Transmission line work ticket intelligent invoicing process automation system based on cloud computing

Abstract

The present invention relates to a cloud computing-based intelligent invoicing process automation system for power transmission line work tickets, and relates to the technical field of cloud computing. According to a preset work ticket review process, an approval person is automatically assigned and a notification is sent, a text description in the work ticket is processed, key information is extracted, and a classification model is constructed to automatically identify risk points in work tasks, and whether safety measures are compliant is judged, and image recognition and processing are performed on pictures in the work ticket to achieve automated equipment status detection. A work panel cabinet is selected on a drawn actual panel cabinet position map of a substation, and the position where the point is made and the position where no operation is performed are distinguished by the depth of color. According to the relative position relationship between the panel cabinets, the adjacent panel cabinets of a certain panel cabinet selected by the user are automatically identified, the system is docked with a real-time monitoring device, line status information is obtained in real time, and relevant information is fed back to a work ticket filling person to ensure the accuracy and safety of the work ticket.

Description

Technical Field

[0001] The present invention relates to the technical field of cloud computing, and in particular to a cloud computing-based intelligent invoicing process automation system for power transmission line work tickets. Background Art

[0002] With the development of cloud computing, facing various difficulties and challenges, the power industry is currently facing the problem of how to use cloud computing technology and artificial intelligence technology to realize the automation and intelligence of the transmission line work ticket issuance process. Summary of the invention

[0003] In view of the technical problems existing in the prior art, the present invention provides a cloud computing-based transmission line work ticket intelligent invoicing process automation system, which performs image recognition and processing on the pictures in the work ticket to realize automatic equipment status detection, so as to solve the problems raised in the above background technology.

[0004] The technical solution of the present invention to solve the above technical problems is as follows: a cloud computing-based intelligent invoicing process automation system for power transmission line work tickets, which specifically includes a process control module, an information analysis module, a risk point identification module, an image recognition and processing module, a panel cabinet location module, a panel cabinet selection module, and a real-time monitoring and feedback module;

[0005] Process control module: used to automatically assign approval personnel and send notifications according to the preset work ticket review process, so as to realize the automatic circulation and approval process of work tickets;

[0006] Information analysis module: used to perform natural language processing on the text description in the work ticket, extract key information, and conduct intelligent analysis;

[0007] Risk point identification module: used to build a classification model based on the information analysis module, automatically identify risk points in work tasks, and determine whether safety measures are compliant;

[0008] Image recognition and processing module: used to perform image recognition and processing on the pictures in the work ticket to realize automatic equipment status detection;

[0009] Cabinet location module: used to select the cabinet to be operated on the actual cabinet location map of the substation, and distinguish the selected location from the location without operation by the depth of color;

[0010] The panel cabinet selection module is used to automatically identify adjacent panel cabinets when the user selects a panel cabinet based on the relative position relationship between the panel cabinets, and generate and display detailed information of these panel cabinets;

[0011] Real-time monitoring and feedback module: used to connect the system with real-time monitoring equipment, obtain line status information in real time, and feed back relevant information to the person filling out the work ticket to ensure the accuracy and security of the work ticket.

[0012] In a preferred embodiment, the process control module automatically assigns approval personnel and sends notifications according to the preset work ticket review process to achieve automatic circulation and approval process of the work ticket, which specifically includes the following steps:

[0013] Step 1: Preset the review process: The administrator presets the review process of the work ticket in the system, including each approval link and the corresponding reviewers, and defines the roles as the initial reviewer, the intermediate reviewer, and the final reviewer. The reviewers are designated by roles to reduce the operating costs when the process is changed;

[0014] Step 2, Work Ticket Submission: After completing the preset review process in step 1, when the user submits a new work ticket, the system receives the work ticket information and sets its status to pending approval. According to the preset review process, the next approval personnel are automatically assigned;

[0015] Step 3, Approver notification: After completing step 2, the system will send a notification to the staff member based on the assigned approver information, through email, SMS and instant messaging tools, to remind the approver to perform the approval operation;

[0016] Step 4, approval process: After completing the approval personnel notification in step 3, log in to the system for approval. The system should provide a corresponding interface to display the work ticket information and related content to be approved. By checking the work content and safety measures information, the approval, rejection, and refusal approval operations can be performed according to the actual situation.

[0017] Step 5, automatic transfer: After completing the approval process in step 4, the system automatically transfers the work ticket to the next link's approver based on the approval result. When the approval of a certain link is rejected or rejected, the system notifies the submitter and provides an opportunity to modify it;

[0018] Step 6, complete the approval: After completing the automatic flow in step 5, when the work ticket successfully passes all approval links, the system will set the work ticket status to completed, notify the staff, and generate approval records and related reports for tracking and counting the approval status.

[0019] In a preferred embodiment, the information analysis module performs natural language processing on the text description in the work ticket, extracts key information, and performs intelligent analysis, which specifically includes the following steps:

[0020] Step 1: Text preprocessing: Preprocess the text description, segment the text description by words, divide the sentence into individual phrases, and remove special characters and punctuation marks to reduce interference with subsequent processing;

[0021] Step 2, entity recognition: After completing the text preprocessing in step 1, use named entity recognition technology to identify key entities in the text, and use the support vector machine algorithm to classify and annotate the corpus containing key entities such as work ticket number, work content, and work time. The specific calculation formula is as follows:

[0022] y h =sign(w T x+b)

[0023] Among them, y h represents the classification result, w represents the weight vector, T represents the parameter, x represents the feature vector, b represents the bias term, and sign represents the sign function, which maps the result of the linear combination to the prediction result of the category label;

[0024] Step 3, information extraction: After completing the entity recognition in step 2, extract key information from the preprocessed and entity recognized text according to the predefined template, use keyword matching technology to extract work content and working time information, and perform text classification and information retrieval by comparing the similarity between the two texts. The specific calculation formula is:

[0025]

[0026] Among them, Q represents the similarity, A and B represent the vector representation of two texts respectively, "·" represents the inner product operation, and "|" represents the modulus of the vector.

[0027] In a preferred embodiment, the risk point identification module, after extracting key information of the work ticket based on the information module, uses the naive Bayes algorithm to build a classification model, automatically identifies risk points in the work task, and determines whether the safety measures are compliant, specifically including the following steps:

[0028] Step 1: Feature extraction: Collect annotated work task description data, including work type, equipment name, work area, and keyword information, convert the text data into a feature vector representation, and use the bag-of-words model to count the frequency of each word in each sample to form a feature vector, including:

[0029] Let D = {d1, d2, ..., d n} represents the sample set, where d i represents the i-th sample, V = {w1,w2,...,w m} represents the set of all words, where w jRepresents the jth word. For a sample d, count the number of occurrences of each word in V to obtain a vector x of length m i =(x i1 ,x i2 ,...,x im ), where x ij Represents word w j In sample d i The number of times it appears in

[0030] Stack the word vectors of all samples together to get an n×m feature matrix X, where each row represents the word vector of a sample and each column represents the frequency of word occurrence;

[0031] Step 2, classification model: After completing the feature extraction in step 1, divide the labeled data set into a training set and a test set. Use the training set to train the naive Bayes classifier, define the features to be independent of each other, and perform classification based on the conditional probability estimation of the features. The specific steps are as follows:

[0032] Step 201, for C categories, D represents the training set, X represents the feature matrix, y represents the category label corresponding to the training set, calculate the prior probability P (y = c), the specific calculation formula is as follows:

[0033]

[0034] Among them, I(y i =c) represents the indicator function, when y i =c is 1, otherwise it is 0, and n represents the total number of samples in the training set;

[0035] Step 202: For each feature x of the feature matrix X j , calculate the conditional probability distribution P(x j |y=c), the specific calculation formula is as follows:

[0036]

[0037] Among them, I(x ij =1andy i =c) represents the indicator function, when x ij =1andy i =c is 1, otherwise it is 0; α is the Laplace smoothing coefficient to avoid the probability of 0; |X| represents the total number of feature dimensions.

[0038] Step 203: For a new sample feature vector x, calculate the posterior probability P(y=c|x) that the sample belongs to each category. The specific calculation formula is as follows:

[0039]

[0040] Among them, m represents the total number of feature dimensions;

[0041] Step 204: Select the category with the maximum posterior probability as the prediction result. The specific calculation formula is as follows:

[0042]

[0043] in, Indicates the category labels corresponding to the training set, including risk points and non-risk points;

[0044] Step 3, model application: After completing the classification model in step 2, use the trained naive Bayes classification model to predict the risk points of the new work task description, convert the new description into feature vector form, and use the model to perform classification prediction to determine whether the work task has risk points.

[0045] In a preferred embodiment, the image recognition and processing module performs image recognition and processing on the image in the work ticket to realize automatic equipment status detection, which specifically includes the following steps:

[0046] Step 1: Image classification: Use the target detection model to perform entity recognition of key elements of equipment and lines in the pictures in the work ticket. Through training and reasoning, identify the objects of interest, and use the loss function Focal Loss to solve the problem of category imbalance in target detection. The specific calculation formula is as follows:

[0047] FL(p_t)=-(1-p_t) gamma ×log(p_t)

[0048] Among them, p_t represents the probability of the model predicting the target, and gamma represents an adjustable parameter;

[0049] When there are few target categories, p_t is small, (1-p_t) gamma Close to 1, the weight of the loss function is small;

[0050] When there are many target categories, p_t is large, (1-p_t) gamma Close to 0, the weight of the loss function is large;

[0051] Step 2: Device status detection: After completing the image classification in step 1, the input of the device status detection is an image X. By stacking multiple convolutional layers, pooling layers, and fully connected layers, a deep convolutional neural network is constructed to achieve the task of device status detection. The specific steps are as follows:

[0052] Step 201, the specific calculation formula of the convolution layer is as follows:

[0053] Z_L=convolve(A_L,W_L)+b_L

[0054] A_L+1=f(Z_L)

[0055] Among them, Z_L represents the output of the neuron, convolve() represents the convolution operation, A_L represents the input feature map, W_L represents the convolution kernel, b_L represents the bias term, L represents the number of layers, and f represents the activation function;

[0056] Step 202, the specific calculation formula of the pooling layer is as follows:

[0057] A_L+1=pool(A_L)

[0058] Among them, A table L represents the input feature map, pool() represents the pooling operation, and L represents the number of layers;

[0059] Step 203, fully connected layer: connect each neuron with all neurons in the previous layer, perform linear transformation through the weight matrix, and introduce nonlinear mapping with the activation function f. The specific calculation formula is as follows:

[0060] Z_L=W_L×A_L+b_L

[0061] A_L+1=f(Z_L)

[0062] Y=g(Z_L)

[0063] Among them, A_L represents the input vector, W_L represents the weight matrix, b_L represents the bias term, L represents the number of layers, and g represents the activation function.

[0064] In a preferred embodiment, the panel cabinet position module selects the panel cabinet to be operated on the drawn actual panel cabinet position map of the substation, and distinguishes the position to be operated from the position not operated by the depth of color, specifically including the following steps:

[0065] Step 1: Collect information: Obtain the floor plan, circuit diagram and panel layout diagram related to the substation from the design documents and on-site survey. Based on the collected information, draw the basic framework of the substation on the drawing software, including the outer boundary, main buildings, roads and surrounding environment. According to the panel layout and circuit diagram, use different symbols to mark the main equipment and lines, including transformers, circuit breakers and disconnectors for distinction;

[0066] Step 2: Draw the actual location map of the substation cabinets: After collecting information in step 1, use rectangles to represent the cabinets according to the cabinet layout diagram, draw the location of each cabinet on the drawing software, and mark each cabinet with a unique number, including the cabinet model, capacity, and detailed information on the purpose;

[0067] Step 3. Select the working cabinet: After completing step 2 and drawing the actual cabinet location map of the substation, use computer drawing software to mark the working cabinet with an ellipse on the actual cabinet location map, including:

[0068] For the cabinets that have been selected for operation, the color of the mark will be darkened to distinguish them from the cabinets that have not been operated;

[0069] For the screen cabinets that do not need to be worked on temporarily, add instructions on the labels, including the words "out of service" and "maintenance", to help staff clearly understand the working status of each screen cabinet.

[0070] In a preferred embodiment, the panel cabinet selection module imports the actual panel cabinet location map of the substation into the computer-aided software, and enters detailed information for each panel cabinet, including the panel cabinet name, model, capacity, and purpose, and records the relative position relationship between the panel cabinets. According to the relative position relationship between the panel cabinets, when the user selects a panel cabinet, the adjacent panel cabinets are automatically identified, and the detailed information of these panel cabinets is generated and displayed. The specific steps are as follows:

[0071] Step 1. Determine the position coordinates of the panel cabinet: Determine the position coordinates of each panel cabinet in the substation layout diagram, using a two-dimensional coordinate system with the panel cabinet in the upper left corner as the origin, the horizontal direction as the X axis, and the vertical direction as the Y axis;

[0072] Step 2, calculate the relative position: After completing step 1 to determine the position coordinates of the screen cabinet, for a screen cabinet A selected by the user, traverse all other screen cabinets B and calculate the relative position relationship between screen cabinets A and B. The Manhattan distance measurement method is used for calculation. The specific calculation formula is as follows:

[0073] distance=|x_B-x_A|+|y_B-y_A|

[0074] Among them, (x_A, y_A) represents the coordinates of cabinet A, (x_B, y_B) represents the coordinates of cabinet B, and distance represents the distance between the two cabinets;

[0075] Step 3, filter adjacent cabinets: After completing step 2 to calculate the relative position, filter out cabinets adjacent to cabinet A by setting a minimum distance threshold. When the distance of cabinet B is less than the threshold, it is determined that cabinet B is adjacent to cabinet A.

[0076] Step 4, generate detailed information: After completing step 3 to filter the adjacent cabinets, obtain their detailed information in the database according to their specific cabinet numbers, including model, capacity, and purpose, and display this information to the user.

[0077] In a preferred embodiment, the real-time monitoring and feedback module connects the system with the real-time monitoring device, obtains line status information in real time, and feeds back relevant information to the person filling out the work ticket to ensure the accuracy and security of the work ticket. The specific steps are as follows:

[0078] Step 1, obtain line status information: interface the sensor with the system, set a periodic time interval through polling, query once every 1 second, send a request to the monitoring device in each time interval, and obtain the response data, and obtain the current, voltage, temperature and humidity information of the line status from the monitoring device in real time;

[0079] Step 2: Line anomaly detection: After completing step 1 to obtain line status information, the current, voltage, temperature and humidity at a certain moment in the future are predicted through historical data analysis. The predicted value is used as a reference to dynamically adjust the threshold. Anomaly detection is performed according to the preset threshold to identify whether there is an abnormality in the line. The specific steps include the following:

[0080] Step 201: Use exponential weighted moving average to calculate the predicted value and adjust the threshold. Select a smoothing factor α to represent the weight ratio of the most recent data. Perform weighted moving average calculation on the historical data in chronological order. For each time point t, the predicted value Y_t is obtained by weighted average of the historical data and the current data. The specific calculation formula is as follows:

[0081] Y_t=α×X_t+(1-α)×Y_{t-1}

[0082] Where Y_t represents the exponentially weighted moving average at time point t, X_t represents the data at time point t, Y_{t-1} represents the exponentially weighted moving average at the previous time point, and α represents the smoothing factor, which ranges from (0<α<1);

[0083] Step 202: Set an adjustable parameter K to adjust the sensitivity of the threshold. Add K times the standard deviation to the average value as the threshold. The specific calculation formula is as follows:

[0084]

[0085]

[0086] Among them, S 上 represents the upper threshold, S 下 represents the lower threshold, represents the mean, S 标 represents standard deviation;

[0087] At each new time point, the current data is compared with the threshold to determine whether there is an abnormality on the line;

[0088] Step 3, triggering the work ticket process: After completing the line abnormality detection in step 2, the work ticket process is triggered according to the abnormal line status detected, and the relevant line status information is passed to the person filling out the work ticket to remind him to initiate the corresponding work ticket;

[0089] Step 4, filling out the work ticket: After completing step 3 to trigger the work ticket process, the person filling out the work ticket fills out the work ticket based on the received line status information, accurately describing the work content, work scope and safety precautions.

[0090] The beneficial effects of the present invention are as follows: according to the preset work ticket review process, approval personnel are automatically assigned and notifications are sent, the text description in the work ticket is processed, key information is extracted, and by building a classification model, risk points in the work task are automatically identified to determine whether safety measures are compliant. Image recognition and processing are performed on the pictures in the work ticket to achieve automated equipment status detection, and the panel cabinets in operation are selected on the drawn actual panel cabinet position map of the substation, and the depth of color is used to distinguish the locations where the points are clicked and the locations where no operations are performed. According to the relative position relationship between the panel cabinets, the adjacent panel cabinets of a panel cabinet selected by the user are automatically identified, the system is connected to the real-time monitoring device, the line status information is obtained in real time, and the relevant information is fed back to the person filling out the work ticket to ensure the accuracy and safety of the work ticket. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] Figure 1 is a system flow chart of the present invention;

[0092] Figure 2 It is a structural block diagram of the present invention. DETAILED DESCRIPTION

[0093] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0094] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, "plurality" means two or more, unless otherwise clearly and specifically defined.

[0095] In the description of the present application, the term "for example" is used to mean "used as an example, illustration or description". Any embodiment described as "for example" in the present application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present application.

[0096] Example 1

[0097] This embodiment provides Figure 2 The cloud computing-based intelligent invoicing process automation system for power transmission line work tickets shown in the figure specifically includes a process control module, an information analysis module, a risk point identification module, an image recognition and processing module, a panel cabinet location module, a panel cabinet selection module, and a real-time monitoring and feedback module;

[0098] Process control module: used to automatically assign approval personnel and send notifications according to the preset work ticket review process, so as to realize the automatic circulation and approval process of work tickets;

[0099] Information analysis module: used to perform natural language processing on the text description in the work ticket, extract key information, and conduct intelligent analysis;

[0100] Risk point identification module: It is used to extract key information from the work ticket based on the information analysis module, and then use the naive Bayes algorithm to build a classification model to automatically identify risk points in the work task and determine whether the safety measures are compliant;

[0101] Image recognition and processing module: used to perform image recognition and processing on the pictures in the work ticket to realize automatic equipment status detection;

[0102] Cabinet location module: used to select the cabinet to be operated on the actual cabinet location map of the substation, and distinguish the selected location from the location without operation by the depth of color;

[0103] The panel cabinet selection module is used to automatically identify adjacent panel cabinets when the user selects a panel cabinet based on the relative position relationship between the panel cabinets, and generate and display detailed information of these panel cabinets;

[0104] Real-time monitoring and feedback module: used to connect the system with real-time monitoring equipment, obtain line status information in real time, and feed back relevant information to the person filling out the work ticket to ensure the accuracy and security of the work ticket.

[0105] In this embodiment, the process control module specifically needs to be explained. The process control module automatically assigns approval personnel and sends notifications according to the preset work ticket review process to realize the automatic circulation and approval process of the work ticket, which specifically includes the following steps:

[0106] Step 1: Preset the review process: The administrator presets the review process of the work ticket in the system, including each approval link and the corresponding reviewers, and defines the roles as the initial reviewer, the intermediate reviewer, and the final reviewer. The reviewers are designated by roles to reduce the operating costs when the process is changed;

[0107] Step 2, Work Ticket Submission: After completing the preset review process in step 1, when the user submits a new work ticket, the system receives the work ticket information and sets its status to pending approval. According to the preset review process, the next approval personnel are automatically assigned;

[0108] Step 3, Approver notification: After completing step 2, the system will send a notification to the staff member based on the assigned approver information, through email, SMS and instant messaging tools, to remind the approver to perform the approval operation;

[0109] Step 4, approval process: After completing the approval personnel notification in step 3, log in to the system for approval. The system should provide a corresponding interface to display the work ticket information and related content to be approved. By checking the work content and safety measures information, the approval, rejection, and refusal approval operations can be performed according to the actual situation.

[0110] Step 5, automatic transfer: After completing the approval process in step 4, the system automatically transfers the work ticket to the next link's approver based on the approval result. When the approval of a certain link is rejected or rejected, the system notifies the submitter and provides an opportunity to modify it;

[0111] Step 6, complete the approval: After completing the automatic flow in step 5, when the work ticket successfully passes all approval links, the system will set the work ticket status to completed, notify the staff, and generate approval records and related reports for tracking and counting the approval status.

[0112] In this embodiment, the information analysis module specifically needs to be explained. The information analysis module performs natural language processing on the text description in the work ticket, extracts key information, and performs intelligent analysis, which specifically includes the following steps:

[0113] Step 1: Text preprocessing: Preprocess the text description, segment the text description by words, divide the sentence into individual phrases, and remove special characters and punctuation marks to reduce interference with subsequent processing;

[0114] Step 2, entity recognition: After completing the text preprocessing in step 1, use named entity recognition technology to identify key entities in the text, and use the support vector machine algorithm to classify and annotate the corpus containing key entities such as work ticket number, work content, and work time. The specific calculation formula is as follows:

[0115] y h =sign(w T x+b)

[0116] Among them, y h represents the classification result, w represents the weight vector, T represents the parameter, x represents the feature vector, b represents the bias term, and sign represents the sign function, which maps the result of the linear combination to the prediction result of the category label;

[0117] Step 3, information extraction: After completing the entity recognition in step 2, extract key information from the preprocessed and entity recognized text according to the predefined template, use keyword matching technology to extract work content and working time information, and perform text classification and information retrieval by comparing the similarity between the two texts. The specific calculation formula is:

[0118]

[0119] Among them, Q represents the similarity, A and B represent the vector representation of two texts respectively, "·" represents the inner product operation, and "|" represents the modulus of the vector.

[0120] In this embodiment, it is specifically necessary to explain the risk point identification module. The risk point identification module, after extracting key information of the work ticket based on the information module, uses the naive Bayes algorithm to build a classification model, automatically identifies the risk points in the work task, and determines whether the safety measures are compliant. Specifically, the following steps are included:

[0121] Step 1: Feature extraction: Collect annotated work task description data, including work type, equipment name, work area, and keyword information, convert the text data into a feature vector representation, and use the bag-of-words model to count the frequency of each word in each sample to form a feature vector, including:

[0122] Let D = {d1, d2, ..., d n} represents the sample set, where d i represents the i-th sample, V = {w1,w2,...,w m} represents the set of all words, where wj Represents the jth word. For a sample d, count the number of occurrences of each word in V to obtain a vector x of length m i =(x i1 ,x i2 ,...,x im ), where x ij Represents word w j In sample d i The number of times it appears in

[0123] Stack the word vectors of all samples together to get an n×m feature matrix X, where each row represents the word vector of a sample and each column represents the frequency of word occurrence;

[0124] Step 2, classification model: After completing the feature extraction in step 1, divide the labeled data set into a training set and a test set. Use the training set to train the naive Bayes classifier, define the features to be independent of each other, and perform classification based on the conditional probability estimation of the features. The specific steps are as follows:

[0125] Step 201, for C categories, D represents the training set, X represents the feature matrix, y represents the category label corresponding to the training set, calculate the prior probability P (y = c), the specific calculation formula is as follows:

[0126]

[0127] Among them, I(y i =c) represents the indicator function, when y i =c is 1, otherwise it is 0, and n represents the total number of samples in the training set;

[0128] Step 202: For each feature x of the feature matrix X j , calculate the conditional probability distribution P(x j |y=c), the specific calculation formula is as follows:

[0129]

[0130] Among them, I(x ij =1andy i =c) represents the indicator function, when x ij =1andy i =c is 1, otherwise it is 0; α is the Laplace smoothing coefficient to avoid the probability of 0; |X| represents the total number of feature dimensions.

[0131] Step 203: For a new sample feature vector x, calculate the posterior probability P(y=c|x) that the sample belongs to each category. The specific calculation formula is as follows:

[0132]

[0133] Among them, m represents the total number of feature dimensions;

[0134] Step 204: Select the category with the maximum posterior probability as the prediction result. The specific calculation formula is as follows:

[0135]

[0136] in, Indicates the category labels corresponding to the training set, including risk points and non-risk points;

[0137] Step 3, model application: After completing the classification model in step 2, use the trained naive Bayes classification model to predict the risk points of the new work task description, convert the new description into feature vector form, and use the model to perform classification prediction to determine whether the work task has risk points.

[0138] In this embodiment, the image recognition and processing module specifically needs to be explained. The image recognition and processing module performs image recognition and processing on the image in the work ticket to realize automatic equipment status detection, which specifically includes the following steps:

[0139] Step 1: Image classification: Use the target detection model to perform entity recognition of key elements of equipment and lines in the pictures in the work ticket. Through training and reasoning, identify the objects of interest, and use the loss function Focal Loss to solve the problem of category imbalance in target detection. The specific calculation formula is as follows:

[0140] FL(p_t)=-(1-p_t) gamma ×log(p_t)

[0141] Among them, p_t represents the probability of the model predicting the target, and gamma represents an adjustable parameter;

[0142] When there are few target categories, p_t is small, (1-p_t) gamma Close to 1, the weight of the loss function is small;

[0143] When there are many target categories, p_t is large, (1-p_t) gamma Close to 0, the weight of the loss function is large;

[0144] Step 2: Device status detection: After completing the image classification in step 1, the input of the device status detection is an image X. By stacking multiple convolutional layers, pooling layers, and fully connected layers, a deep convolutional neural network is constructed to achieve the task of device status detection. The specific steps are as follows:

[0145] Step 201, the specific calculation formula of the convolution layer is as follows:

[0146] Z_L=convolve(A_L,W_L)+b_L

[0147] A_L+1=f(Z_L)

[0148] Among them, Z_L represents the output of the neuron, convolve() represents the convolution operation, A_L represents the input feature map, W_L represents the convolution kernel, b_L represents the bias term, L represents the number of layers, and f represents the activation function;

[0149] Step 202, the specific calculation formula of the pooling layer is as follows:

[0150] A_L+1=pool(A_L)

[0151] Among them, A_L represents the input feature map, pool() represents the pooling operation, and L represents the number of layers;

[0152] Step 203, fully connected layer: connect each neuron with all neurons in the previous layer, perform linear transformation through the weight matrix, and introduce nonlinear mapping with the activation function f. The specific calculation formula is as follows:

[0153] Z_L=W_L×A_L+b_L

[0154] A_L+1=f(Z_L)

[0155] Y=g(Z_L)

[0156] Among them, A_L represents the input vector, W_L represents the weight matrix, b_L represents the bias term, L represents the number of layers, and g represents the activation function.

[0157] In this embodiment, it is specifically necessary to explain the panel cabinet position module, which selects the panel cabinet to be operated on the drawn actual panel cabinet position map of the substation, and distinguishes the position to be operated from the position without operation by the depth of color, and specifically includes the following steps:

[0158] Step 1: Collect information: Obtain the floor plan, circuit diagram and panel layout diagram related to the substation from the design documents and on-site survey. Based on the collected information, draw the basic framework of the substation on the drawing software, including the outer boundary, main buildings, roads and surrounding environment. According to the panel layout and circuit diagram, use different symbols to mark the main equipment and lines, including transformers, circuit breakers and disconnectors for distinction;

[0159] Step 2: Draw the actual location map of the substation cabinets: After collecting information in step 1, use rectangles to represent the cabinets according to the cabinet layout diagram, draw the location of each cabinet on the drawing software, and mark each cabinet with a unique number, including the cabinet model, capacity, and detailed information on the purpose;

[0160] Step 3. Select the working cabinet: After completing step 2 and drawing the actual cabinet location map of the substation, use computer drawing software to mark the working cabinet with an ellipse on the actual cabinet location map, including:

[0161] For the cabinets that have been selected for operation, the color of the mark will be darkened to distinguish them from the cabinets that have not been operated;

[0162] For the screen cabinets that do not need to be worked on temporarily, add instructions on the labels, including the words "out of service" and "maintenance", to help staff clearly understand the working status of each screen cabinet.

[0163] In this embodiment, it is specifically necessary to explain the panel cabinet selection module. The panel cabinet selection module imports the actual panel cabinet location map of the substation into the computer-aided software, and enters detailed information for each panel cabinet, including the panel cabinet name, model, capacity, and purpose. At the same time, the relative position relationship between the panel cabinets is recorded. According to the relative position relationship between the panel cabinets, when the user selects a panel cabinet, the adjacent panel cabinets are automatically identified, and the detailed information of these panel cabinets is generated and displayed. The specific steps are as follows:

[0164] Step 1. Determine the position coordinates of the panel cabinet: Determine the position coordinates of each panel cabinet in the substation layout diagram, using a two-dimensional coordinate system with the panel cabinet in the upper left corner as the origin, the horizontal direction as the X axis, and the vertical direction as the Y axis;

[0165] Step 2, calculate the relative position: After completing step 1 to determine the position coordinates of the screen cabinet, for a screen cabinet A selected by the user, traverse all other screen cabinets B and calculate the relative position relationship between screen cabinets A and B. The Manhattan distance measurement method is used for calculation. The specific calculation formula is as follows:

[0166] distance=|x_B-x_A|+|y_B-y_A|

[0167] Among them, (x_A, y_A) represents the coordinates of cabinet A, (x_B, y_B) represents the coordinates of cabinet B, and distance represents the distance between the two cabinets;

[0168] Step 3, filter adjacent cabinets: After completing step 2 to calculate the relative position, filter out cabinets adjacent to cabinet A by setting a minimum distance threshold. When the distance of cabinet B is less than the threshold, it is determined that cabinet B is adjacent to cabinet A.

[0169] Step 4, generate detailed information: After completing step 3 to filter the adjacent cabinets, obtain their detailed information in the database according to their specific cabinet numbers, including model, capacity, and purpose, and display this information to the user.

[0170] In this embodiment, the real-time monitoring and feedback module specifically needs to be explained. The real-time monitoring and feedback module connects the system with the real-time monitoring device, obtains line status information in real time, and feeds back relevant information to the person filling out the work ticket to ensure the accuracy and security of the work ticket. The specific steps are as follows:

[0171] Step 1, obtain line status information: interface the sensor with the system, set a periodic time interval through polling, query once every 1 second, send a request to the monitoring device in each time interval, and obtain the response data, and obtain the current, voltage, temperature and humidity information of the line status from the monitoring device in real time;

[0172] Step 2: Line anomaly detection: After completing step 1 to obtain line status information, the current, voltage, temperature and humidity at a certain moment in the future are predicted through historical data analysis. The predicted value is used as a reference to dynamically adjust the threshold. Anomaly detection is performed according to the preset threshold to identify whether there is an abnormality in the line. The specific steps include the following:

[0173] Step 201: Use exponential weighted moving average to calculate the predicted value and adjust the threshold. Select a smoothing factor α to represent the weight ratio of the most recent data. Perform weighted moving average calculation on the historical data in chronological order. For each time point t, the predicted value Y_t is obtained by weighted average of the historical data and the current data. The specific calculation formula is as follows:

[0174] Y_t=α×X_t+(1-α)×Y_{t-1}

[0175] Where Y_t represents the exponentially weighted moving average at time point t, X_t represents the data at time point t, Y_{t-1} represents the exponentially weighted moving average at the previous time point, and α represents the smoothing factor, which ranges from (0<α<1);

[0176] Step 202: Set an adjustable parameter K to adjust the sensitivity of the threshold. Add K times the standard deviation to the average value as the threshold. The specific calculation formula is as follows:

[0177]

[0178]

[0179] Among them, S 上 represents the upper threshold, S下 represents the lower threshold, represents the mean, S 标 represents standard deviation;

[0180] At each new time point, the current data is compared with the threshold to determine whether there is an abnormality on the line;

[0181] Step 3, triggering the work ticket process: After completing the line abnormality detection in step 2, the work ticket process is triggered according to the abnormal line status detected, and the relevant line status information is passed to the person filling out the work ticket to remind him to initiate the corresponding work ticket;

[0182] Step 4, filling out the work ticket: After completing step 3 to trigger the work ticket process, the person filling out the work ticket fills out the work ticket based on the received line status information, accurately describing the work content, work scope and safety precautions.

[0183] The cloud computing-based transmission line work ticket intelligent invoicing process automation system of the present invention stores and manages a large amount of work ticket data through a cloud computing platform, and provides efficient data access and processing capabilities. At the same time, through the natural language processing, image recognition and other functions in artificial intelligence technology, intelligent analysis and processing of work ticket related information is realized.

[0184] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0185] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0186] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0187] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0188] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0189] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0190] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. The cloud computing-based intelligent invoicing process automation system for power transmission line work tickets is characterized by: Specifically, it includes process control module, information analysis module, risk point identification module, image recognition and processing module, screen cabinet location module, screen cabinet selection module, and real-time monitoring and feedback module; Process control module: used to automatically assign approval personnel and send notifications according to the preset work ticket review process, so as to realize the automatic circulation and approval process of work tickets; Information analysis module: used to perform natural language processing on the text description in the work ticket, extract key information, and conduct intelligent analysis; Risk point identification module: used to build a classification model based on the information analysis module, automatically identify risk points in work tasks, and determine whether safety measures are compliant; including: Step 1: Feature extraction: Collect annotated work task description data, including work type, equipment name, work area, and keyword information, convert the text data into a feature vector representation, and use the bag-of-words model to count the frequency of each word in each sample to form a feature vector, including: Let D = {d1, d2, ..., d n } represents the sample set, where d i represents the i-th sample, V = {w1,w2,...,w m } represents the set of all words, where w j Represents the jth word. For a sample d, count the number of occurrences of each word in V to obtain a vector x of length m i =(x i1 ,x i2 ,...,x im ), where x ij Represents word w j In sample d i The number of times it appears in Stack the word vectors of all samples together to get an n×m feature matrix X, where each row represents the word vector of a sample and each column represents the frequency of word occurrence; Step 2, classification model: After completing the feature extraction in step 1, divide the labeled data set into a training set and a test set. Use the training set to train the naive Bayes classifier, define the features to be independent of each other, and perform classification based on the conditional probability estimation of the features. The specific steps are as follows: Step 201, for C categories, D represents the training set, X represents the feature matrix, y represents the category label corresponding to the training set, calculate the prior probability P (y = c), the specific calculation formula is as follows: Among them, I(y i =c) represents the indicator function, when y i =c is 1, otherwise it is 0, n represents the number of sample sets; Step 202: For each feature x of the feature matrix X j , calculate the conditional probability distribution P(x j |y=c), the specific calculation formula is as follows: Among them, I(x ij =1andy i =c) represents the indicator function, when x ij =1andy i =c is 1, otherwise it is 0; α1 is the Laplace smoothing coefficient; |X| represents the total number of feature dimensions; Step 203: For a new sample feature vector x, calculate the posterior probability P(y=c|x) that the sample belongs to each category. The specific calculation formula is as follows: Among them, m represents the total number of feature dimensions; Step 204: Select the category with the maximum posterior probability as the prediction result. The specific calculation formula is as follows: in, Indicates the category labels corresponding to the training set, including risk points and non-risk points; Step 3: Model application: After completing the classification model in step 2, use the trained naive Bayes classification model to predict the risk points of the new work task description, convert the new description into a feature vector form, and use the model to perform classification prediction to determine whether the work task has risk points; Image recognition and processing module: used to perform image recognition and processing on the pictures in the work ticket to realize automatic equipment status detection; Panel cabinet location module: used to select the panel cabinet to be operated on the actual panel cabinet location map of the substation, and distinguish the selected location from the location without operation by the depth of color; The panel cabinet selection module is used to automatically identify adjacent panel cabinets when the user selects a panel cabinet based on the relative position relationship between the panel cabinets, and generate and display detailed information of these panel cabinets; Real-time monitoring and feedback module: used to connect the system with real-time monitoring equipment, obtain line status information in real time, and feed back relevant information to the person filling out the work ticket to ensure the accuracy and security of the work ticket; including: Step a1, obtaining line status information: interface the sensor with the system, set a periodic time interval through polling, query once every 1 second, send a request to the monitoring device in each time interval, and obtain response data, and obtain the current, voltage, temperature and humidity information of the line status from the monitoring device in real time; Step a2, line anomaly detection: After completing step a1 to obtain line status information, the current, voltage, temperature and humidity at a certain moment in the future are predicted through historical data analysis. The predicted value is used as a reference to dynamically adjust the threshold, and anomaly detection is performed according to the preset threshold to identify whether there is an abnormality in the line. Specifically, the following steps are included: Step a201, use exponential weighted moving average to calculate the predicted value and adjust the threshold, select a smoothing factor α, which represents the weight ratio of the most recent data, and perform weighted moving average calculation on the historical data in chronological order. For each time point t, the predicted value Y_t is obtained by weighted average of the historical data and the current data. The specific calculation formula is as follows: Y_t=α2×X_t+(1-α2)×Y_{t-1} Among them, Y_t represents the exponentially weighted moving average at time point t, X_t represents the data at time point t, Y_{t-1} represents the exponentially weighted moving average at the previous time point, and α2 represents the smoothing factor; Step a202: set an adjustable parameter K to adjust the sensitivity of the threshold. Add K times the standard deviation to the average value as the threshold. The specific calculation formula is as follows: Among them, S 上 represents the upper threshold, S 下 represents the lower threshold, represents the mean, S 标 represents standard deviation; At each new time point, the current data is compared with the threshold to determine whether there is an abnormality on the line; Step a3, triggering the work ticket process: after completing the line abnormality detection in step a2, the work ticket process is triggered according to the abnormal line status detected, and the relevant line status information is passed to the person filling out the work ticket, reminding him to initiate the corresponding work ticket; Step a4, filling in the work ticket: After completing step a3 to trigger the work ticket process, the person filling in the work ticket fills in the work ticket based on the received line status information, accurately describing the work content, work scope and safety precautions.

2. The cloud computing-based intelligent invoicing process automation system for power transmission line work tickets according to claim 1 is characterized by: The process control module presets the review process of the work ticket in the system. After receiving the information of the work ticket, the work ticket status is set to pending approval, and a notification is sent to the staff to perform the approval operation. According to the approval result, the work ticket is automatically transferred to the approval personnel of the next link. When the work ticket successfully passes all approval links, the work ticket status is set to completed, the staff is notified, and approval records and related reports are generated for tracking and counting the approval status.

3. The cloud computing-based intelligent invoicing process automation system for power transmission line work tickets according to claim 1 is characterized by: The information analysis module preprocesses the text description of the work ticket, uses named entity recognition technology to identify key entities in the text, and uses the support vector machine algorithm to classify and label the corpus containing key entities such as work ticket number, work content, and work time, and extracts key information from the preprocessed and entity-recognized text.

4. The cloud computing-based intelligent invoicing process automation system for power transmission line work tickets according to claim 1 is characterized in that: The image recognition and processing module uses the target detection model to perform entity recognition of key elements of equipment and lines in the pictures in the work ticket, identifies objects of interest through training and reasoning, and uses the loss function Focal Loss to solve the category imbalance problem in target detection. After completing image classification, it performs equipment status detection and builds a deep convolutional neural network to achieve the task of equipment status detection.

5. The cloud computing-based intelligent invoicing process automation system for power transmission line work tickets according to claim 4 is characterized in that: When performing device status detection, the input of the device status detection is an image X. By stacking multiple convolutional layers, pooling layers, and fully connected layers, a deep convolutional neural network is constructed to achieve the task of device status detection.

6. The cloud computing-based intelligent invoicing process automation system for power transmission line work tickets according to claim 1 is characterized in that: The panel cabinet location module obtains the floor plan, line diagram and panel cabinet layout diagram information related to the substation from the design documents and on-site surveys, draws the location of each panel cabinet on the drawing software based on the collected information, and marks a unique number for each panel cabinet. The panel cabinets in operation are marked with ellipses on the actual panel cabinet location map, and the color of the mark is deepened to distinguish the panel cabinets that have not been operated. For the panel cabinets that do not need to be operated temporarily, instructions are added to the marks to help staff clearly understand the working status of each panel cabinet.

7. The cloud computing-based intelligent invoicing process automation system for power transmission line work tickets according to claim 1 is characterized in that: The panel cabinet selection module imports the actual panel cabinet location map of the substation into the computer-aided software, and the person in charge of the work enters the specific name and the location of the small room into the work ticket. For a panel cabinet A selected by the user, all other panel cabinets B are traversed, and the relative position relationship between panel cabinet A and panel cabinet B is calculated, and then adjacent panel cabinets are screened to generate detailed information.

Images