

[0090] It should be noted that, see Figure 3 As shown, the first sentence module and the second sentence module are both sentence models, and the sentence model includes a vector conversion module, a position encoding module and 6 encoding modules.

[0091] In the embodiment of the present invention, a sentence model is used to perform text conversion operation on the target traffic violation text to obtain the target traffic violation features.

[0092] S12. Perform text conversion on the preset historical traffic violation texts in sequence through the second sentence module to obtain historical traffic violation features corresponding to each preset historical traffic violation text.

[0093] In the embodiment of the present invention, a sentence model is used to perform text conversion operations on preset historical traffic violation texts in sequence to obtain historical traffic violation features corresponding to each preset historical traffic violation text.

[0094] Step 206: Use the detection network to perform feature matching on the target violation feature and each historical violation feature in turn, obtain the matching degree corresponding to each historical violation feature, and issue a risk warning for the work plan associated with the work ticket based on all the matching degrees.

[0095] Further, step 206 includes the following sub-steps:

[0096] S21, using the detection network to perform feature matching on the target violation feature and each historical violation feature in turn, and obtaining the matching degree corresponding to each historical violation feature;

[0097] In an embodiment of the present invention, a detection network is used to perform feature matching on a target traffic violation feature and each historical traffic violation feature in turn to obtain a matching degree corresponding to each historical traffic violation feature. For example, the historical traffic violation feature includes a first historical feature and a second historical feature. The detection network is used to perform feature matching on the first historical feature and the target traffic violation feature to obtain a first matching degree, and then the detection network is used to perform feature matching on the second historical feature and the target traffic violation feature to obtain a second matching degree.

[0098] It should be noted that, see Figure 3 As shown, the detection network includes a first feature extraction module, a second feature extraction module and a detection module. The specific processing process of the detection network is:

[0099] A1. Fusing the input target text features and historical text features through a first feature extraction module to obtain a first fused feature, wherein the first feature extraction module includes two average pooling layers and a first feature extraction branch;

[0100] Furthermore, A1 includes the following sub-steps:

[0101] A11, performing mean processing on the input target text feature through an average pooling layer to obtain a first text feature;

[0102] In the embodiment of the present invention, a pooling operation is performed on the input target text feature through an average pooling layer to obtain a first text feature.

[0103] A12, performing mean processing on the input historical text features through an average pooling layer to obtain a second text feature;

[0104] In the embodiment of the present invention, a pooling operation is performed on the input historical text feature through an average pooling layer to obtain a second text feature.

[0105] A13. Use the first feature extraction branch to fuse the first text feature and the second text feature to obtain a first fused feature, wherein the first feature extraction branch includes a feature fusion layer, a fully connected layer, a RELU activation layer, a fully connected layer, a RELU activation layer, a fully connected layer and a RELU activation layer connected in sequence.

[0106] In an embodiment of the present invention, the first text feature and the second text feature are feature fused through a first feature extraction branch to obtain a first fused feature, wherein the first feature extraction branch includes a cat function layer, a fully connected layer, a RELU activation layer, a fully connected layer, a RELU activation layer, a fully connected layer and a RELU activation layer connected in sequence.

[0107] A2, performing feature fusion on the target text feature and the historical text feature through a second feature extraction module to obtain a second fusion feature, wherein the second feature extraction module includes a vector network, a recurrent network and a second feature extraction branch;

[0108] Furthermore, A2 includes the following sub-steps:

[0109] A21. Perform feature fusion of the target text features and the historical text features through a vector network, and output multiple fused feature vectors layer by layer;

[0110] In an embodiment of the present invention, a vector network is used to fuse target text features and historical text features, and multiple fused feature vectors are output layer by layer, wherein the vector network includes multiple word vector modules.

[0111] A22, using LSTM recurrent network to perform feature fusion and extraction operations on all fused feature vectors, and output multiple third text features layer by layer;

[0112] In an embodiment of the present invention, an LSTM recurrent network is used to perform feature fusion and extraction operations on all fused feature vectors, and multiple third text features are output layer by layer, wherein the LSTM recurrent network includes multiple LSTM modules and 4 fully connected layers.

[0113] A23. Perform feature fusion on all third text features through the second feature extraction branch to obtain a second fused feature, wherein the second feature extraction branch includes a feature fusion layer, a fully connected layer, and a RELU activation layer connected in sequence.

[0114] In an embodiment of the present invention, the four third text features are fused through the second feature extraction branch to obtain a second fused feature, wherein the second feature extraction branch includes a feature fusion layer, a fully connected layer and a RELU activation layer connected in sequence.

[0115] A3. Use a detection module to perform feature matching on the first fusion feature and the second fusion feature to obtain a corresponding matching degree, wherein the detection module includes a feature fusion layer and an activation function layer connected in sequence.

[0116] Furthermore, A3 includes the following sub-steps:

[0117] A31. Fusing the first fused feature and the second fused feature through a feature fusion layer to obtain a third fused feature.

[0118] In the embodiment of the present invention, the first fusion feature and the second fusion feature are fused through a cat function layer to obtain a third fusion feature.

[0119] A32. Use the activation function layer to perform matching analysis on the third fusion feature to obtain the corresponding matching degree.

[0120] In the embodiment of the present invention, a sigmoid activation layer is used to perform matching analysis on the third fusion feature to obtain a corresponding matching degree.

[0121] S22, determining in turn whether each matching degree is greater than a preset matching threshold;

[0122] In the embodiment of the present invention, it is determined whether each matching degree is greater than a preset matching threshold.

[0123] In another embodiment, the maximum value of each matching degree is selected as the target matching degree, and it is determined whether the target matching degree is greater than a preset matching threshold.

[0124] S23. If any matching degree is greater than the matching threshold, a risk warning is issued for the work plan associated with the work ticket;

[0125] In an embodiment of the present invention, if any matching degree is greater than 90%, the work ticket corresponding to the plan number, management unit, work unit, specialty, risk problem, operation type, operation steps, voltage level, risk basis, and similar historical accident event details are output. Among them, the risk basis includes the historical violation of the operation planner. The associated accident events mainly include the violation code corresponding to the accident event, violation description, accident process, direct and indirect causes of the accident, measures, exposure issues, etc.

[0126] In another embodiment, if the target matching degree is greater than 90%, a risk warning is issued for the work plan associated with the work ticket. The specific process of risk warning is: obtaining the phone number of the person in charge of the work plan, calling the SMS platform to send a warning SMS to complete the warning notification.

[0127] S24. If all matching degrees are less than or equal to the matching threshold, a normal analysis result of the operation plan associated with the work ticket is generated.

[0128] In the embodiment of the present invention, if each matching degree is less than or equal to 90%, it is determined that there is no abnormal analysis result in the work plan associated with the work ticket.

[0129] In another embodiment, if the target matching degree is less than or equal to 90%, an analysis result is generated indicating that the work plan associated with the work ticket has no abnormalities.

[0130] In an embodiment of the present invention, the present invention extracts keywords from a work ticket, inputs the keywords into a preset violation information database, matches the corresponding target violation text, and then uses a text matching model to compare the target violation text with preset historical violation texts one by one to obtain a corresponding matching degree, and performs a risk warning on the work plan associated with the work ticket based on the matching degree. A rapid risk warning assessment can be performed on the work plan associated with the work ticket based on the historical violation text, overcoming the existing technical problem that risk warning assessments for work tickets mostly rely on expert experience, and when a large number of work tickets require risk warning assessments, a large amount of manpower and material resources are consumed, resulting in low efficiency of risk warning assessments.

[0131] See also Figure 4 , Figure 4 This is a structural block diagram of a work ticket-based operation risk warning system provided in Example 3 of the present invention.

[0132] The present invention provides a work ticket-based operation risk early warning system, comprising:

[0133] The acquisition module 301 is used to obtain a work ticket, extract words from the work ticket, and obtain corresponding keywords;

[0134] A matching module 302, for inputting keywords into a preset traffic violation information database to match corresponding target traffic violation texts;

[0135] A comparison module 303 is used to sequentially input a plurality of preset historical violation texts and a target violation text into a pre-trained text matching model, wherein the text matching model includes a sentence network and a detection network;

[0136] A conversion module 304 is used to perform text conversion on each preset historical traffic violation text and the target traffic violation text in sequence through a sentence network to obtain the target traffic violation feature and the historical traffic violation features corresponding to each preset historical traffic violation text;

[0137] The early warning module 305 is used to use the detection network to perform feature matching on the target violation feature and each historical violation feature in turn, obtain the matching degree corresponding to each historical violation feature, and issue a risk warning to the work plan associated with the work ticket based on all the matching degrees.

[0138] Furthermore, the acquisition module 301 includes:

[0139] The query submodule is used to query the work order based on the job plan basic library and the preset query conditions to obtain the corresponding work ticket;

[0140] The extraction submodule is used to extract keywords in the work ticket according to a preset dictionary, wherein the keywords include unit and personnel names.

[0141] Furthermore, the conversion module 304 includes:

[0142] A first conversion submodule, configured to perform text conversion on the target traffic violation text through a first sentence module to obtain a target traffic violation feature;

[0143] The second conversion submodule is used to perform text conversion on the preset historical traffic violation texts in sequence through the second sentence module to obtain historical traffic violation features corresponding to each preset historical traffic violation text.

[0144] Furthermore, the early warning module 305 includes:

[0145] The analysis submodule is used to determine in turn whether each matching degree is greater than a preset matching threshold;

[0146] The first warning submodule is used to issue a risk warning to the work plan associated with the work ticket if any matching degree is greater than a matching threshold;

[0147] The second early warning submodule is used to generate an analysis result indicating that the operation plan associated with the work ticket is normal if all matching degrees are less than or equal to the matching threshold.

[0148] Furthermore, the detection network includes a first feature extraction module, a second feature extraction module and a detection module, and the specific processing process of the detection network is:

[0149] The first feature extraction module is used to fuse the input target text features and the historical text features to obtain a first fused feature, wherein the first feature extraction module includes two average pooling layers and a first feature extraction branch;

[0150] The target text feature and the historical text feature are subjected to feature fusion by a second feature extraction module to obtain a second fusion feature, wherein the second feature extraction module includes a vector network, a recurrent network and a second feature extraction branch;

[0151] A detection module is used to perform feature matching on the first fusion feature and the second fusion feature to obtain a corresponding matching degree, wherein the detection module includes a feature fusion layer and an activation function layer connected in sequence.

[0152] Furthermore, the step of fusing the input target text features and historical text features by a first feature extraction module to obtain a first fused feature includes:

[0153] The input target text feature is averaged through the average pooling layer to obtain the first text feature;

[0154] The input historical text features are processed by the average pooling layer to obtain the second text features;

[0155] The first feature extraction branch is used to fuse the first text feature and the second text feature to obtain a first fused feature, wherein the first feature extraction branch includes a feature fusion layer, a fully connected layer, a RELU activation layer, a fully connected layer, a RELU activation layer, a fully connected layer and a RELU activation layer connected in sequence.

[0156] Furthermore, the step of fusing the target text features and the historical text features through a second feature extraction module to obtain a second fused feature includes:

[0157] The target text features and historical text features are fused through a vector network, and multiple fused feature vectors are output layer by layer;

[0158] The LSTM recurrent network is used to perform feature fusion and extraction operations on all fused feature vectors, and multiple third text features are output layer by layer;

[0159] The second feature extraction branch is used to perform feature fusion on all third text features to obtain a second fused feature, wherein the second feature extraction branch includes a feature fusion layer, a fully connected layer, and a RELU activation layer connected in sequence.

[0160] Furthermore, the step of using a detection module to perform feature matching on the first fusion feature and the second fusion feature to obtain a corresponding matching degree includes:

[0161] The first fusion feature and the second fusion feature are fused through a feature fusion layer to obtain a third fusion feature;

[0162] The activation function layer is used to perform matching analysis on the third fusion feature to obtain the corresponding matching degree.

[0163] See also Figure 5 , Figure 5 This is a structural block diagram of a computer device provided in Embodiment 4 of the present invention.

[0164] An electronic device according to an embodiment of the present invention includes: a memory 401 and a processor 402, wherein the memory 402 stores a computer program; when the computer program is executed by the processor 402, the processor 402 executes a work ticket-based operation risk warning method as in any of the above embodiments.

[0165] The memory 401 may be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk or a ROM. The memory 401 has a storage space 403 for a program code 413 for executing any method step in the above method. For example, the storage space 403 for the program code may include individual program codes 413 for implementing the various steps in the above method, respectively. These program codes may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards or floppy disks. The program code may be compressed, for example, in an appropriate form. When these codes are run by a computing and processing device, the computing and processing device performs the various steps in the method described above. These program codes may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards or floppy disks. The program code may be compressed, for example, in an appropriate form. When these codes are executed by a computing and processing device, they cause the computing and processing device to execute the various steps in the work ticket-based operation risk warning method described above.

[0166] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0167] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0168] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0169] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0170] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0171] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for early warning of operation risks based on work tickets, It is characterized in that include: Obtain a work ticket, extract words from the work ticket, and obtain corresponding keywords; Input the keyword into a preset violation information database to match the corresponding target violation text; Inputting a plurality of preset historical violation texts and the target violation text into a pre-trained text matching model in sequence, wherein the text matching model includes a sentence network and a detection network; Performing text conversion on each of the preset historical traffic violation texts and the target traffic violation text in turn through the sentence network to obtain the target traffic violation features and the historical traffic violation features corresponding to each of the preset historical traffic violation texts; The detection network is used to perform feature matching on the target violation feature and each of the historical violation features in turn to obtain a matching degree corresponding to each of the historical violation features, and a risk warning is issued for the work plan associated with the work ticket based on all the matching degrees.

2. The operation risk early warning method based on work ticket according to claim 1, It is characterized in that The step of obtaining a work ticket, extracting words from the work ticket, and obtaining corresponding keywords includes: Based on the basic database of job plans, query the work orders according to the preset query conditions and obtain the corresponding work tickets; Keywords in the work ticket are extracted according to a preset dictionary, wherein the keywords include unit and personnel names.

3. The operation risk early warning method based on work ticket according to claim 1, It is characterized in that The sentence network includes a first sentence module and a second sentence module. The step of performing text conversion on each of the preset historical traffic violation texts and the target traffic violation text in turn through the sentence network to obtain the target traffic violation features and the historical traffic violation features corresponding to each of the preset historical traffic violation texts includes: Performing text conversion on the target traffic violation text by using the first sentence module to obtain target traffic violation features; The preset historical traffic violation texts are sequentially converted by the second sentence module to obtain historical traffic violation features corresponding to each of the preset historical traffic violation texts.

4. The operation risk early warning method based on work ticket according to claim 1, It is characterized in that The detection network includes a first feature extraction module, a second feature extraction module and a detection module. The specific processing process of the detection network is as follows: The first feature extraction module is used to fuse the input target text features and the historical text features to obtain a first fused feature, wherein the first feature extraction module includes two average pooling layers and a first feature extraction branch; The target text feature and the historical text feature are subjected to feature fusion by the second feature extraction module to obtain a second fusion feature, wherein the second feature extraction module includes a vector network, a recurrent network, and a second feature extraction branch; The detection module is used to perform feature matching on the first fusion feature and the second fusion feature to obtain a corresponding matching degree, wherein the detection module includes a feature fusion layer and an activation function layer connected in sequence.

5. The method for early warning of operation risks based on work tickets according to claim 4, It is characterized in that The step of fusing the input target text features and historical text features by the first feature extraction module to obtain a first fused feature includes: The input target text feature is averaged through the average pooling layer to obtain the first text feature; The input historical text features are processed by the average pooling layer to obtain the second text features; A first feature extraction branch is used to perform feature fusion on the first text feature and the second text feature to obtain a first fusion feature, wherein the first feature extraction branch includes a feature fusion layer, a fully connected layer, a RELU activation layer, a fully connected layer, a RELU activation layer, a fully connected layer and a RELU activation layer connected in sequence.

6. The method for early warning of operation risks based on work tickets according to claim 4, It is characterized in that The step of fusing the target text feature and the historical text feature by the second feature extraction module to obtain a second fused feature comprises: Performing feature fusion on the target text features and the historical text features through a vector network, and outputting multiple fused feature vectors layer by layer; Using an LSTM recurrent network to perform feature fusion and extraction operations on all the fused feature vectors, and outputting multiple third text features layer by layer; The second feature extraction branch is used to perform feature fusion on all the third text features to obtain a second fused feature, wherein the second feature extraction branch includes a feature fusion layer, a fully connected layer and a RELU activation layer connected in sequence.

7. The method for early warning of operation risks based on work tickets according to claim 4, It is characterized in that The step of using the detection module to perform feature matching on the first fusion feature and the second fusion feature to obtain a corresponding matching degree includes: Performing feature fusion on the first fused feature and the second fused feature through a feature fusion layer to obtain a third fused feature; An activation function layer is used to perform matching analysis on the third fusion feature to obtain a corresponding matching degree.

8. The method for early warning of operation risks based on work tickets according to claim 1, It is characterized in that The step of providing risk warning for the work plan associated with the work ticket according to all the matching degrees includes: Determining in turn whether each of the matching degrees is greater than a preset matching threshold; If any of the matching degrees is greater than the matching threshold, a risk warning is issued for the work plan associated with the work ticket; If each of the matching degrees is less than or equal to the matching threshold, an analysis result indicating that the operation plan associated with the work ticket is normal is generated.

9. A work risk early warning system based on work tickets, It is characterized in that include: A collection module is used to obtain a work ticket, extract words from the work ticket, and obtain corresponding keywords; A matching module, used for inputting the keyword into a preset violation information database to match the corresponding target violation text; A comparison module, used for sequentially inputting a plurality of preset historical violation texts and the target violation text into a pre-trained text matching model, wherein the text matching model includes a sentence network and a detection network; A conversion module, used to perform text conversion on each of the preset historical traffic violation texts and the target traffic violation text in sequence through the sentence network to obtain the target traffic violation feature and the historical traffic violation features corresponding to each of the preset historical traffic violation texts; The early warning module is used to use the detection network to perform feature matching on the target violation feature and each of the historical violation features in turn, obtain the matching degree corresponding to each of the historical violation features, and issue a risk warning for the work plan associated with the work ticket based on all the matching degrees.

10. An electronic device, It is characterized in that It includes a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the work ticket-based operation risk warning method as described in any one of claims 1-8.

Images