Remote supervision and result analysis system based on ORM system

Abstract

The invention provides a remote supervision and result analysis system based on an ORM system. The remote supervision and result analysis system at least comprises a data acquisition module and a data arrangement and analysis module, the data acquisition module acquires information at least including illegal supervision and discovery conditions, video perception rate, video responsibility special notification and operation execution conditions of "four different from two", cleans data and stores the data in a database; and the data arrangement and analysis module realizes system integration of the security risk management and control supervision platform by using ORM, analyzes, studies and judges the data, mines the association between the data, generates accurate statistical results and reports according to predetermined rules and standards, and is used for identifying the type and frequency of violation behaviors, calculating a video perception rate, and sending the video perception rate to the security risk management and control supervision platform. And the operation execution condition is evaluated.

Description

Technical Field

[0001] The present invention relates to the fields of information technology, data processing and management, and in particular to a remote inspection and result analysis system based on an ORM system. Background Art

[0002] With the expansion of the scale of power grids and the increase in the number of operations, the pressure of remote safety management and control of operations is increasing. The traditional remote inspection work mode and big data application have drawbacks such as low efficiency and lack of intelligence.

[0003] In the work of power grid safety inspection information management, staff usually manually obtain relevant inspection result data from different files, reports or systems, and manually perform simple data entry and sorting. The analysis of data is mostly based on manual experience judgment and simple statistical methods. The disadvantages of this model are mainly reflected in low efficiency and insufficient intelligence. On the one hand, such repetitive operations can easily make staff mentally fatigued and easily introduce human errors. On the other hand, various types of data lack systematic integration, and the relationship between data lacks effective mining, so information cannot be fully utilized.

[0004] For example, in the existing technology, the compilation of information such as violations found during safety inspections, video perception rates, special reports on video fulfillment of responsibilities, and job execution status mainly relies on manual operations. Usually, staff manually collect various types of data, such as obtaining relevant inspection result data from different files, reports or systems. Then, simple data entry and compilation are performed using office software (such as Excel). For data analysis, most of them are based on manual experience judgment and simple statistical methods.

[0005] The main disadvantages of the existing technology include: 1. Inefficiency: The process of manually collecting and organizing data is very time-consuming, especially when the amount of data is large, which requires a lot of manpower and time costs. For example, when processing a large amount of "four no's and two direct's" inspection violation data, staff need to check records from different sources one by one, which is prone to omissions and errors.

[0006] For the special notification data on video perception rate and video fulfillment, manual analysis of video content to determine the accuracy of perception rate and fulfillment is also affected by human factors and is extremely inefficient.

[0007] 2. Lack of accuracy: Manual data entry is prone to errors, such as mis-entry and omission of data, etc. This may lead to deviations in the final statistical results and analysis, affecting the accurate judgment of the actual situation.

[0008] When evaluating job performance, manual analysis may fail to fully consider a variety of complex factors and indicators, leading to inaccurate conclusions.

[0009] 3. Lack of systematic integration. In existing technologies, different types of inspection result data (such as violations, video perception rate, and job execution) are often processed separately, and there is a lack of a unified system to integrate these data. This makes it difficult to explore the correlation between data and to fully and deeply understand the overall inspection situation. Summary of the invention

[0010] In view of the defects and shortcomings of the above-mentioned prior art, the present invention provides a remote inspection and result analysis system based on the ORM system, which automatically collects information from different sources and cleans the data with algorithms. The required tables are established in the database and the data is stored in the database. Then, different intelligent algorithms are used to analyze, judge and mine the data. The user interface is a user interaction page, and the query window is responsible for the user's request operation. The received data is submitted to the ORM, and the ORM analyzes the requested data, obtains the database where the data is located, and sends a data request to the database, and finally passes the request result to the user interface view. In addition, the system also provides a report download function so that users can download reports for offline viewing and sharing.

[0011] The functions to be achieved include: 1. Automated data collection It realizes the automatic collection of data such as violations discovered during the "Four No's and Two Direct" inspections, video perception rates, special reports on video fulfillment of responsibilities, and work execution status, without the need for manual collection, thus improving the efficiency and accuracy of data acquisition.

[0012] 2. Intelligent data organization and analysis Develop an intelligent algorithm that can automatically organize and analyze different types of collected data and generate accurate statistical results and reports according to predetermined rules and standards. For example, it can automatically identify the type and frequency of violations, accurately calculate the video perception rate, and conduct a comprehensive assessment of job execution.

[0013] 3. Unified data management system Establish a unified data management platform based on the safety risk management and supervision platform system to integrate inspection result data from different sources and types to facilitate data storage, query and sharing. At the same time, through the characteristics of the safety risk management and supervision platform system, the mapping of data and database table structure is realized to improve the efficiency and flexibility of data management.

[0014] It specifically includes the following technical contents: A remote inspection and result analysis system based on an ORM system comprises at least: a data acquisition module and a data collation and analysis module; the data acquisition module collects information including at least violations found during the "four no's and two direct's" inspection, video perception rate, special video responsibility reports, and job execution status, cleans the data, and stores it in a database; the data collation and analysis module uses ORM to realize system integration of a safety risk management and control supervision platform, analyzes, determines, and mines the associations between data, generates accurate statistical results and reports according to predetermined rules and standards, and is used to identify the types and frequencies of violations, calculate video perception rates, and evaluate job execution status.

[0015] Furthermore, the interface of the data acquisition module includes: The “Four No’s and Two Direct’s” inspection system interface establishes a connection with the database of the “Four No’s and Two Direct’s” inspection system, uses SQL query statements to obtain violation data, and uses JSON format to transmit data; Video-related system interface, obtains video number, shooting time, and location information through specific API calls; obtains binding information by querying the video and related business association table; extracts key data by analyzing the special notification document on video fulfillment and using the corresponding document parsing library; The job execution monitoring system interface obtains the start time and end time data of the job through the log file or database of the job execution monitoring system; by querying the job step status table, it determines whether each job step is completed and the completion timestamp.

[0016] Furthermore, the data cleaning process of the data acquisition module includes: Deduplication: Use a unique index in a hash table or database to detect duplicate data. First, a unique hash value is calculated for each piece of data. Then, the hash value is inserted into the hash table or database index. If the insertion fails, it is deleted as duplicate data. Correction of incorrect data format: Use regular expressions to match date strings and then convert them to standard date formats; for numeric data, if non-numeric characters appear, try to remove the non-numeric part and convert it to numeric type; if the conversion fails, set the default value or mark it as incorrect data according to the context of the data; Missing data filling: For data with time series characteristics, calculate the average value of historical data to fill missing values. First, filter out the historical data set based on the time range or other relevant characteristics of the data, and then calculate the average value of the target field in the set; for non-numeric data or data that is difficult to fill with historical data, set a default value based on the data type and business rules.

[0017] Furthermore, the data collation and analysis module realizes the system integration of the security risk management and control supervision platform, including: data mapping and ORM function realization; The data mapping includes: “Four No’s and Two Direct’s” Inspection Violation Data Mapping: The violation data collected and cleaned from the “Four No’s and Two Direct’s” inspection system will be inserted into the violation table according to the field correspondence; Video perception rate data mapping: inserting the calculated video perception rate data into a video perception rate table, wherein the video perception rate table includes at least a perception rate value and a timestamp of perception rate calculation; Job execution data mapping: associate the job execution data with the job ID and insert it into the corresponding field of the job execution table, using a specific code to indicate the completion status of the job steps; The ORM function implementation includes: Through the ORM framework, the data objects are mapped to the database table structure, so that when storing data, only the data objects need to be operated. The ORM framework automatically converts the state of the object into the corresponding SQL statement and executes the database operation. Use the query interface provided by the ORM framework to perform data query operations.

[0018] Furthermore, the data sorting and analysis module uses a support vector machine (SVM) algorithm to classify traffic violations, including the following steps: Data preprocessing: The cleaned violation data is divided into a training set and a test set, and feature engineering is performed on the data; Model training: Use the training set data to train the SVM model; Model evaluation and optimization: Use the test set data to evaluate the performance of the trained SVM model; if the performance does not meet the requirements, further adjust the hyperparameters or increase the amount of data, optimize feature engineering, and retrain the model; Classification application: Apply the trained optimal SVM model to the newly collected violation data, predict the category of violation behavior, and count the frequency of occurrence of different types of violation behavior.

[0019] Furthermore, the data sorting and analysis module uses a multivariate linear regression algorithm to analyze factors affecting the video perception rate, including the following steps: Data preparation: Collect data related to video perception rate, including light intensity data, object size data, and corresponding video perception rate data, and organize them into a data set, where each row of data represents a video sample, including light intensity, object size, and perception rate; Model construction and training: The data set is divided into a training set and a test set; a multivariate linear regression model is constructed using the training set data to determine the linear relationship coefficients between the independent variables including light intensity and target size and the dependent variables including perception rate; Model evaluation and testing: Evaluate the trained model, calculate the model's R-squared value and mean square error index to evaluate the model's fit to the data and prediction accuracy; perform hypothesis testing, check the significance of independent variables, exclude insignificant independent variables, and retrain the model to improve model performance; Factor analysis and prediction: Based on the trained model, analyze the impact of each independent variable on the video perception rate through the size and positive and negative analysis of the regression coefficient; use the model to predict the perception rate of new video data, and adjust the parameters of the video surveillance equipment or optimize the video acquisition environment based on the prediction results to improve the video perception rate.

[0020] Furthermore, the data collation and analysis module uses a process mining algorithm based on Petri nets to analyze the compliance and efficiency of the operation process to describe the activities, sequence relationships, resource allocation, and concurrent execution in the operation process. By mining the operation execution log data, the differences between the actual operation process and the preset process model are discovered, and the compliance and efficiency of the process are evaluated, including the following steps: Data preprocessing: Extract the job step name, start time, end time, and execution resource information from the job log data obtained from the job execution monitoring system; then convert it into a format suitable for process mining algorithm processing; Model discovery: Use process mining algorithms to analyze event log data and build a job process model represented by a Petri net; automatically discover the structure of the job process based on the rules of mining the sequential and concurrent relationships between job steps. Compliance check: Compare the mining process model with the preset standard process model to check whether the actual process meets the standard requirements; by calculating the difference index between the two, determine the illegal operations or deviations from the standard process in the process; Efficiency analysis: Analyze the execution efficiency of the job process based on the job process model and job log data; calculate the average execution time of the job, the average waiting time of each job step, and the indicators of resource utilization.

[0021] Furthermore, it also includes: a user interaction interface module, a query window operates according to the user's request, submits the received data to the ORM, the ORM analyzes the requested data, obtains the database where the data is located and sends a data request to the database, and passes the request result to the user interface view.

[0022] And, a remote inspection and result analysis method based on the ORM system: by collecting information including at least "four no's and two direct" inspections of violations, video perception rates, special reports on video responsibility, and job execution, and cleaning the data and storing it in a database; using ORM to implement system integration of the safety risk management and supervision platform, analyze, judge, and mine the relationships between data, and generate accurate statistical results and reports according to predetermined rules and standards to identify the types and frequencies of violations, calculate video perception rates, and evaluate job execution.

[0023] Furthermore, according to the user's request operation, the received data is submitted to the ORM, which analyzes the requested data, obtains the database where the data is located, sends a data request to the database, and passes the request result to the user interface view.

[0024] Compared with the prior art, the present invention and its preferred solutions have at least the following outstanding advantages: 1. High degree of automation 1. It can automatically collect, organize and analyze inspection result data, greatly reducing manual intervention and improving work efficiency. In contrast, existing technologies mainly rely on manual operation, which is inefficient and prone to errors.

[0025] 2. For the video perception rate and video responsibility special report data, the perception rate and responsibility status can be obtained more accurately and quickly by automatically analyzing the video content. The existing technology of manual video analysis has poor accuracy and efficiency.

[0026] 2. Improved data accuracy 1. Through the data cleaning function of the data acquisition module and the application of intelligent algorithms, data entry errors can be effectively avoided and data accuracy can be improved. The existing technology of manual data entry is prone to problems such as mis-entry and missed input.

[0027] 2. When evaluating the execution of a job, the use of algorithms such as process mining can more comprehensively and accurately analyze the compliance and efficiency of the job process, while the existing manual analysis technology has great limitations.

[0028] 3. Systematic integration advantages A unified data management platform based on the safety risk management and supervision platform system has been established to integrate different types of inspection result data to facilitate correlation analysis and sharing between data. In existing technologies, data is often processed separately and lacks systematic integration. Through a unified data management platform and intelligent algorithms, a more comprehensive and in-depth inspection result analysis report can be provided, which helps to better understand the overall inspection situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments: Figure 1 A schematic diagram of a query statement for obtaining traffic violation data using an SQL query statement according to an embodiment of the present invention; Figure 2 A schematic diagram of a sentence of traffic violation data obtained in an embodiment of the present invention; Figure 3 This is a schematic diagram of the getVideoInfo function according to an embodiment of the present invention; Figure 4 The embodiment of the present invention connects the video service association table and the service information representation through SQL; Figure 5 A schematic diagram of using the pandas library to read table data in an embodiment of the present invention; Figure 6 This is a schematic diagram of SQL query according to an embodiment of the present invention; Figure 7 Querying the status of a job step for an embodiment of the present invention; Figure 8 This is a schematic diagram of a unified implementation of the date format according to an embodiment of the present invention; Fig. 9 This is a schematic diagram of using SQL query to calculate the average value of job execution time in the past month in an embodiment of the present invention; Fig.10 A schematic diagram of an SQL insert statement used in an embodiment of the present invention; Fig.11 A schematic diagram of a mapping class for defining "four no's and two direct's" inspection violation data using SQLAlchemy in Python according to an embodiment of the present invention; Fig.12 A schematic diagram of querying traffic violation data within a certain time period according to an embodiment of the present invention; Fig.13 This is a schematic diagram of using the sklearn library in Python to train an SVM model in an embodiment of the present invention; Fig.14 A schematic diagram of constructing a multiple linear regression model using the statsmodels library in Python in an embodiment of the present invention; Fig.15 This is a schematic diagram of evaluating a trained model according to an embodiment of the present invention; Fig.16 This is a schematic diagram of using the pm4py library in Python to perform process mining in an embodiment of the present invention; Fig.17 A schematic diagram of determining illegal operations or deviations from standard procedures in an operation process according to an embodiment of the present invention; Fig.18 This is a schematic diagram of calculating the average execution time by analyzing the start time and end time of a job step according to an embodiment of the present invention; Fig.19 This is a schematic diagram of a remote inspection and result analysis system based on an ORM system according to an embodiment of the present invention; Fig. 20 This is a sample diagram of interface design used in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to make the features and advantages of the present invention more clearly understood, the following embodiments are specifically described in detail with reference to the accompanying drawings: It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention. In addition, the technical features involved in each embodiment of the present invention described below can be combined with each other as long as there is no conflict between them.

[0031] like Fig.19 As shown, the remote inspection and result analysis system based on the ORM system provided by the embodiment of the present invention automatically collects information from different sources and cleans the data with algorithms. The required tables are established in the database and the data is stored in the database. Then, different intelligent algorithms are used to analyze, judge and mine the data. The user interface is a user interaction page, and the query window is responsible for the user's request operation. The received data is submitted to the ORM, which analyzes the requested data, obtains the database where the data is located, sends a data request to the database, and finally passes the request result to the user interface view. The specific implementation of each module of the system is further introduced below.

[0032] 1. Data acquisition module 1. Interface design (1) “Four No’s and Two Direct” Inspection System Interface Data acquisition method: This embodiment establishes a connection with the database of the "Four No's and Two Direct" inspection system and uses SQL query statements to obtain violation data. The query statement is as follows: Figure 1 shown.

[0033] Data transmission format: This embodiment uses JSON format to transmit data, so as to facilitate data exchange between different systems. For example, a piece of violation data obtained is represented as Figure 2 The statement.

[0034] (2) Video-related system interface Obtaining basic video information: This embodiment targets the database of the video management system and obtains information such as video number, shooting time, location, etc. through specific API calls. For example, as a preferred solution, the getVideoInfo function provided by the video system is used, and the video ID is passed in as a parameter. The function returns a JSON object containing basic video information, such as Figure 3 shown.

[0035] Video and service binding information acquisition: This embodiment obtains binding information by querying the video and related service association table. For example, the video service association table and the service information table are connected through SQL, such as Figure 4 shown.

[0036] Video accountability special notification data extraction: Analyze the video accountability special notification document (such as PDF or Excel format), and use the corresponding document parsing library (such as Python's pdfminer or pandas library) to extract key data. For example, for a notification file in Excel format, use the pandas library to read the table data, such as Figure 5 shown.

[0037] (3) Job execution monitoring system interface Job time data acquisition: Get the job start and end time data from the job execution monitoring system's log file or database. If it is a log file, you can use text parsing technology to extract time information; if it is a database, use something like Figure 6 The SQL query shown.

[0038] Obtaining the completion status of job steps: query the job step status table to determine whether each job step is completed and the completion timestamp. The implementation example is as follows: Figure 7 shown.

[0039] 2. Data cleaning (1) Deduplication Algorithm flow: Use a unique index of a hash table or database to detect duplicate data. First, calculate a unique hash value for each piece of data (for example, based on a combination of key fields in the data). Then, insert the hash value into the hash table or database index. If the insertion fails (indicating that the hash value already exists), the data is duplicated and is deleted.

[0040] Processing steps: 1. Traverse the collected data set.

[0041] 2. For each piece of data, calculate the hash value based on its key fields (such as the violation time, location, and behavior category in the “Four No’s and Two Direct” inspection data; the video number, shooting time, and location in the video data; the job ID and start time in the job data, etc.).

[0042] 3. Try to insert the hash value into a unique index of a hash table or database.

[0043] 4. If the insertion is successful, continue to process the next data; if the insertion fails, delete the current data record.

[0044] (2) Correction of incorrect data format Date format unification: Use regular expressions to match date strings and then convert them to a standard date format (such as ISO 8601: YYYY-MM-DDTHH:mm:ss.sssZ). For example, in Python, this can be done as Figure 8 shown.

[0045] Other data type error handling: For numeric data, if non-numeric characters appear, you can try to remove the non-numeric part and convert it to a numeric type; if the conversion fails, a reasonable default value is set or marked as error data according to the context of the data. For example, for the string data "100pcs" representing the quantity, remove the character "pcs" and convert it to the integer 100.

[0046] (3) Missing data filling Historical data mean filling: For data with time series characteristics (such as job execution time), calculate the average value of historical data to fill missing values. First, filter out the historical data set based on the time range or other relevant characteristics of the data, and then calculate the average value of the target field in the set. For example, use SQL query to calculate the average value of job execution time in the past month, such as Fig. 9 shown.

[0047] Default value filling: For some non-numeric data or data that is difficult to fill with historical data (such as some description information of the video), set the default value according to the data type and business rules. For example, for the shooting device information of the video, if it is missing, it can be set to "unknown device".

[0048] 2. Data collation and analysis module 1. Security risk management and supervision platform system integration (1) Data Mapping "Four No's and Two Direct" Inspection Violation Data Mapping: Create a "Violation Status Table" containing fields such as violation_id (auto-increment primary key), violation_time, violation_location, violation_type, etc. The violation data collected and cleaned from the "Four No's and Two Direct" inspection system is inserted into the "Violation Status Table" according to the field correspondence. For example, use an SQL insert statement such as Fig.10 shown.

[0049] Video perception rate data mapping: Create a "video perception rate table" containing fields such as video_id, perception_rate, and measurement_time. Insert the calculated video perception rate data into the table, where the perception_rate field stores the perception rate value and the measurement_time field records the timestamp of the perception rate calculation.

[0050] Job execution status data mapping: Design a "job execution status table" with fields such as job_id, start_time, end_time, step_completion_status, etc. Associate the job execution data according to the job ID and insert it into the corresponding field. The step_completion_status field can use a specific code to indicate the completion status of the job step (such as 0 for unfinished and 1 for completed).

[0051] (2) Object-Relational Mapping (ORM) Function Utilization Data storage optimization: Use the ORM framework (such as Hibernate in Java or SQLAlchemy in Python) to map data objects to database table structures. In this way, when storing data, you only need to operate the data object, and the ORM framework will automatically convert the object state into the corresponding SQL statement and execute the database operation, improving the efficiency of data storage and the maintainability of the code. For example, in Python, use SQLAlchemy to define the mapping class of the "four no's and two straight's" inspection violation data, such as Fig.11 shown.

[0052] Convenience of data query: The query interface provided by the ORM framework can be used to conveniently perform data query operations. Query conditions can be constructed through object attributes and methods without writing complex SQL query statements. For example, querying the violation data within a certain period of time, such as Fig.12 shown.

[0053] 2. Application of intelligent algorithms (1) Traffic violation data classification algorithm Algorithm selection: Support vector machine (SVM) algorithm is used to classify violations. SVM algorithm classifies different types of data by constructing a hyperplane, and has good performance in dealing with small samples and nonlinear classification problems.

[0054] Algorithm flow: 1. Data preprocessing: The cleaned violation data is divided into a training set and a test set, usually in a ratio of 70:30. Feature engineering is performed on the data, for example, the violation location is encoded and converted into numerical features, and the violation time is extracted into features such as working days, holidays, and time periods, which together with the violation behavior category form a feature vector.

[0055] 2. Model training: Use the training set data to train the SVM model. Select a suitable kernel function (such as radial basis function RBF), and adjust the model's hyperparameters (such as penalty parameter C and kernel function parameter gamma) through cross-validation to obtain the best classification performance. For example, use the sklearn library in Python to train the SVM model, such as Fig.13 shown.

[0056] 3. Model evaluation and optimization: Use the test set data to evaluate the performance of the trained SVM model and calculate indicators such as accuracy, recall, and F1 value. If the performance does not meet the requirements, further adjust the hyperparameters or increase the amount of data, optimize feature engineering, and retrain the model.

[0057] 4. Classification application: Apply the trained optimal SVM model to the newly collected violation data, predict the category of violation behavior, and count the frequency of occurrence of different types of violation behavior.

[0058] (2) Video Perception Rate Data Regression Analysis Algorithm Algorithm selection: Multiple linear regression algorithm is used to analyze the factors affecting video perception rate. Multiple linear regression can establish a linear relationship model between multiple independent variables (such as lighting conditions, target size, etc.) and the dependent variable (video perception rate).

[0059] Algorithm flow: 1. Data preparation: Collect data related to video perception rate, including light intensity data (which can be obtained through sensors or extracted from video metadata), target size data (the pixel area or actual size estimation of the target in the video screen is calculated by the target detection algorithm), and the corresponding video perception rate data. Organize these data into a dataset, where each row of data represents a video sample, including columns such as light intensity, target size, and perception rate.

[0060] 2. Model construction and training: Divide the data set into a training set and a test set. Use the training set data to build a multivariate linear regression model and determine the linear relationship coefficient between the independent variables (light intensity, target size, etc.) and the dependent variable (perception rate). For example, use the statsmodels library in Python to build a multivariate linear regression model, such as Fig.14 shown.

[0061] 3. Model evaluation and testing: Evaluate the trained model and calculate the model's R-squared value, mean square error (MSE) and other indicators to evaluate the model's fit to the data and prediction accuracy. At the same time, perform hypothesis testing, check the significance of independent variables, exclude insignificant independent variables, and retrain the model to improve the model's performance, such as Fig.15 shown.

[0062] 4. Factor analysis and prediction: Based on the trained model, analyze the influence of each independent variable (lighting conditions, target size, etc.) on the video perception rate (through the size and positive and negative of the regression coefficient). Use the model to predict the perception rate of new video data, and adjust the parameters of the video surveillance equipment or optimize the video acquisition environment based on the prediction results to improve the video perception rate.

[0063] (3) Job execution data process mining algorithm Algorithm selection: Use the process mining algorithm based on Petri net to analyze the compliance and efficiency of the operation process. Petri net can intuitively describe the activities, sequence relationships, resource allocation, and concurrent execution in the operation process. By mining the operation execution log data, the difference between the actual operation process and the preset process model can be found, and the compliance and efficiency of the process can be evaluated.

[0064] Algorithm flow: 1. Data preprocessing: Extract information such as job step name, start time, end time, and execution resources from the job log data obtained from the job execution monitoring system. Convert this information into a format suitable for process mining algorithm processing. For example, construct an event log table, where each row records an execution event of a job step, including fields such as event ID, job ID, step name, timestamp, and resource ID.

[0065] 2. Model discovery: Use process mining algorithms (such as the Alpha algorithm or the Heuristic Miner algorithm) to analyze event log data and build a job process model (represented in the form of a Petri net). These algorithms automatically discover the structure of the job process based on mining rules such as the sequence relationship and concurrency relationship between job steps. For example, in Python, use the pm4py library for process mining, such as Fig.163. Compliance check: Compare the mining process model with the preset standard process model to check whether the actual process meets the standard requirements. By calculating the difference index between the two (such as edit distance or behavior profile difference), determine the illegal operation or deviation from the standard process in the process, such as Fig.17 shown.

[0066] 4. Efficiency analysis: Based on the job process model and job log data, analyze the execution efficiency of the job process. Calculate indicators such as the average execution time of the job, the average waiting time of each job step, and resource utilization. For example, calculate the average execution time by analyzing the start time and end time of the job step, such as Fig.18 shown.

[0067] 5. Process optimization suggestions: Based on the results of compliance inspection and efficiency analysis, we will propose optimization suggestions for the operation process. For example, if the waiting time for a certain operation step is too long, we can consider adjusting resource allocation or optimizing the order of operation steps; if there are frequent violations, we can strengthen the training of operators or improve the monitoring mechanism of the operation process.

[0068] (III) User interface module 1. Data query interface (1) Interface layout design Search bar: Located in a prominent position at the top of the interface, it is used by users to enter keywords for data query. There is a search button next to the search bar. Users can start the query operation by clicking the button after entering keywords. The search bar should have an automatic prompt function. When users enter some keywords, the system automatically displays a list of possible matching keywords based on the existing data information, which is convenient for users to quickly select.

[0069] Filter area: Set up a filter area below the search bar to further filter the query results. The filter conditions include department name, time range, inspection type and other dimensions. Department name filtering can display a list of all searchable departments through a drop-down menu, and users can click to select; time range filtering provides a date selector, and users can select the start date and end date to determine the data within a specific time period; inspection type filtering also presents different inspection type options in the form of a drop-down menu, such as work progress inspection, quality inspection, compliance inspection, etc.

[0070] Data display area: occupies the main part of the interface, used to display the query and filtered result data. The data is presented in a table format, with each row representing an inspection result record, and the columns displaying detailed information of the record, such as inspection number, department, inspection time, inspection result description, and relevant responsible persons. The table should have an adaptive width function to ensure that the data content can be clearly displayed under different screen resolutions to avoid horizontal scroll bars that affect the user's viewing experience.

[0071] Sorting button: Located above or below the data display area, it provides multiple sorting options, such as ascending / descending order by inspection time, alphabetical order by department name, etc. After the user clicks the corresponding sorting button, the data in the data display area will be rearranged according to the selected sorting rule, making it easier for users to quickly locate and compare data. An example of an interface design is shown below: Fig. 20 shown.

[0072] (II) Query processing flow Keyword input and verification: After the user enters a keyword in the search bar, the system first verifies the keyword. Check whether the keyword conforms to the grammatical rules and data format requirements. If the keyword is wrong or incomplete, the system pops up a prompt box to inform the user of the error information and suggest the user to correct it. For example, if the time format entered by the user is incorrect, the system prompts the user to re-enter it in the correct date format (such as yyyy-MM-dd).

[0073] Query request sending: When the keyword verification is passed, the system constructs a query request based on the keyword entered by the user and sends it to the backend database. The query request contains keyword information and the currently set filtering conditions (if any), so that the backend can accurately retrieve the data that meets the requirements from the database.

[0074] Data retrieval and return: After receiving the query request, the backend database retrieves data in the database according to the conditions in the request. The database query engine uses efficient indexing and query algorithms to quickly locate data records that match keywords and filter conditions. After the retrieval is completed, the query result data is returned to the front-end interface in a specific data format (such as JSON format).

[0075] Data display and update: After the front-end interface receives the query result data returned by the backend, it parses the data and fills it into the table in the data display area. During the data filling process, the corresponding field values ​​of each record are correctly displayed in the corresponding column according to the setting of the table column. At the same time, if the interface sets the default sorting rule, the data will be sorted according to the sorting rule before displaying. If the user modifies the filter conditions or sorting rules during the data display process, the system repeats the above query processing flow to re-acquire and display the data that meets the new conditions.

[0076] (III) Data screening and processing steps Filter condition setting: The user selects or enters filter conditions in the filter area, such as selecting a specific department name, setting a time range, or selecting an inspection type. Each filter condition setting operation should be fed back to the system in real time. For example, when the user selects a department in the department name drop-down menu, the system immediately updates the interface display to indicate that the current query will only be performed on the data of the selected department.

[0077] Filter request construction and sending: When the user has completed the setting of all filter conditions, click the "Apply Filter" button (or similar function button) in the filter area. The system constructs a filter request based on the filter conditions set by the user, combines it with the current query keyword (if any), and sends it to the backend database. The filter request clearly specifies the specific value and logical relationship of each filter condition (such as AND or OR connecting multiple conditions) to ensure that the backend database can accurately understand the user's filter intention.

[0078] Backend filtering and data return: After receiving the filtering request, the backend database further filters the data according to the conditions in the filtering request based on the previous query results (if there are already query results) or directly in the full data (if it is the first filtering). The filtering process follows the query optimization principle of the database and uses indexes and appropriate query statements to efficiently filter out data records that meet the conditions. After the filtering is completed, the filtered result data is returned to the front-end interface again.

[0079] Interface data update: The front-end interface receives the filtered data returned by the backend, updates the table content in the data display area, and displays the latest filtered data results. At the same time, the interface should retain the current filter condition settings so that users can clearly know the conditions under which the current data is filtered, and it is convenient for users to modify the filter conditions and re-filter when necessary.

[0080] (IV) Data sorting and processing flow Sorting option selection: The user clicks the sort button above or below the data display area and selects the desired sorting rule, such as ascending order by inspection time, descending order by department name, etc. When the user clicks the sort button, the system records the selected sorting rule information.

[0081] Sorting request construction and sending: The system constructs a sorting request based on the sorting rules selected by the user and sends it to the backend database together with the current query keywords and filter conditions (if any). The sorting request clearly specifies the sorting fields and sorting direction (ascending or descending) so that the backend database can sort the data as required.

[0082] Backend sorting and data return: After receiving the sorting request, the backend database sorts the current query result data (if there is already a query result) or the full amount of data (if it is the first sorting) according to the specified sorting field and direction. The database uses the built-in sorting algorithm to efficiently sort the data to ensure that the data can be arranged quickly and orderly according to user requirements. After the sorting is completed, the sorted result data is returned to the front-end interface.

[0083] Interface data update: The front-end interface receives the sorted data returned by the backend, updates the table content in the data display area, and re-displays the data in the sorted order. Users can intuitively see that the data is sorted according to the selected rules, which is convenient for comparing and analyzing the data. If the user selects a different sorting rule again, the system repeats the above sorting process, re-sorts the data and updates the interface display.

[0084] 2. Report Viewing Interface (1) Interface layout design Report list area: located on the left side of the interface, it displays all available reports in a list format. Each report item displays key information such as the report name, generation time, and report type. The report list should have a scroll bar function so that users can easily browse and find specific reports when there are a large number of reports.

[0085] Report preview area: occupies most of the space on the right side of the interface, used to preview the content of the selected report. The preview area supports displaying report information in a combination of intuitive charts (such as bar charts, line charts, pie charts, etc.) and text. For example, for a work progress inspection report, a bar chart may be used to show the proportion of work completion progress of different departments, and at the same time, the specific work progress, existing problems and recommended measures of each department are described in detail in text below the chart. The chart should have interactive functions, such as displaying a detailed data information prompt box when the mouse hovers over a chart data point, so that users can view specific values.

[0086] Download button: Located above or below the report preview area, after the user selects a report in the report list and views it in the preview area, if the user needs to download the report for offline viewing or sharing, click the download button. The file format of the report (such as PDF, DOCX, etc.) can be displayed next to the download button so that the user can understand the file type after downloading.

[0087] Back button: Set a back button in the upper left or upper right corner of the report preview area. When the user wants to return to the report list to view other reports during report preview, click the back button to return to the report list interface.

[0088] (2) Report review process Report list loading: When the interface is initialized, the system obtains all generated report information from the background and loads it into the report list area for display. The system sends a request to the background to obtain the report list. The background queries the report records stored in the database, including the report name, generation time, report type and other information, and organizes this information into a suitable data format (such as JSON array) and returns it to the front-end interface. After the front-end interface receives the report list data, it traverses the data and displays each report item in the report list area in the form of a list. At the same time, a click event processing function is added to each report item so that the report preview operation can be triggered when the user clicks the report item.

[0089] Report preview request and loading: After the user clicks a report item in the report list area, the system triggers a report preview request. The request carries the unique identifier of the selected report (such as the report number) and is sent to the backend database. The backend reads the corresponding report content data from the database according to the report number, including chart data, text descriptions and other information in the report. These data are sorted into a format suitable for front-end display (such as HTML fragments or specific JSON structures) and returned to the front-end interface. After receiving the report preview data, the front-end interface parses and displays the report content in the report preview area. For chart data, use a suitable chart library (such as Echarts, Highcharts, etc.) to render the data into the corresponding chart and display it in the preview area; for the text description part, directly fill the text content into the corresponding position for display.

[0090] Report download processing: When the user clicks the download button in the report preview area, the system first determines the file format of the current preview report (such as based on the report information returned by the background or the pre-set default format). Then, the system sends a report download request to the background, which carries the report number and the required file format information. The background obtains the original data of the report from the database according to the report number, and converts and generates the download file according to the specified file format. For example, if a PDF file is generated, the background uses the PDF generation library to convert the report content into a file stream in PDF format, and returns the file stream to the front-end interface. After the front-end interface receives the file stream returned by the background, it triggers the download function of the browser, saves the file to the local path specified by the user, and completes the report download operation.

[0091] Return to report list operation: When the user clicks the back button in the report preview area, the system clears the current report content display in the report preview area, reloads the report list data in the report list area, and displays all report information so that the user can select other reports to view.

[0092] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt 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 codes.

[0093] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 processor, 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.

[0094] 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.

[0095] 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.

[0096] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0097] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, various changes, modifications, substitutions and variations of these embodiments are made without departing from the principles and spirit of the present invention, and still fall within the scope of protection of the present invention.

[0098] The present invention is not limited to the above-mentioned optimal implementation mode. Anyone can derive other various forms of remote inspection and result analysis systems based on the ORM system under the inspiration of the present invention. All equal changes and modifications made according to the scope of the patent application of the present invention should fall within the scope of the present invention.

Claims

1. A remote inspection and result analysis system based on ORM system, characterized by: At least: Data collection module and data collation and analysis module; The data collection module collects information including at least violations discovered during the "four no's and two direct's" inspection, video perception rate, special video responsibility reports, and job execution status, cleans the data, and stores it in a database; the data collation and analysis module uses ORM to achieve system integration of the safety risk management and supervision platform, analyzes, determines, and mines the relationships between data, and generates accurate statistical results and reports according to predetermined rules and standards to identify the types and frequencies of violations, calculate video perception rates, and evaluate job execution.

2. The remote inspection and result analysis system based on the ORM system according to claim 1 is characterized in that: The interface of the data acquisition module includes: The "Four No's and Two Direct" inspection system interface establishes a connection with the database of the "Four No's and Two Direct" inspection system, uses SQL query statements to obtain violation data, and uses JSON format to transmit data; Video-related system interface, obtains video number, shooting time, and location information through specific API calls; obtains binding information by querying the video and related business association table; extracts key data by analyzing the special notification document on video fulfillment and using the corresponding document parsing library; The job execution monitoring system interface obtains the start time and end time data of the job through the log file or database of the job execution monitoring system; by querying the job step status table, it determines whether each job step is completed and the completion timestamp.

3. The remote inspection and result analysis system based on the ORM system according to claim 1 is characterized in that: The data cleaning process of the data acquisition module includes: Deduplication: Use a unique index in a hash table or database to detect duplicate data. First, a unique hash value is calculated for each piece of data. Then, the hash value is inserted into the hash table or database index. If the insertion fails, it is deleted as duplicate data. Correction of incorrect data format: Use regular expressions to match date strings and then convert them to standard date formats; for numeric data, if non-numeric characters appear, try to remove the non-numeric part and convert it to numeric type; if the conversion fails, set the default value or mark it as incorrect data according to the context of the data; Missing data filling: For data with time series characteristics, calculate the average value of historical data to fill missing values. First, filter out the historical data set based on the time range or other relevant characteristics of the data, and then calculate the average value of the target field in the set; for non-numeric data or data that is difficult to fill with historical data, set a default value based on the data type and business rules.

4. The remote inspection and result analysis system based on the ORM system according to claim 1 is characterized by: The data collation and analysis module realizes the system integration of the security risk management and control supervision platform, including: data mapping and ORM function realization; The data mapping includes: "Four No's and Two Direct" inspection violation data mapping: the violation data collected and cleaned from the "Four No's and Two Direct" inspection system will be inserted into the violation table according to the field correspondence; Video perception rate data mapping: inserting the calculated video perception rate data into a video perception rate table, wherein the video perception rate table includes at least a perception rate value and a timestamp of perception rate calculation; Job execution data mapping: associate the job execution data with the job ID and insert it into the corresponding field of the job execution table, using a specific code to indicate the completion status of the job steps; The ORM function implementation includes: Through the ORM framework, the data objects are mapped to the database table structure, so that when storing data, only the data objects need to be operated. The ORM framework automatically converts the state of the object into the corresponding SQL statement and executes the database operation. Use the query interface provided by the ORM framework to perform data query operations.

5. The remote inspection and result analysis system based on the ORM system according to claim 1 is characterized in that: The data sorting and analysis module uses the support vector machine (SVM) algorithm to classify traffic violations, including the following steps: Data preprocessing: The cleaned violation data is divided into a training set and a test set, and feature engineering is performed on the data; Model training: Use the training set data to train the SVM model; Model evaluation and optimization: Use the test set data to evaluate the performance of the trained SVM model; if the performance does not meet the requirements, further adjust the hyperparameters or increase the amount of data, optimize feature engineering, and retrain the model; Classification application: Apply the trained optimal SVM model to the newly collected violation data, predict the category of violation behavior, and count the frequency of occurrence of different types of violation behavior.

6. The remote inspection and result analysis system based on the ORM system according to claim 1 is characterized by: The data sorting and analysis module uses a multivariate linear regression algorithm to analyze factors that affect video perception rate, including the following steps: Data preparation: Collect data related to video perception rate, including light intensity data, object size data, and corresponding video perception rate data, and organize them into a data set, where each row of data represents a video sample, including light intensity, object size, and perception rate; Model construction and training: The data set is divided into a training set and a test set; a multivariate linear regression model is constructed using the training set data to determine the linear relationship coefficients between the independent variables including light intensity and target size and the dependent variables including perception rate; Model evaluation and testing: Evaluate the trained model, calculate the model's R-squared value and mean square error index to evaluate the model's fit to the data and prediction accuracy; perform hypothesis testing, check the significance of independent variables, exclude insignificant independent variables, and retrain the model to improve model performance; Factor analysis and prediction: Based on the trained model, analyze the impact of each independent variable on the video perception rate through the size and positive and negative analysis of the regression coefficient; use the model to predict the perception rate of new video data, and adjust the parameters of the video surveillance equipment or optimize the video acquisition environment based on the prediction results to improve the video perception rate.

7. The remote inspection and result analysis system based on the ORM system according to claim 1 is characterized in that: The data collation and analysis module uses a process mining algorithm based on Petri nets to analyze the compliance and efficiency of the operation process to describe the activities, sequence relationships, resource allocation, and concurrent execution in the operation process. By mining the operation execution log data, the differences between the actual operation process and the preset process model are discovered, and the compliance and efficiency of the process are evaluated, including the following steps: Data preprocessing: Extract the job step name, start time, end time, and execution resource information from the job log data obtained from the job execution monitoring system; then convert it into a format suitable for process mining algorithm processing; Model discovery: Use process mining algorithms to analyze event log data and build a job process model represented by a Petri net; automatically discover the structure of the job process based on the rules of mining the sequential and concurrent relationships between job steps. Compliance check: Compare the mining process model with the preset standard process model to check whether the actual process meets the standard requirements; by calculating the difference index between the two, determine the illegal operations or deviations from the standard process in the process; Efficiency analysis: Analyze the execution efficiency of the job process based on the job process model and job log data; calculate the average execution time of the job, the average waiting time of each job step, and the indicators of resource utilization.

8. The remote inspection and result analysis system based on the ORM system according to claim 1 is characterized by: Also includes: The user interaction interface module, the query window operates according to the user's request, submits the received data to the ORM, and the ORM analyzes the requested data, obtains the database where the data is located, sends a data request to the database, and passes the request result to the user interface view.

9. A remote inspection and result analysis method based on an ORM system, characterized in that: By collecting information including at least "four no's and two direct" inspections of violations, video perception rates, special reports on video responsibilities, and job execution, and cleaning the data and storing it in the database; using ORM to implement system integration of the safety risk management and supervision platform, analyze, judge, and mine the relationships between data, and generate accurate statistical results and reports according to predetermined rules and standards to identify the types and frequencies of violations, calculate video perception rates, and evaluate job execution.

10. The remote inspection and result analysis method based on the ORM system according to claim 9, characterized in that: According to the user's request operation, the received data is submitted to ORM, which analyzes the requested data, obtains the database where the data is located, sends a data request to the database, and passes the request result to the user interface view.

Images