An operation and maintenance efficiency evaluation method based on an operation and maintenance quality management database

Abstract

The application discloses an operation and maintenance efficiency evaluation method based on an operation and maintenance quality management database, and comprises the following steps: S1, establishing an operation and maintenance quality management database associated with asset and configuration information; S2, scanning a two-dimensional code to automatically call a security certificate and generate a push work order; S3, inputting a GPS displacement feature vector into an extended isolated forest model to calculate an abnormal score and lock a violation task through random slope hyperplane cutting; S4, extracting a response time length and a satisfaction index vector; S5, using an improved game theory combination weighting method and introducing an asymmetric game strategy based on the abnormal score, dynamically adjusting a penalty coefficient to solve an optimal weight, and generating a KPI quantitative score; and S6, aggregating and analyzing data to output a visual report. Through the random slope cutting and the asymmetric game mechanism, the operation and maintenance abnormality recognition precision and the examination fairness are effectively improved.

Description

Technical Field

[0001] This invention relates to the fields of intelligent operation and maintenance and computer data processing, and in particular to a method for evaluating operation and maintenance efficiency based on an operation and maintenance quality management database. Background Technology

[0002] Operations efficiency evaluation methods based on operations quality management databases have been widely applied in recent years in fields such as intelligent operations and maintenance, equipment management, and asset monitoring due to their ability to achieve accurate quantitative evaluation through multi-source data fusion and intelligent algorithms, becoming an important development direction for improving operations and maintenance management. However, in practical applications, large-scale operations and maintenance scenarios face many challenges, such as a wide variety of equipment assets, discrete and random operations and maintenance behaviors, and complex fault handling environments, and the deployment effectiveness of operations and maintenance efficiency evaluation models is still constrained by many factors.

[0003] Currently, most operational efficiency assessment methods rely on single-dimensional statistical indicators, making it difficult to fully utilize multi-source information such as time points, GPS movement trajectories, asset attributes, and user satisfaction. This results in a lack of comprehensiveness in the modeling of operational behavior. Some systems only use fixed-weight scoring standards to evaluate work order processing results, ignoring the dynamic correlation between multiple dimensions such as response time, processing efficiency, task difficulty, and subjective evaluation, thus limiting the adaptive adjustment capability of the assessment strategy. Furthermore, the anomaly judgment logic lacks a transparent explanation path, making it difficult to provide managers with specific criteria for judging abnormal behavior or inefficiency, affecting the credibility and usability of the assessment results.

[0004] Furthermore, existing evaluation methods mostly use static design for weight allocation in comprehensive score calculation. They fail to dynamically adjust the game relationship between subjective and objective weights based on the scores of abnormal behaviors detected during operation and maintenance. This results in some cheating behaviors or inefficient operations not being effectively punished in the score, making it difficult to adapt to the continuous changes and evolution of operation and maintenance quality risks, and seriously affecting the practical value and stability of the model in real-world scenarios.

[0005] Therefore, how to provide a method for evaluating operational efficiency based on an operational quality management database is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] One objective of this invention is to propose an operation and maintenance efficiency evaluation method based on an operation and maintenance quality management database. This invention fully integrates key steps such as work order lifecycle monitoring, GPS displacement feature extraction, extended isolated forest model anomaly detection, and an improved game theory-based combined weighting method, constructing an intelligent evaluation process with multi-source data association, spatiotemporal behavioral anomaly identification, dynamic constraints on anomaly scoring, and KPI-quantified scoring. This invention improves the accuracy of behavioral anomaly identification by introducing a random slope hyperplane into the extended isolated forest model, and dynamically adjusts the deviation penalty coefficient of subjective and objective weights using an asymmetric game strategy based on anomaly scoring, achieving adaptive correction of the evaluation mechanism. This invention has advantages such as rich anomaly detection dimensions, flexible game strategy response, and fair and transparent evaluation results. It can significantly improve the accuracy of behavioral monitoring, the adaptability of the scoring model, and the credibility of the final results in complex operation and maintenance scenarios, thereby effectively solving problems such as single behavioral modeling, static weight allocation, and difficulty in quantifying the impact of anomalies in existing methods.

[0007] An operation and maintenance efficiency evaluation method based on an operation and maintenance quality management database according to an embodiment of the present invention includes the following steps:

[0008] S1. Collect multi-source data, classify and store asset identification information, operation and maintenance plan configuration information, and core equipment security credentials, and output the operation and maintenance quality management database.

[0009] S2. Scan the device's identity QR code to trigger a repair request, parse the index and call the corresponding asset attribute information from the operation and maintenance quality management database to generate a basic work order, retrieve the matching core device security certificate and embed it into the basic work order, push it to the operation and maintenance engineer's terminal, and output the initial work order.

[0010] S3. Real-time monitoring of the entire lifecycle of the initial work order, recording key time nodes such as the order acceptance time, arrival time and completion time, combining GPS displacement to generate feature vector input to expand the isolated forest model, constructing a random slope hyperplane to cut the feature space, calculating the abnormal path length and score, verifying timeliness, locking the task if timeout or score exceeds the limit, and outputting operation and maintenance process data.

[0011] S4. Write the operation and maintenance process data into the operation and maintenance quality management database, generate operation and maintenance records, and perform data cleaning and vectorization on the time dimension and rating dimension in the operation and maintenance records. Extract the response time vector, processing time vector, asset type weight and user satisfaction scalar and output them.

[0012] S5. Improved game theory combined weighting method by inputting response time vector, processing time vector, asset type weight and user satisfaction scalar input, introduces asymmetric game strategy based on abnormal scores, dynamically calculates objective penalty coefficient and subjective penalty coefficient according to abnormal scores, constructs and solves a system of equations for the asymmetric error sum of squares matrix, calculates the optimal comprehensive weight vector, and generates KPI quantitative score.

[0013] S6. Extract the total number of work orders and fault type data from the operation and maintenance quality management database, perform aggregation analysis, calculate the month-on-month growth rate of the data in the current period and the historical period, identify the development trend of operation and maintenance events, and generate operation and maintenance quality analysis reports and visualization charts by combining KPI quantitative scoring and output them.

[0014] Optionally, S1 specifically includes:

[0015] S11. Obtain the original equipment file data, operation and maintenance scheduling rule data, and equipment login credential data through the data interface respectively;

[0016] S12. Parse the unique equipment identifier in the equipment file data, extract the text-formatted equipment identity QR code characters that correspond one-to-one with the unique equipment identifier, extract the geographical latitude and longitude coordinates data in the equipment file data as the physical location, and extract the equipment model and specifications parameters in the equipment file data as asset attributes.

[0017] S13. Parse the time limit parameters in the operation and maintenance scheduling rule data, extract the maximum time range allowed for inspection operations as the preset time window threshold, extract the standard inspection item list of inspection tasks as the inspection standard template, and extract the time interval rules for generating inspection tasks as periodic task rules.

[0018] S14. Parse the encrypted string in the device login credential data and extract the username and password combination used to log in to the core server as the core device security credential.

[0019] S15. Construct a related data table containing asset fields, plan fields, and voucher fields. Map and write the equipment identification QR code, physical location, asset attributes, preset time window threshold, inspection standard template, periodic task rules, and core equipment security voucher into the corresponding fields. Establish a one-to-one index relationship between the equipment identification QR code, the operation and maintenance plan configuration information, and the core equipment security voucher, and output the operation and maintenance quality management database.

[0020] Optionally, S2 specifically includes:

[0021] S21. Use the mobile device's camera to capture an image of the label pasted on the device surface, decode the label image, extract the string data contained in the label image, and use it as the device's identity QR code.

[0022] S22. Parse the device identification QR code to obtain the device's unique identification character. Use the unique identification character as a search index. Traverse the asset fields in the operation and maintenance quality management database, read the records that match the unique identification character, and extract the device model, specifications, and purchase date information stored in the records as asset attribute information.

[0023] S23. Create a blank record sheet containing the equipment unique identifier, equipment model, specifications and parameters and fault description fields. Automatically fill the extracted asset attribute information into the corresponding equipment model field, specifications and parameters field and purchase date field of the blank record sheet to generate a basic work order.

[0024] S24. Read the device model field in the basic work order, query the preset permission configuration table based on the device model field, obtain the operation permission level corresponding to the device model, find the corresponding private key file in the key library according to the operation permission level, decrypt the device login credential data using the private key file, restore the plaintext format username string and password string, and use the combination of username string and password string as the core device security credential.

[0025] S25. Add a hidden security field to the data structure of the basic work order, write the core equipment security certificate into the security field, encrypt and encapsulate the basic work order containing the core equipment security certificate, and generate an encrypted data packet.

[0026] S26. The encrypted data packet is sent to the mobile device corresponding to the preset maintenance engineer terminal identifier via the wireless communication network. The maintenance engineer terminal receives and decrypts the data packet, displays the work order details containing device information and core device security credentials, and outputs the initial work order.

[0027] Optionally, S3 specifically includes:

[0028] S31. Monitor the operation event stream of the maintenance engineer's terminal. When the button signal to confirm the order is detected, read and record the current server date and time as the order acceptance time. When the button signal to clock in is detected, read and record the current server date and time as the arrival time. When the button signal to submit the task is detected, read and record the current server date and time as the completion time.

[0029] S32. Read the data output by the global positioning sensor of the maintenance engineer terminal according to the preset sampling frequency, extract the longitude and latitude values ​​at the current sampling time, and extract the longitude and latitude values ​​at the previous sampling time. Convert the difference between the latitude value at the previous sampling time and the latitude value at the current sampling time into radians, and convert the difference between the longitude value at the previous sampling time and the longitude value at the current sampling time into radians.

[0030] S33. Take the sine value of each of the two radian values ​​and calculate the sum of squares. Take the cosine value of the latitude value at the previous sampling time and the latitude value at the current sampling time and multiply them. Take the cosine value of the difference between the two radian values. Subtract the cosine value from 1 and multiply it with the product of the cosine values ​​of the two radian values ​​to obtain an intermediate variable. Take the arcsine value of the intermediate variable and convert the radian value to an angle. Multiply the angle value by the Earth's radius and use the calculated value as GPS displacement data.

[0031] S34. Convert the order acceptance time, arrival time, and completion time into timestamp values. Convert the GPS displacement data calculated from all sampling times into displacement values. Concatenate the three timestamp values ​​with all displacement values ​​in chronological order to form a single high-dimensional numerical vector as a feature vector.

[0032] S35. Construct an extended isolated forest model. In the feature space, arbitrarily select a feature dimension and a value under the feature dimension as the split point. Read the millisecond value of the current server's system timestamp as the intermediate seed number. Square the intermediate seed number to obtain the square result. Take the middle four digits of the square result as the new seed number. Divide the new seed number by 10000 to obtain a division quotient between 0 and 1.

[0033] S36. Multiply the division quotient by 2 and subtract 1 to get a value between -1 and 1. Substitute the value into the arctangent function to calculate the slope value. Use the hyperplane that passes through the value and has the slope value to divide the space where the feature vector is located into two subspaces. Recursively execute the operation of arbitrarily selecting dimensions and values ​​and cutting within each subspace to build a binary tree structure until each subspace contains only one feature vector or reaches the preset maximum tree depth.

[0034] S37. Calculate the total number of edges traversed by the feature vector in the constructed binary tree structure from the root node to the leaf node. Use the total as the path length. Calculate the average of the path lengths of all binary trees. Use an exponential function to map the average to the interval between 0 and 1 to obtain the anomaly score.

[0035] S38. Read the server date and time when the inspection task is submitted as the current system time. Compare the current system time with the end time set by the preset time window threshold. If the current system time is later than the end time or the value of the abnormal score is greater than the preset threshold, change the status of the initial work order to locked, prohibit the modification or deletion of the work order data, and output the operation and maintenance process data.

[0036] Optionally, S4 specifically includes:

[0037] S41. Write the operation and maintenance process data into the corresponding data record table in the operation and maintenance quality management database. Subtract the order acceptance time from the completion time timestamp to obtain the total duration value. Record the total duration value as the time span. Extract the click operation records generated by the operation and maintenance engineer in the three stages of order acceptance, on-site arrival and completion as operation logs. Extract the user's rating star value after the task is completed as the result score. Combine the time span, operation logs and result score to generate operation and maintenance records.

[0038] S42. Extract the order acceptance time and arrival time from the operation and maintenance records. Subtract the order acceptance time from the arrival time to obtain the first time difference. Arrange the first time difference of all work orders in order to form a numerical sequence as the response duration vector.

[0039] S43. Extract the arrival time and completion time from the operation and maintenance records, subtract the arrival time from the completion time to obtain the second time difference, and arrange the second time difference of all work orders in order to form a numerical sequence as a processing time vector.

[0040] S44. Extract the equipment model field from the asset attribute information in the operation and maintenance records, query the preset equipment difficulty configuration table, read the value corresponding to the equipment model as the equipment difficulty coefficient, normalize the equipment difficulty coefficient, and use the normalized value as the asset type weight.

[0041] S45. Extract the result rating field from the operation and maintenance records, read the evaluation star rating value, divide the evaluation star rating value by the maximum star rating value to obtain the decimal ratio, and use the ratio as a scalar of user satisfaction.

[0042] S46. Output the response duration vector, processing duration vector, asset type weight, and user satisfaction scalar in the format of the data columns.

[0043] Optionally, S5 specifically includes:

[0044] S51. An improved game theory combined weighting method is used to input response time vector, processing time vector, asset type weight and user satisfaction scalar. The analytic hierarchy process is used to construct a judgment matrix and compare the four indicators of response time, processing time, asset type weight and user satisfaction in pairs.

[0045] S52. If indicator i is more important than indicator j, assign the value 2 to the element in the i-th row and j-th column of the judgment matrix; if they are equally important, assign the value 1; if indicator i is less important than indicator j, assign the value 0.5. Calculate the product of all elements in each column of the judgment matrix, take the fourth root of each column's product to obtain the geometric mean of each column, add the four geometric means to get the sum, divide each geometric mean by the sum, and arrange the four calculated ratios into a column vector in row order to obtain the subjective weight vector.

[0046] S53. The objective weight vector is calculated using the entropy method. The original values ​​in the response time vector, processing time vector, and user satisfaction scalar are divided by the sum of the values ​​in the corresponding columns to obtain the proportion matrix. The natural logarithm of each value in the proportion matrix is ​​taken and the result is multiplied by the value itself. The negative sum is obtained by summing the values ​​column by column. The entropy value of each indicator is obtained by subtracting the negative sum from 1 and dividing by the natural logarithm of the total number of indicators. The difference coefficient is obtained by subtracting the entropy value from 1. The difference coefficient of each indicator is divided by the sum of all difference coefficients to obtain the objective weight vector.

[0047] S54. Construct the objective function and introduce an asymmetric game strategy based on abnormal scoring. Calculate the dynamic objective penalty coefficient and the dynamic subjective penalty coefficient, solve to generate the optimal comprehensive weight vector, and calculate and output the KPI quantitative score.

[0048] Optionally, S54 specifically includes:

[0049] S541. Construct an objective function. Introduce an asymmetric game strategy based on anomaly scoring into the objective function. Read the anomaly scores in the operation and maintenance process data. If the anomaly score is less than the preset normal threshold, multiply the Euclidean norm of the subjective weight vector by the first initial penalty coefficient and multiply the Euclidean norm of the objective weight vector by the second initial penalty coefficient. Add the two products to construct a symmetric error sum of squares function as the objective function.

[0050] S542. If the abnormal score is greater than or equal to the preset normal threshold, calculate the difference between the abnormal score and the normal threshold, and use it as an index to be substituted into an exponential function with the natural constant e as the base to calculate the nonlinear growth factor. Multiply the second initial penalty coefficient by the nonlinear growth factor to obtain the dynamic objective penalty coefficient, and subtract the product of the nonlinear growth factor and the first initial penalty coefficient from the first initial penalty coefficient to obtain the dynamic subjective penalty coefficient.

[0051] S543. Set a combination coefficient vector to be solved consisting of four unknown variables. Calculate the sum of squares of the differences between corresponding elements of the subjective weight vector and the combination coefficient vector to be solved, and multiply the result by the dynamic subjective penalty coefficient. Calculate the sum of squares of the differences between corresponding elements of the objective weight vector and the combination coefficient vector to be solved, and multiply the result by the dynamic objective penalty coefficient. Add the two products to construct an asymmetric error sum of squares function.

[0052] S544. Take the derivative of the asymmetric error sum of squares matrix equation system and set the derivative to zero. Solve the linear equation system to obtain the combination coefficient vector to be solved as the optimal combination coefficient. Multiply the subjective weight vector with the subjective coefficient in the optimal combination coefficient, multiply the objective weight vector with the objective coefficient in the optimal combination coefficient, add the corresponding elements of the two product vectors, and calculate the optimal comprehensive weight vector.

[0053] S545. Read the equipment model field in the asset attribute information, query the preset difficulty configuration table to obtain the difficulty correction coefficient corresponding to the equipment model, multiply the first component of the optimal comprehensive weight vector with the element value of the corresponding work order in the response time vector, multiply the second component with the element value of the corresponding work order in the processing time vector, multiply the third component with the asset type weight, and multiply the fourth component with the element value of the corresponding work order in the user satisfaction scalar. Add the four products to obtain the basic score, divide the basic score by the difficulty correction coefficient, and generate the KPI quantitative score of the current work order.

[0054] Optionally, S6 specifically includes:

[0055] S61. Read the system clock to obtain the current date, and calculate the start and end time points backward from the current date according to the preset statistical period. Use the date subtraction function to calculate the date that is shifted forward by one period length from the start time point as the start and end time points of the previous period.

[0056] S62. Send a query command to the operation and maintenance quality management database, extract the number of all records that have been completed between the start and end time of the current cycle as the total number of work orders processed, and extract the fault type field stored in each record. Iterate through the fault type field and use a counting function to accumulate the frequency of each fault text.

[0057] S63. Read the total number of work orders processed in the previous period, subtract the total number of work orders processed in the previous period from the total number of work orders processed in the current period to get the numerical difference, divide the numerical difference by the total number of work orders processed in the previous period to get the ratio, and multiply the ratio by 100% to calculate the month-on-month growth rate of the data.

[0058] S64. Arrange the total number of work orders processed each day in the current period into a time series array in chronological order. Use the least squares method to perform linear fitting on the time series array to calculate the slope value. If the slope value is greater than 0, the development trend is determined to be upward. If the slope value is less than 0, the development trend is determined to be downward. Identify the development trend of operation and maintenance events.

[0059] S65. Read the KPI quantitative scores corresponding to all work order records in the current period from the operation and maintenance quality management database, add up all KPI quantitative scores and divide by the total number of work order records to obtain the arithmetic mean, write the development trend of operation and maintenance events, the frequency statistics of fault types and the arithmetic mean of KPI quantitative scores into the preset text template, and generate an operation and maintenance quality analysis report.

[0060] S66. Read the total number of work orders processed, the frequency of fault types, and the quantitative KPI score data from the operation and maintenance quality analysis report. Call the plotting library functions to draw a line chart with date as the horizontal axis and total amount as the vertical axis, a pie chart with fault type as the category, and a bar chart with engineer identification as the horizontal axis and score as the vertical axis. Insert the line chart, pie chart, and bar chart at the bottom of the operation and maintenance quality analysis report and output them.

[0061] The beneficial effects of this invention are:

[0062] This invention addresses the issues of single-dimensional O&M behavior data and low anomaly detection accuracy by constructing an O&M quality management database and extending the isolated forest model. It employs a high-dimensional feature vector constructed by fusing GPS displacement and key time nodes, introduces a random slope hyperplane to perform spatial cutting and path length calculation, and outputs accurate anomaly scores. Through multi-dimensional data cleaning and vectorization of response time, processing time, and other dimensions, a standardized evaluation index system is constructed. The standardized indicators are input into an improved game theory-based weighting method, and a penalty coefficient is dynamically calculated using an asymmetric game strategy based on anomaly scores. An asymmetric error sum-of-squares matrix equation system is constructed and solved to output the optimal comprehensive weight vector. In the scoring calculation stage, a weighted summation and correction are performed by combining the optimal comprehensive weight vector with a difficulty correction coefficient to generate a quantitative KPI score. Furthermore, historical periodic data is used to calculate the month-on-month growth rate and perform trend analysis to generate a visualized O&M quality analysis report. Ultimately, this achieves closed-loop management of real-time behavior monitoring, accurate anomaly identification, and quantitative efficiency evaluation throughout the entire lifecycle of O&M work orders, effectively improving the sensitivity of anomaly detection, the adaptability of weight allocation, and the scientific fairness of evaluation results. Attached Figure Description

[0063] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0064] Figure 1 This is a flowchart of an operation and maintenance efficiency evaluation method based on an operation and maintenance quality management database proposed in this invention;

[0065] Figure 2 This is a flowchart of the spatiotemporal feature vector generation and anomaly scoring calculation based on the extended isolated forest model proposed in this invention;

[0066] Figure 3 This is a flowchart of the improved game theory combinatorial weighting method for optimization and KPI quantification scoring based on asymmetric game strategy proposed in this invention. Detailed Implementation

[0067] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0068] refer to Figures 1-3 An operation and maintenance efficiency evaluation method based on an operation and maintenance quality management database includes the following steps:

[0069] S1. Collect multi-source data and classify, store and associate asset identification information, operation and maintenance plan configuration information and core equipment security credentials. The asset identification information includes equipment identity QR code, physical location and asset attributes. The operation and maintenance plan configuration information includes preset time window threshold, inspection standard template and periodic task rules. Output an operation and maintenance quality management database with established association relationships.

[0070] S2. Scan the device's identity QR code to trigger a repair request, parse the index and call the corresponding asset attribute information from the operation and maintenance quality management database to generate a basic work order. Based on the preset permission policy and security encryption algorithm, retrieve the matching core device security credentials and embed them into the basic work order, then push it to the operation and maintenance engineer's terminal to output the initial work order.

[0071] S3. Real-time monitoring of the entire lifecycle of the initial work order, recording key time nodes such as the order acceptance time, arrival time, and completion time, combining GPS displacement to generate feature vectors to expand the isolated forest model, constructing a random slope hyperplane to cut the feature space, calculating the abnormal path length and score, verifying timeliness, locking the task if timeout or score exceeds the limit, and outputting operation and maintenance process data containing the key time nodes, abnormal scores, and timeliness verification.

[0072] S4. Write the operation and maintenance process data into the operation and maintenance quality management database to generate operation and maintenance records containing time span, operation logs and result scores. Then, perform data cleaning and vectorization on the time dimension and score dimension in the operation and maintenance records to extract response time vector, processing time vector, asset type weight and user satisfaction scalar and output them.

[0073] S5. Improved game theory combined weighting method by inputting response time vector, processing time vector, asset type weight and user satisfaction scalar input, introduces asymmetric game strategy based on abnormal scores, dynamically calculates objective penalty coefficient and subjective penalty coefficient according to abnormal scores, constructs and solves a system of equations for the asymmetric error sum of squares matrix, calculates the optimal comprehensive weight vector, and generates KPI quantitative score.

[0074] S6. Based on the preset statistical period, extract the total number of work orders and fault type data from the operation and maintenance quality management database, perform aggregation analysis, calculate the month-on-month growth rate of the current period and the historical period, identify the development trend of operation and maintenance events, and generate and output operation and maintenance quality analysis reports and visualization charts by combining KPI quantitative scoring.

[0075] This invention significantly improves the precision of operations and maintenance (O&M) management and the accuracy of performance evaluation. By establishing an O&M quality management database, it achieves the associated configuration of multi-source data, asset identifiers, and security credentials, ensuring the security of work order generation and permission access. Utilizing an extended isolated forest model for real-time monitoring of the entire lifecycle, and constructing a hyperplane with random slope to cut the feature space, it accurately identifies abnormal paths and calculates abnormal scores, effectively locking down timed-out or non-compliant tasks. Combined with an improved game theory-based weighting method, it introduces an asymmetric game strategy based on abnormal scores to dynamically adjust subjective and objective weights, generating high-precision KPI quantitative scores and solving the problem of excessive subjectivity in traditional assessments. Simultaneously, the system performs aggregate analysis and year-on-year calculations on the total number of work orders and fault types, automatically outputting a visualized O&M quality analysis report. This method significantly improves the detection rate of false performance and the first-time fault repair rate while significantly reducing response time and customer complaint rates, achieving intelligent and dynamic management of O&M quality assessment.

[0076] In this embodiment, S1 specifically includes:

[0077] S11. Obtain the original equipment file data, operation and maintenance scheduling rule data, and equipment login credential data through the data interface respectively;

[0078] S12. Parse the unique equipment identifier in the equipment file data, extract the text-formatted equipment identity QR code characters that correspond one-to-one with the unique equipment identifier, extract the geographical latitude and longitude coordinates data in the equipment file data as the physical location, and extract the equipment model and specifications parameters in the equipment file data as asset attributes.

[0079] S13. Parse the time limit parameters in the operation and maintenance scheduling rule data, extract the maximum time range allowed for inspection operations as the preset time window threshold, extract the standard inspection item list of inspection tasks as the inspection standard template, and extract the time interval rules for generating inspection tasks as periodic task rules.

[0080] S14. Parse the encrypted string in the device login credential data and extract the username and password combination used to log in to the core server as the core device security credential.

[0081] S15. Construct a related data table containing asset fields, plan fields, and voucher fields. Map and write the equipment identification QR code, physical location, asset attributes, preset time window threshold, inspection standard template, periodic task rules, and core equipment security voucher into the corresponding fields. Establish a one-to-one index relationship between the equipment identification QR code, the operation and maintenance plan configuration information, and the core equipment security voucher. Output the operation and maintenance quality management database with the established relationship.

[0082] In this embodiment, S2 specifically includes:

[0083] S21. Use the mobile device's camera to capture an image of the label pasted on the device surface, decode the label image, extract the string data contained in the label image, and use it as the device's identity QR code.

[0084] S22. Parse the device identification QR code to obtain the device's unique identification character. Use the unique identification character as a search index. Traverse the asset fields in the operation and maintenance quality management database, read the records that match the unique identification character, and extract the device model, specifications, and purchase date information stored in the records as asset attribute information.

[0085] S23. Create a blank record sheet containing the equipment unique identifier, equipment model, specifications and parameters and fault description fields. Automatically fill the extracted asset attribute information into the corresponding equipment model field, specifications and parameters field and purchase date field of the blank record sheet to generate a complete basic work order.

[0086] S24. Read the device model field in the basic work order, query the preset permission configuration table based on the device model field, obtain the operation permission level corresponding to the device model, find the corresponding private key file in the key library according to the operation permission level, decrypt the device login credential data using the private key file, restore the plaintext format username string and password string, and use the combination of username string and password string as the core device security credential.

[0087] S25. Add a hidden security field to the data structure of the basic work order, write the core equipment security certificate into the security field, encrypt and encapsulate the basic work order containing the core equipment security certificate, and generate an encrypted data packet.

[0088] S26. The encrypted data packet is sent to the mobile device corresponding to the preset maintenance engineer terminal identifier via the wireless communication network. The maintenance engineer terminal receives and decrypts the data packet, displays the work order details containing device information and core device security credentials, and outputs the initial work order.

[0089] In this embodiment, S3 specifically includes:

[0090] S31. Monitor the operation event stream of the maintenance engineer's terminal. When the button signal to confirm the order is detected, read and record the current server date and time as the order acceptance time. When the button signal to clock in is detected, read and record the current server date and time as the arrival time. When the button signal to submit the task is detected, read and record the current server date and time as the completion time.

[0091] S32. Read the data output by the global positioning sensor of the maintenance engineer terminal according to the preset sampling frequency, extract the longitude and latitude values ​​at the current sampling time, and extract the longitude and latitude values ​​at the previous sampling time. Convert the difference between the latitude value at the previous sampling time and the latitude value at the current sampling time into radians, and convert the difference between the longitude value at the previous sampling time and the longitude value at the current sampling time into radians. The preset sampling frequency is once per second.

[0092] S33. Take the sine value of each of the two radian values ​​and calculate the sum of squares. Take the cosine value of the latitude value at the previous sampling time and the latitude value at the current sampling time and multiply them. Take the cosine value of the difference between the two radian values. Subtract the cosine value from 1 and multiply it with the product of the cosine values ​​of the two radian values ​​to obtain an intermediate variable. Take the arcsine value of the intermediate variable and convert the radian value to an angle. Multiply the angle value by the Earth's radius and use the calculated value as GPS displacement data.

[0093] S34. Convert the order acceptance time, arrival time, and completion time into timestamp values. Convert the GPS displacement data calculated from all sampling times into displacement values. Concatenate the three timestamp values ​​with all displacement values ​​in chronological order to form a single high-dimensional numerical vector as a feature vector.

[0094] S35. Construct an extended isolated forest model. In the feature space, arbitrarily select a feature dimension and a value under the feature dimension as the split point. Read the millisecond value of the current server's system timestamp as the intermediate seed number. Square the intermediate seed number to obtain the square result. Take the middle four digits of the square result as the new seed number. Divide the new seed number by 10000 to obtain a division quotient between 0 and 1.

[0095] S36. Multiply the division quotient by 2 and subtract 1 to obtain a value between -1 and 1. Substitute the value into the arctangent function to calculate the slope value. Use a hyperplane that passes through the value and has the slope value to divide the space where the feature vector is located into two subspaces. Recursively perform the operation of arbitrarily selecting dimensions and values ​​and cutting within each subspace to construct a binary tree structure until each subspace contains only one feature vector or reaches the preset maximum tree depth, which is 8.

[0096] S37. Calculate the total number of edges traversed by the feature vector in the constructed binary tree structure from the root node to the leaf node. Use the total as the path length. Calculate the average of the path lengths of all binary trees. Use an exponential function to map the average to the interval between 0 and 1 to obtain the abnormal score of the operation and maintenance behavior data.

[0097] S38. Read the server date and time when the inspection task is submitted as the current system time. Compare the current system time with the end time set by the preset time window threshold. If the current system time is later than the end time or the value of the abnormal score is greater than the preset threshold, change the status of the initial work order to locked, prohibit modification or deletion of the work order data, and output the operation and maintenance process data containing key time nodes, abnormal scores and timeliness verification. The preset threshold is 0.6.

[0098] This implementation method, by constructing an extended isolated forest model and introducing a random slope hyperplane cutting mechanism, offers significant advantages over traditional isolated forests and other anomaly detection algorithms. Traditional isolated forests typically employ a cutting method perpendicular to the coordinate axes, which struggles to effectively capture the nonlinear coupling relationships between features when processing complex spatiotemporal trajectory and time series mixed data, resulting in low accuracy in identifying concealed violations. In contrast, this invention utilizes random seeds generated by server timestamps to construct a dynamic slope hyperplane, achieving multi-angle, non-orthogonal cutting of the feature space, significantly improving the separation and detection sensitivity of anomalous trajectories in high-dimensional feature spaces.

[0099] Simultaneously, by combining precise GPS displacement calculation with real-time monitoring of key time nodes, this method can accurately identify false performance of duties and process violations, quantifying anomaly scores to a range of 0 to 1. Based on a 0.6 threshold judgment logic and a timeout locking mechanism, it effectively prevents data tampering and subsequent data entry. This solution significantly improves the detection rate of anomalies in operational behavior while achieving automated and intelligent supervision of the entire lifecycle of work orders, ensuring the authenticity and timeliness of operational data.

[0100] In this embodiment, S4 specifically includes:

[0101] S41. Write the operation and maintenance process data, which includes key time nodes, abnormal scores and has been verified for timeliness, into the corresponding data record table in the operation and maintenance quality management database. Subtract the order acceptance time from the completion time timestamp to obtain the total duration value. Record the total duration value as the time span. Extract the click operation records generated by the operation and maintenance engineer in the three stages of order acceptance, on-site arrival and completion as operation logs. Extract the user's rating star value after the task is completed as the result score. Combine the time span, operation logs and result scores to generate operation and maintenance records.

[0102] S42. Extract the order acceptance time and arrival time from the operation and maintenance records. Subtract the order acceptance time from the arrival time to obtain the first time difference. Arrange the first time difference of all work orders in order to form a numerical sequence as the response duration vector.

[0103] S43. Extract the arrival time and completion time from the operation and maintenance records, subtract the arrival time from the completion time to obtain the second time difference, and arrange the second time difference of all work orders in order to form a numerical sequence as a processing time vector.

[0104] S44. Extract the equipment model field from the asset attribute information in the operation and maintenance records, query the preset equipment difficulty configuration table, read the value corresponding to the equipment model as the equipment difficulty coefficient, normalize the equipment difficulty coefficient, and use the normalized value as the asset type weight.

[0105] S45. Extract the result rating field from the operation and maintenance records, read the evaluation star rating value, divide the evaluation star rating value by the maximum star rating value to obtain the decimal ratio, and use the ratio as a scalar of user satisfaction.

[0106] S46. Output the response duration vector, processing duration vector, asset type weight, and user satisfaction scalar in the format of data columns for subsequent evaluation calculations.

[0107] In this embodiment, S5 specifically includes:

[0108] S51. An improved game theory combined weighting method is used to input response time vector, processing time vector, asset type weight and user satisfaction scalar. The analytic hierarchy process is used to construct a judgment matrix and compare the four indicators of response time, processing time, asset type weight and user satisfaction in pairs.

[0109] S52. If indicator i is more important than indicator j, assign the value 2 to the element in the i-th row and j-th column of the judgment matrix; if they are equally important, assign the value 1; if indicator i is less important than indicator j, assign the value 0.5. Calculate the product of all elements in each column of the judgment matrix, take the fourth root of each column's product to obtain the geometric mean of each column, add the four geometric means to get the sum, divide each geometric mean by the sum, and arrange the four calculated ratios into a column vector in row order to obtain the subjective weight vector.

[0110] S53. The objective weight vector is calculated using the entropy method. The original values ​​in the response time vector, processing time vector, and user satisfaction scalar are divided by the sum of the values ​​in the corresponding columns to obtain the proportion matrix. The natural logarithm of each value in the proportion matrix is ​​taken and the result is multiplied by the value itself. The negative sum is obtained by summing the values ​​column by column. The entropy value of each indicator is obtained by subtracting the negative sum from 1 and dividing by the natural logarithm of the total number of indicators. The difference coefficient is obtained by subtracting the entropy value from 1. The difference coefficient of each indicator is divided by the sum of all difference coefficients to obtain the objective weight vector.

[0111] S54. Construct the objective function and introduce an asymmetric game strategy based on abnormal scoring. Calculate the dynamic objective penalty coefficient and the dynamic subjective penalty coefficient, solve to generate the optimal comprehensive weight vector, and calculate and output the KPI quantitative score.

[0112] In this embodiment, S54 specifically includes:

[0113] S541. Construct an objective function that minimizes the deviation between the combined weights and the subjective and objective weights. Introduce an asymmetric game strategy based on abnormal scores into the objective function. Read the abnormal scores in the operation and maintenance process data. If the abnormal score is less than the preset normal threshold, multiply the Euclidean norm of the subjective weight vector by a first initial penalty coefficient and multiply the Euclidean norm of the objective weight vector by a second initial penalty coefficient. Add the two products to construct a symmetric error sum of squares function as the objective function. The preset normal threshold is 0.6, the first initial penalty coefficient is 0.6, and the second initial penalty coefficient is 0.4.

[0114] S542. If the abnormal score is greater than or equal to the preset normal threshold, calculate the difference between the abnormal score and the normal threshold, and use it as an index to be substituted into an exponential function with the natural constant e as the base to calculate the nonlinear growth factor. Multiply the second initial penalty coefficient by the nonlinear growth factor to obtain the dynamic objective penalty coefficient, and subtract the product of the nonlinear growth factor and the first initial penalty coefficient from the first initial penalty coefficient to obtain the dynamic subjective penalty coefficient.

[0115] S543. Set a combination coefficient vector to be solved consisting of four unknown variables. Calculate the sum of squares of the differences between corresponding elements of the subjective weight vector and the combination coefficient vector to be solved, and multiply the result by the dynamic subjective penalty coefficient. Calculate the sum of squares of the differences between corresponding elements of the objective weight vector and the combination coefficient vector to be solved, and multiply the result by the dynamic objective penalty coefficient. Add the two products to construct an asymmetric error sum of squares function.

[0116] S544. Take the derivative of the asymmetric error sum of squares matrix equation system and set the derivative to zero. Solve the linear equation system to obtain the combination coefficient vector to be solved as the optimal combination coefficient. Multiply the subjective weight vector with the subjective coefficient in the optimal combination coefficient, multiply the objective weight vector with the objective coefficient in the optimal combination coefficient, add the corresponding elements of the two product vectors, and calculate the optimal comprehensive weight vector.

[0117] S545. Read the equipment model field in the asset attribute information, query the preset difficulty configuration table to obtain the difficulty correction coefficient corresponding to the equipment model, multiply the first component of the optimal comprehensive weight vector with the element value of the corresponding work order in the response time vector, multiply the second component with the element value of the corresponding work order in the processing time vector, multiply the third component with the asset type weight, and multiply the fourth component with the element value of the corresponding work order in the user satisfaction scalar. Add the four products to obtain the basic score, divide the basic score by the difficulty correction coefficient, and generate the KPI quantitative score of the current work order. The preset difficulty correction coefficient is 1.2.

[0118] This implementation method achieves dynamic and precise allocation of KPI assessment weights by introducing an improved game theory-based combined weighting method. Compared to the traditional fixed-weight assignment method, this method combines the analytic hierarchy process (AHP) to obtain subjective weights with the entropy method to calculate objective weights, effectively balancing managerial experience with the value of objective data. In particular, the introduction of an asymmetric game strategy based on abnormal scores automatically generates a non-linear growth factor using an exponential function when abnormal scores exceed a threshold, significantly increasing the objective weight penalty and reducing the influence of subjective weights, ensuring zero tolerance for violations in the assessment. The basic scores are differentiated by incorporating an equipment difficulty correction coefficient, making the quantitative results more consistent with actual operational and maintenance difficulties. This mechanism significantly improves the fairness and incentive effect of performance evaluation while suppressing human cheating and ensuring data authenticity.

[0119] The improved game theory combinatorial weighting method of this invention is similar to the traditional game theory combinatorial weighting method in that both retain the core architecture of combinatorial weighting, namely, both use the analytic hierarchy process to calculate the subjective weight vector, use the entropy method to calculate the objective weight vector, and construct an objective function to minimize the deviation between the combined weight and the subjective and objective weights. Finally, both obtain the optimal comprehensive weight vector by solving the system of equations of the sum of squared errors.

[0120] The difference lies in that this invention breaks away from the limitations of traditional methods that use fixed weight coefficients or symmetric error functions for game theory, and introduces an asymmetric game strategy based on anomaly scoring. Traditional methods typically use static penalty coefficients to construct a symmetric error sum-of-squares function, which cannot adapt to dynamic anomalies during operation and maintenance. In contrast, this invention introduces a dynamic adjustment mechanism in step S54 using the anomaly score output from step S3. Specifically, by substituting the difference between the anomaly score and the normal threshold into an exponential function to calculate a nonlinear growth factor, the second initial penalty coefficient is multiplied and amplified to obtain a dynamic objective penalty coefficient. Simultaneously, the first initial penalty coefficient is reduced to obtain a dynamic subjective penalty coefficient, thereby constructing an asymmetric error sum-of-squares function.

[0121] Based on the aforementioned improvements, the beneficial effect of this invention lies in the introduction of an asymmetric game strategy based on anomaly scoring. The improved game theory-based weighting method can dynamically adjust the game position of subjective and objective weights according to the degree of anomaly in operational behavior. When a high anomaly score is detected, the system automatically amplifies the game weight of objective data and significantly suppresses subjective weights, effectively preventing operational personnel from using subjective evaluations to cover up violations and significantly improving the fairness and anti-interference capability of the assessment. This dynamic penalty mechanism enables KPI quantitative scoring to accurately reflect operational quality, solving the problem of evaluation distortion in traditional static methods when dealing with false performance or operational anomalies.

[0122] In this embodiment, S6 specifically includes:

[0123] S61. Read the system clock to obtain the current date, and determine the start and end time points by calculating backwards from the current date according to the preset statistical period. Use the date subtraction function to calculate the date that is shifted forward by one period length from the start time point as the start and end time points of the previous period. The preset statistical period is 30 days.

[0124] S62. Send a query command to the operation and maintenance quality management database, extract the number of all records that have been completed between the start and end time of the current cycle as the total number of work orders processed, and extract the fault type field stored in each record. Iterate through the fault type field and use a counting function to accumulate the frequency of each fault text.

[0125] S63. Read the total number of work orders processed in the previous period, subtract the total number of work orders processed in the previous period from the total number of work orders processed in the current period to get the numerical difference, divide the numerical difference by the total number of work orders processed in the previous period to get the ratio, and multiply the ratio by 100% to calculate the month-on-month growth rate of the data.

[0126] S64. Arrange the total number of work orders processed each day in the current period into a time series array in chronological order. Use the least squares method to perform linear fitting on the time series array to calculate the slope value. If the slope value is greater than 0, the development trend is determined to be upward. If the slope value is less than 0, the development trend is determined to be downward. Identify the development trend of operation and maintenance events.

[0127] S65. Read the KPI quantitative scores corresponding to all work order records in the current period from the operation and maintenance quality management database, add up all KPI quantitative scores and divide by the total number of work order records to obtain the arithmetic mean, write the development trend of operation and maintenance events, the frequency statistics of fault types and the arithmetic mean of KPI quantitative scores into the preset text template, and generate an operation and maintenance quality analysis report.

[0128] S66. Read the total number of work orders processed, the frequency of fault types, and the quantitative KPI score data from the operation and maintenance quality analysis report. Call the plotting library functions to draw a line chart with date as the horizontal axis and total amount as the vertical axis, a pie chart with fault type as the category, and a bar chart with engineer identification as the horizontal axis and score as the vertical axis. Insert the line chart, pie chart, and bar chart at the bottom of the operation and maintenance quality analysis report and output them.

[0129] Example 1: To verify the practical application feasibility of this invention in the field of operation and maintenance efficiency assessment, it was applied to the smart municipal infrastructure operation and maintenance management platform of a large coastal city. This platform is responsible for the daily inspection and emergency repair maintenance of over 2,400 kilometers of gas pipelines, 560 pressure regulating stations, and over 32,000 outdoor municipal devices in the city's main urban area. It strictly adheres to "standardized and compliant operation and maintenance management processes and systems," aiming to effectively monitor the entire operation and maintenance work order execution process and time nodes, thereby helping operation and maintenance managers effectively supervise the details of the operation and maintenance personnel's work process, thus improving daily operation and maintenance efficiency and reducing the investment in operation and maintenance manpower costs.

[0130] The operations and maintenance (O&M) team in this scenario is large, comprising 12 regional maintenance centers and over 4,500 frontline O&M engineers. In actual operations, traditional management methods face significant challenges, particularly for critical security devices such as core servers and bastion hosts. Traditional account and password management methods are prone to leakage or loss due to the reliance on documents or paper records. Furthermore, the lack of effective oversight during routine inspections often leads to employees slacking off or falsifying inspection plans before they are completed. This invention, by deeply integrating O&M big data analytics capabilities and utilizing an extended isolated forest model combined with an improved game theory-based weighting method, constructs an intelligent and traceable quality management system.

[0131] In practical applications, this invention fully covers the entire lifecycle from "event reporting" to "approval." When equipment malfunctions, on-site personnel scan the equipment identification QR code provided by the asset management system to report the problem. The system automatically dispatches the faulty equipment's asset information to the maintenance engineer's mobile phone, effectively simplifying the asset repair process. During the work order processing stage, the system first calls an extended isolated forest model to perform in-depth analysis of the collected GPS displacement data and time node vectors. By introducing a random slope hyperplane to cut the feature space, it accurately identifies abnormal behaviors such as trajectory anomalies or processing times that deviate significantly from the model, and calculates and outputs anomaly scores in real time.

[0132] Meanwhile, this invention introduces an improved game theory-based combined weighting method to address dynamic changes. The algorithm utilizes the anomaly scores from the above outputs to construct a penalty mechanism based on an asymmetric game strategy: when the anomaly score is high, the game weight of the objective evaluation dimension is dynamically increased, while the weight of the subjective evaluation dimension is decreased, thereby constructing and solving an asymmetric error sum-of-squares function to obtain the optimal comprehensive weight vector. Especially when processing "change" or "maintenance" work orders involving core equipment, the algorithm automatically strengthens the weight proportion of high-risk operations by combining monitoring data of secure encryption operations. Finally, the system uses the optimal comprehensive weight vector, combined with indicators such as response time and processing time, to calculate a quantitative KPI score, urging maintenance personnel to process work orders promptly and saving maintenance records for later traceability.

[0133] For the system's pre-set periodic tasks, this invention automatically filters and summarizes the number of maintenance work orders generated by various types of equipment based on equipment classification. For different types of work orders such as "maintenance, change, inspection, and rack / remove," the system can automatically identify their characteristic differences. Especially when handling "change" or "maintenance" work orders involving core equipment, the system automatically strengthens the monitoring weight of permission settings and security encryption algorithm operations to ensure the compliance of high-security operations. After maintenance is completed, the system calculates data from multiple dimensions, including work order response time, on-site arrival time, maintenance processing time, and user quality evaluation, to comprehensively evaluate the performance of maintenance engineers, generate KPI statistical reports, urge maintenance personnel to process work orders promptly, and save maintenance records for later traceability.

[0134] Table 1 below shows the core performance comparison data between the method of this invention and the traditional manual sampling method in four typical operation and maintenance areas during a three-month trial operation period. This data fully demonstrates the significant achievements of this invention in "improving efficiency, reducing costs," and realizing "intelligent operation and maintenance trend analysis."

[0135] Table 1. Multi-dimensional comparison data between the present invention and traditional methods

[0136]

[0137] Based on the comparative data shown in Table 1, it can be seen that the operation and maintenance assessment method based on extended isolated forest and asymmetric game strategy proposed in this invention has significant performance advantages over the traditional manual sampling method in the operation and maintenance management of municipal facilities. In particular, it has achieved comprehensive improvement in key indicators such as work order processing efficiency, risk identification capability, maintenance quality and customer satisfaction.

[0138] In terms of work order processing efficiency, this invention achieved a significant performance improvement in all four typical operation and maintenance areas. The average daily work order volume increased from 102.5 orders using the traditional method to 126.3 orders, indicating that the system significantly improved operation and maintenance throughput through intelligent scheduling. Simultaneously, the average response time was drastically reduced from 52.2 minutes using the traditional method to 28.7 minutes, a response speedup of nearly 45%. For example, in the geographically vast and task-heavy area D (remote suburban pipeline network), the traditional response time averaged 62.3 minutes, while this invention only required 33.5 minutes, greatly improving the real-time handling capability for sudden faults in remote areas.

[0139] In terms of risk identification and process control, this invention effectively addresses the industry pain point of difficulty in supervising "false performance of duties" through precise GPS displacement tracking and anomaly scoring models. The overall false performance detection rate has increased from 3.9% using traditional methods to 17.2%, particularly in Region D, where the detection rate reaches as high as 21.3%. This effectively eliminates the phenomena of "absenteeism" and "overdue inspection registration," ensuring the authenticity and compliance of the operation and maintenance execution process.

[0140] Regarding repair quality and customer satisfaction, this invention guides employees to focus on effectiveness through scientific KPI assessments. The overall first-time fault repair rate has increased from 81.2% to 93.5%, while the overall customer complaint rate has decreased from 4.4% to 1.15%. This data shows that this invention not only compels employees to improve their technical skills through objective data supervision, but also significantly enhances the user experience through standardized service processes.

[0141] Overall, this invention achieves efficient, accurate, and interpretable operation and maintenance quality assessment by integrating multi-source data to construct dynamic state maps in differentiated scenarios such as "commercial core areas", "old residential areas", "industrial parks" and "suburban pipeline networks", and by combining an improved game theory mechanism. It has broad practical application value.

[0142] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for evaluating operational efficiency based on an operational quality management database, characterized in that, Includes the following steps: S1. Collect multi-source data, classify and store asset identification information, operation and maintenance plan configuration information, and core equipment security credentials, and output the operation and maintenance quality management database. S2. Scan the device's identity QR code to trigger a repair request, parse the index and call the corresponding asset attribute information from the operation and maintenance quality management database to generate a basic work order, retrieve the matching core device security certificate and embed it into the basic work order, push it to the operation and maintenance engineer's terminal, and output the initial work order. S3. Real-time monitoring of the entire lifecycle of the initial work order, recording key time nodes such as the order acceptance time, arrival time and completion time, combining GPS displacement to generate feature vector input to expand the isolated forest model, constructing a random slope hyperplane to cut the feature space, calculating the abnormal path length and score, verifying timeliness, locking the task if timeout or score exceeds the limit, and outputting operation and maintenance process data. S4. Write the operation and maintenance process data into the operation and maintenance quality management database, generate operation and maintenance records, and perform data cleaning and vectorization on the time dimension and rating dimension in the operation and maintenance records. Extract the response time vector, processing time vector, asset type weight and user satisfaction scalar and output them. S5. Improved game theory combined weighting method by inputting response time vector, processing time vector, asset type weight and user satisfaction scalar input, introduces asymmetric game strategy based on abnormal scores, dynamically calculates objective penalty coefficient and subjective penalty coefficient according to abnormal scores, constructs and solves a system of equations for the asymmetric error sum of squares matrix, calculates the optimal comprehensive weight vector, and generates KPI quantitative score. S6. Extract the total number of work orders and fault type data from the operation and maintenance quality management database, perform aggregation analysis, calculate the month-on-month growth rate of the data in the current period and the historical period, identify the development trend of operation and maintenance events, and generate operation and maintenance quality analysis reports and visualization charts by combining KPI quantitative scoring and output them.

2. The method for evaluating operation and maintenance efficiency based on an operation and maintenance quality management database according to claim 1, characterized in that, S1 specifically includes: S11. Obtain the original equipment file data, operation and maintenance scheduling rule data, and equipment login credential data through the data interface respectively; S12. Parse the unique equipment identifier in the equipment file data, extract the text-formatted equipment identity QR code characters that correspond one-to-one with the unique equipment identifier, extract the geographical latitude and longitude coordinates data in the equipment file data as the physical location, and extract the equipment model and specifications parameters in the equipment file data as asset attributes. S13. Parse the time limit parameters in the operation and maintenance scheduling rule data, extract the maximum time range allowed for inspection operations as the preset time window threshold, extract the standard inspection item list of inspection tasks as the inspection standard template, and extract the time interval rules for generating inspection tasks as periodic task rules. S14. Parse the encrypted string in the device login credential data and extract the username and password combination used to log in to the core server as the core device security credential. S15. Construct a related data table containing asset fields, plan fields, and voucher fields. Map and write the equipment identification QR code, physical location, asset attributes, preset time window threshold, inspection standard template, periodic task rules, and core equipment security voucher into the corresponding fields. Establish a one-to-one index relationship between the equipment identification QR code, the operation and maintenance plan configuration information, and the core equipment security voucher, and output the operation and maintenance quality management database.

3. The method for evaluating operation and maintenance efficiency based on an operation and maintenance quality management database according to claim 1, characterized in that, S2 specifically includes: S21. Use the mobile device's camera to capture an image of the label pasted on the device surface, decode the label image, extract the string data contained in the label image, and use it as the device's identity QR code. S22. Parse the device identification QR code to obtain the device's unique identification character. Use the unique identification character as a search index. Traverse the asset fields in the operation and maintenance quality management database, read the records that match the unique identification character, and extract the device model, specifications, and purchase date information stored in the records as asset attribute information. S23. Create a blank record sheet containing the equipment unique identifier, equipment model, specifications and parameters and fault description fields. Automatically fill the extracted asset attribute information into the corresponding equipment model field, specifications and parameters field and purchase date field of the blank record sheet to generate a basic work order. S24. Read the device model field in the basic work order, query the preset permission configuration table based on the device model field, obtain the operation permission level corresponding to the device model, find the corresponding private key file in the key library according to the operation permission level, decrypt the device login credential data using the private key file, restore the plaintext format username string and password string, and use the combination of username string and password string as the core device security credential. S25. Add a hidden security field to the data structure of the basic work order, write the core equipment security certificate into the security field, encrypt and encapsulate the basic work order containing the core equipment security certificate, and generate an encrypted data packet. S26. The encrypted data packet is sent to the mobile device corresponding to the preset maintenance engineer terminal identifier via the wireless communication network. The maintenance engineer terminal receives and decrypts the data packet, displays the work order details containing device information and core device security credentials, and outputs the initial work order.

4. The method for evaluating operation and maintenance efficiency based on an operation and maintenance quality management database according to claim 1, characterized in that, S3 specifically includes: S31. Monitor the operation event stream of the maintenance engineer's terminal. When the button signal to confirm the order is detected, read and record the current server date and time as the order acceptance time. When the button signal to clock in is detected, read and record the current server date and time as the arrival time. When the button signal to submit the task is detected, read and record the current server date and time as the completion time. S32. Read the data output by the global positioning sensor of the maintenance engineer terminal according to the preset sampling frequency, extract the longitude and latitude values ​​at the current sampling time, and extract the longitude and latitude values ​​at the previous sampling time. Convert the difference between the latitude value at the previous sampling time and the latitude value at the current sampling time into radians, and convert the difference between the longitude value at the previous sampling time and the longitude value at the current sampling time into radians. S33. Take the sine value of each of the two radian values ​​and calculate the sum of squares. Take the cosine value of the latitude value at the previous sampling time and the latitude value at the current sampling time and multiply them. Take the cosine value of the difference between the two radian values. Subtract the cosine value from 1 and multiply it with the product of the cosine values ​​of the two radian values ​​to obtain an intermediate variable. Take the arcsine value of the intermediate variable and convert the radian value to an angle. Multiply the angle value by the Earth's radius and use the calculated value as GPS displacement data. S34. Convert the order acceptance time, arrival time, and completion time into timestamp values. Convert the GPS displacement data calculated from all sampling times into displacement values. Concatenate the three timestamp values ​​with all displacement values ​​in chronological order to form a single high-dimensional numerical vector as a feature vector. S35. Construct an extended isolated forest model. In the feature space, arbitrarily select a feature dimension and a value under the feature dimension as the split point. Read the millisecond value of the current server's system timestamp as the intermediate seed number. Square the intermediate seed number to obtain the square result. Take the middle four digits of the square result as the new seed number. Divide the new seed number by 10000 to obtain a division quotient between 0 and 1. S36. Multiply the division quotient by 2 and subtract 1 to get a value between -1 and 1. Substitute the value into the arctangent function to calculate the slope value. Use the hyperplane that passes through the value and has the slope value to divide the space where the feature vector is located into two subspaces. Recursively execute the operation of arbitrarily selecting dimensions and values ​​and cutting within each subspace to build a binary tree structure until each subspace contains only one feature vector or reaches the preset maximum tree depth. S37. Calculate the total number of edges traversed by the feature vector in the constructed binary tree structure from the root node to the leaf node. Use the total as the path length. Calculate the average of the path lengths of all binary trees. Use an exponential function to map the average to the interval between 0 and 1 to obtain the anomaly score. S38. Read the server date and time when the inspection task is submitted as the current system time. Compare the current system time with the end time set by the preset time window threshold. If the current system time is later than the end time or the value of the abnormal score is greater than the preset threshold, change the status of the initial work order to locked, prohibit the modification or deletion of the work order data, and output the operation and maintenance process data.

5. The method for evaluating operation and maintenance efficiency based on an operation and maintenance quality management database according to claim 1, characterized in that, S4 specifically includes: S41. Write the operation and maintenance process data into the corresponding data record table in the operation and maintenance quality management database. Subtract the order acceptance time from the completion time timestamp to obtain the total duration value. Record the total duration value as the time span. Extract the click operation records generated by the operation and maintenance engineer in the three stages of order acceptance, on-site arrival and completion as operation logs. Extract the user's rating star value after the task is completed as the result score. Combine the time span, operation logs and result score to generate operation and maintenance records. S42. Extract the order acceptance time and arrival time from the operation and maintenance records. Subtract the order acceptance time from the arrival time to obtain the first time difference. Arrange the first time difference of all work orders in order to form a numerical sequence as the response duration vector. S43. Extract the arrival time and completion time from the operation and maintenance records, subtract the arrival time from the completion time to obtain the second time difference, and arrange the second time difference of all work orders in order to form a numerical sequence as a processing time vector. S44. Extract the equipment model field from the asset attribute information in the operation and maintenance records, query the preset equipment difficulty configuration table, read the value corresponding to the equipment model as the equipment difficulty coefficient, normalize the equipment difficulty coefficient, and use the normalized value as the asset type weight. S45. Extract the result rating field from the operation and maintenance records, read the evaluation star rating value, divide the evaluation star rating value by the maximum star rating value to obtain the decimal ratio, and use the ratio as a scalar of user satisfaction. S46. Output the response duration vector, processing duration vector, asset type weight, and user satisfaction scalar in the format of the data columns.

6. The method for evaluating operation and maintenance efficiency based on an operation and maintenance quality management database according to claim 1, characterized in that, S5 specifically includes: S51. An improved game theory combined weighting method is used to input response time vector, processing time vector, asset type weight and user satisfaction scalar. The analytic hierarchy process is used to construct a judgment matrix and compare the four indicators of response time, processing time, asset type weight and user satisfaction in pairs. S52. If indicator i is more important than indicator j, assign the value 2 to the element in the i-th row and j-th column of the judgment matrix; if they are equally important, assign the value 1; if indicator i is less important than indicator j, assign the value 0.

5. Calculate the product of all elements in each column of the judgment matrix, take the fourth root of each column's product to obtain the geometric mean of each column, add the four geometric means to get the sum, divide each geometric mean by the sum, and arrange the four calculated ratios into a column vector in row order to obtain the subjective weight vector. S53. The objective weight vector is calculated using the entropy method. The original values ​​in the response time vector, processing time vector, and user satisfaction scalar are divided by the sum of the values ​​in the corresponding columns to obtain the proportion matrix. The natural logarithm of each value in the proportion matrix is ​​taken and the result is multiplied by the value itself. The negative sum is obtained by summing the values ​​column by column. The entropy value of each indicator is obtained by subtracting the negative sum from 1 and dividing by the natural logarithm of the total number of indicators. The difference coefficient is obtained by subtracting the entropy value from 1. The difference coefficient of each indicator is divided by the sum of all difference coefficients to obtain the objective weight vector. S54. Construct the objective function and introduce an asymmetric game strategy based on abnormal scoring. Calculate the dynamic objective penalty coefficient and the dynamic subjective penalty coefficient, solve to generate the optimal comprehensive weight vector, and calculate and output the KPI quantitative score.

7. The method for evaluating operation and maintenance efficiency based on an operation and maintenance quality management database according to claim 6, characterized in that, Specifically, S54 includes: S541. Construct an objective function. Introduce an asymmetric game strategy based on anomaly scoring into the objective function. Read the anomaly scores in the operation and maintenance process data. If the anomaly score is less than the preset normal threshold, multiply the Euclidean norm of the subjective weight vector by the first initial penalty coefficient and multiply the Euclidean norm of the objective weight vector by the second initial penalty coefficient. Add the two products to construct a symmetric error sum of squares function as the objective function. S542. If the abnormal score is greater than or equal to the preset normal threshold, calculate the difference between the abnormal score and the normal threshold, and use it as an index to be substituted into an exponential function with the natural constant e as the base to calculate the nonlinear growth factor. Multiply the second initial penalty coefficient by the nonlinear growth factor to obtain the dynamic objective penalty coefficient, and subtract the product of the nonlinear growth factor and the first initial penalty coefficient from the first initial penalty coefficient to obtain the dynamic subjective penalty coefficient. S543. Set a combination coefficient vector to be solved consisting of four unknown variables. Calculate the sum of squares of the differences between corresponding elements of the subjective weight vector and the combination coefficient vector to be solved, and multiply the result by the dynamic subjective penalty coefficient. Calculate the sum of squares of the differences between corresponding elements of the objective weight vector and the combination coefficient vector to be solved, and multiply the result by the dynamic objective penalty coefficient. Add the two products to construct an asymmetric error sum of squares function. S544. Take the derivative of the asymmetric error sum of squares matrix equation system and set the derivative to zero. Solve the linear equation system to obtain the combination coefficient vector to be solved as the optimal combination coefficient. Multiply the subjective weight vector with the subjective coefficient in the optimal combination coefficient, multiply the objective weight vector with the objective coefficient in the optimal combination coefficient, add the corresponding elements of the two product vectors, and calculate the optimal comprehensive weight vector. S545. Read the equipment model field in the asset attribute information, query the preset difficulty configuration table to obtain the difficulty correction coefficient corresponding to the equipment model, multiply the first component of the optimal comprehensive weight vector with the element value of the corresponding work order in the response time vector, multiply the second component with the element value of the corresponding work order in the processing time vector, multiply the third component with the asset type weight, and multiply the fourth component with the element value of the corresponding work order in the user satisfaction scalar. Add the four products to obtain the basic score, divide the basic score by the difficulty correction coefficient, and generate the KPI quantitative score of the current work order.

8. The method for evaluating operation and maintenance efficiency based on an operation and maintenance quality management database according to claim 1, characterized in that, S6 specifically includes: S61. Read the system clock to obtain the current date, and calculate the start and end time points backward from the current date according to the preset statistical period. Use the date subtraction function to calculate the date that is shifted forward by one period length from the start time point as the start and end time points of the previous period. S62. Send a query command to the operation and maintenance quality management database, extract the number of all records that have been completed between the start and end time of the current cycle as the total number of work orders processed, and extract the fault type field stored in each record. Iterate through the fault type field and use a counting function to accumulate the frequency of each fault text. S63. Read the total number of work orders processed in the previous period, subtract the total number of work orders processed in the previous period from the total number of work orders processed in the current period to get the numerical difference, divide the numerical difference by the total number of work orders processed in the previous period to get the ratio, and multiply the ratio by 100% to calculate the month-on-month growth rate of the data. S64. Arrange the total number of work orders processed each day in the current period into a time series array in chronological order. Use the least squares method to perform linear fitting on the time series array to calculate the slope value. If the slope value is greater than 0, the development trend is determined to be upward. If the slope value is less than 0, the development trend is determined to be downward. Identify the development trend of operation and maintenance events. S65. Read the KPI quantitative scores corresponding to all work order records in the current period from the operation and maintenance quality management database, add up all KPI quantitative scores and divide by the total number of work order records to obtain the arithmetic mean, write the development trend of operation and maintenance events, the frequency statistics of fault types and the arithmetic mean of KPI quantitative scores into the preset text template, and generate an operation and maintenance quality analysis report. S66. Read the total number of work orders processed, the frequency of fault types, and the quantitative KPI score data from the operation and maintenance quality analysis report. Call the plotting library functions to draw a line chart with date as the horizontal axis and total amount as the vertical axis, a pie chart with fault type as the category, and a bar chart with engineer identification as the horizontal axis and score as the vertical axis. Insert the line chart, pie chart, and bar chart at the bottom of the operation and maintenance quality analysis report and output them.

Images