A performance appraisal method and system based on security operation and maintenance platform

Abstract

The present invention discloses a performance appraisal method and system based on a security operation and maintenance platform, which relates to the technical field of performance appraisal and includes the following steps: S1, determining associated sample data, S2, screening samples to be inspected, S3, recording index data, S4, analyzing a first weight value, S5, calculating an actual weight value, and S6, outputting a total performance appraisal score. The present invention monitors the type and status of real-time data, and determines the number of warning index data and type of abnormal index data of the operation and maintenance personnel according to the response time required for processing warning data and abnormal data. This method can greatly reflect the actual problem-solving ability of each operation and maintenance personnel, and is combined with other indicators to enrich the content of the performance appraisal and increase the rationality of the appraisal. At the same time, corresponding weights are set for each indicator, so that the total score of the performance appraisal can more accurately reflect the various aspects of the operation and maintenance personnel's capabilities.

Description

Technical Field

[0001] The present invention relates to the technical field of performance appraisal, and in particular to a performance appraisal method and system based on a security operation and maintenance platform. Background Art

[0002] In order to ensure the stable operation of the security system and timely respond to various types of faults, the performance appraisal method of the security operation and maintenance platform accurately evaluates the work performance of the operation and maintenance personnel through quantitative indicators and objective evaluation standards, thereby improving the operation and maintenance efficiency and service quality. The invention patent with application number 202311536741.7 discloses "a performance appraisal method and system based on a performance appraisal indicator system, involving the field of performance appraisal. The performance appraisal method based on the performance appraisal indicator system includes the following steps: S1. Obtain performance appraisal indicator system parameters, and obtain appraisal indicator data based on the performance appraisal indicator system parameters; S2. Construct a basic performance appraisal plan based on the appraisal indicator data, and generate primary performance indicators based on the basic performance appraisal plan; S3. Analyze the primary performance indicators to obtain original performance appraisal parameters and original update frequency parameters; S4. Perform primary performance appraisal based on the primary performance indicators, obtain basic performance appraisal parameters, and calculate basic performance appraisal prediction parameters based on the basic performance appraisal parameters. The present invention improves the accuracy and effectiveness of performance appraisal from obtaining performance appraisal indicator system parameters to generating advanced performance appraisal plans."

[0003] The above-mentioned existing technology solves the problem of slow update efficiency due to the inability to adjust the time of performance appraisal. However, during the execution of this method, since various types of sample data in the operation and maintenance platform are not collected, this method cannot construct a decision tree to analyze the real-time data in the platform, nor can it accurately reflect the operation and maintenance personnel's ability to handle abnormal data and warning data, making it impossible for performance appraisal to more comprehensively reflect the work status of the operation and maintenance personnel. Summary of the Invention

[0004] The purpose of the present invention is to provide a performance evaluation method and system based on a security operation and maintenance platform to solve the problems raised in the above background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a performance evaluation method based on a security operation and maintenance platform, comprising the following steps:

[0006] S1. Determine associated sample data: After obtaining different types of sample data, transfer the different types of samples to a low-volume sample set, an equal-volume sample set, and a high-volume sample set according to the number of samples. Divide all samples of the same type in the low-volume sample set into groups of six, select three target samples from each group, calculate the data of the center point of the group based on the target sample data in the same group, use a sample fusion algorithm to analyze the center point data to determine the associated sample data corresponding to each target sample, and store the associated sample data in the low-volume sample set.

[0007] S2. Screening samples to be tested: After selecting a sample as a sample to be tested from the high-volume sample set, count all neighboring samples of the sample to be tested, determine the ratio between the number of neighboring samples with different monitoring methods and the total number of neighboring samples, and if the ratio is greater than a threshold, delete the sample data to be tested. Repeat the operation until the number of samples of all types in the high-volume sample set is equal to the preset value, extract sample data from the high-volume sample set and the low-volume sample set and transfer them to the equal-volume sample set. After determining the number of decision trees, randomly sample the data from the equal-volume sample set to construct a training set and a test set. Use the data in the training set to complete the decision tree training, input the data in the test set into all decision trees, use the averaging method to analyze the type and status of the current sample data, and adjust the parameters of the decision tree according to the deviation between the actual sample result and the predicted result;

[0008] S3. Record indicator data: After obtaining real-time data and the corresponding recording time, transmit it to each decision tree model for analysis to obtain the status corresponding to each type of data. If the status of a data type is warning, set a time interval. If no confirmation information is returned within the time interval, the original warning number indicator data is automatically increased by one. If the status of a data type is abnormal, set a time interval according to the abnormal data type. If the abnormal state is not converted to normal state within the time interval, the timer corresponding to the current abnormal data starts timing until the state is converted to normal state. Output the abnormal processing time of this type, and add the original type abnormality indicator data to the current abnormal processing time to obtain the new type abnormality indicator data;

[0009] S4. Analyze the first weight value: Determine all performance evaluation indicators for operation and maintenance personnel, use an expert panel to rate the degree of correlation between each indicator and other indicators, and after obtaining the correlation level, use the influence coefficient generation algorithm to calculate the average influence coefficient between each indicator and other indicators. Based on the average influence coefficient, construct a comprehensive influence matrix, and use the comprehensive influence matrix to deduce the first weight value corresponding to each indicator;

[0010] S5. Calculate the actual weight value: Determine the initial data corresponding to all performance evaluation indicators, use the initial data to construct a standard numerical matrix, calculate the correlation coefficients between different indicators according to the values ​​in the standard numerical matrix, use the correlation coefficients to analyze the value parameters corresponding to each indicator, determine the second weight value of the indicator based on the value parameters, and use the weight analysis algorithm to combine the first weight value and the second weight value of each indicator to obtain the actual weight value of each indicator;

[0011] S6. Output the total performance appraisal score: Use deep learning technology to convert the indicator data of each operation and maintenance personnel into a corresponding score, calculate the total performance appraisal score of each operation and maintenance personnel based on the indicator score and the actual weight value, and output it through a visual interface.

[0012] Preferably, the step S1 specifically includes the following steps:

[0013] S101. After obtaining sample data of different types, count the number of samples corresponding to the same type. If the number of samples is lower than a preset value, transfer the samples corresponding to the current type to a low-volume sample set. If the number of samples is equal to the preset value, transfer the samples corresponding to the current type to an equal-volume sample set. If the number of samples is higher than the preset value, transfer the samples corresponding to the current type to a high-volume sample set. The types include video monitoring, environmental monitoring, infrared monitoring, communication monitoring, and personnel information monitoring.

[0014] S102, dividing all samples of the same type in the low-volume sample set into groups of six. If the number of remaining samples after grouping is less than three, the remaining samples are added to any group; otherwise, the remaining samples are combined into one group.

[0015] S103. Randomly select three samples from each group. As the target sample, determine the target sample The corresponding value of each monitoring method included in is Calculate the center point of the group based on the corresponding values ​​of different monitoring methods contained in the three target samples in the same group The value of ;

[0016] S104: Combine each target sample data in the same group with the center point data, analyze the combined data using a sample fusion algorithm, obtain associated sample data corresponding to each target sample data, store the associated sample data in a low-volume sample set, and repeat the operation until the number of samples of all types in the low-volume sample set is equal to a preset value. The sample fusion algorithm is specifically as follows:

[0017] ;

[0018] in, Represents associated sample data, Represents the center point data, represents the target sample data, Represents a random function.

[0019] Preferably, the step S2 specifically includes the following steps:

[0020] S201: After selecting an unlabeled sample from a large sample set as a sample to be tested, the similarity between the sample to be tested and other samples of the same type in the sample set is calculated. If the similarity is higher than a preset value, the sample is determined to be a neighboring sample of the sample to be tested;

[0021] S202. Compare the monitoring method contained in the sample to be tested with the monitoring methods of all adjacent samples, count the number of samples with different monitoring methods from the sample to be tested, and calculate the ratio between the number of samples with different monitoring methods and the total number of adjacent samples. If the ratio is greater than a threshold, it is determined that the current sample to be tested has redundant data, and the data of the sample to be tested is deleted. Otherwise, it is determined that the current sample to be tested does not have redundant data, and the sample to be tested is marked. Repeat the operation until the number of samples of all types in the high-volume sample set is equal to the preset value.

[0022] Preferably, the step S2 further includes the following steps:

[0023] S203: After extracting different types of sample data from the high-volume sample set and the low-volume sample set, transfer them to an equal-volume sample set, add type labels and status labels to all samples in the sample set, determine the number of decision trees, randomly extract 80% of the data from the equal-volume sample set, store the extracted data in different training sets, and extract 20% of the data from the equal-volume sample set in a test set. The status includes normal, warning, and abnormal.

[0024] S204. After setting a corresponding training set for each decision tree model, complete the decision tree training using the data in the training set, input the data in the test set into all decision trees, use the averaging method to analyze the type and status of the current sample data through the prediction results of all decision trees, and adjust the parameters of the decision tree according to the deviation value between the sample label information and the prediction result.

[0025] Preferably, the step S3 specifically includes the following steps:

[0026] S301, after obtaining real-time data and the corresponding recording time, transfer it to each decision tree model for analysis, use the averaging method to obtain the state corresponding to each type of data based on the prediction results of all decision tree models, and after determining the result output time, transfer the recording time, result output time, data corresponding to each type, and state to the database for storage. The real-time data includes video monitoring data, environmental monitoring data, infrared monitoring data, communication monitoring data, and personnel information monitoring data;

[0027] S302: If the status of a data type is warning, the recording time, result output time, data corresponding to the type, and status are output through a visual interface, and a time interval is set. If no confirmation information is returned within the time interval, the corresponding operation and maintenance personnel indicator data is queried in the database based on the result output time, the original warning number indicator data is incremented by one, and the new warning number indicator data is obtained and transmitted to the database;

[0028] S303. If the status of a data type is abnormal, the recording time, result output time, data and status corresponding to the type will be output through the visual interface, and the corresponding time interval will be set according to the abnormal data type. If the abnormal state is not converted to the normal state within the time interval, the corresponding timer will be set for the current abnormal data until the abnormal state is converted to the normal state and the timing ends, thereby obtaining the abnormal processing time of this type. According to the result output time, the indicator data of the corresponding operation and maintenance personnel is queried in the database, and the original corresponding type abnormal indicator data is added to the current abnormal processing time to obtain the new type abnormal indicator data, which is transmitted to the database.

[0029] Preferably, the step S4 specifically includes the following steps:

[0030] S401. Determine all performance evaluation indicators for operation and maintenance personnel, and use an expert panel to rate the degree of correlation between each indicator and other indicators. For indicators and indicators The set association level Afterwards, Convert to triangular fuzzy number ,in is a triangular fuzzy number The parameter values ​​are analyzed to obtain three standard coefficients ,in The performance evaluation indicators include warning number indicators, video anomaly indicators, environmental anomaly indicators, infrared anomaly indicators, communication anomaly indicators and personnel anomaly indicators, task completion rate indicators, normal operation indicators, customer satisfaction indicators, cooperation number indicators, communication ability indicators, knowledge score indicators and skill operation indicators;

[0031] S402: After determining the standard coefficients corresponding to all indicators, analyze the standard coefficients of each indicator using an influence coefficient generation algorithm to determine the average influence coefficient between each indicator and other indicators, and construct a comprehensive influence matrix based on the average influence coefficient. The influence coefficient generation algorithm is specifically as follows:

[0032] ;

[0033] ;

[0034] in, Indicates expert Generate indicators and indicators The corresponding influence coefficient between Indicates expert Generate indicators and indicators The first standard coefficient corresponding to Indicates expert Generate indicators and indicators The corresponding second standard coefficient between Indicates expert Generate indicators and indicators The corresponding third standard coefficient between Indicates expert Generate indicators and indicators The corresponding influence coefficient between Indicates expert Generate indicators and indicators The corresponding influence coefficient between Indicator and indicators The average influence coefficient between Represents parameters;

[0035] S403. Add up all the data in the same row of the comprehensive influence coefficient matrix to obtain the sum of the influence corresponding to each row, then add up all the data in the same column of the comprehensive influence coefficient matrix to obtain the sum of the influence corresponding to each column, use the sum of the influence of each row and each column to calculate the degree value corresponding to each indicator, and determine the corresponding first weight value according to the degree value of each indicator.

[0036] Preferably, the step S5 specifically includes the following steps:

[0037] S501. Determine the initial data corresponding to all performance evaluation indicators, standardize the initial data, use the processed data to construct a standard value matrix, and determine the mean value and standard deviation corresponding to each indicator data according to each value in the standard value matrix;

[0038] S502. Calculate the correlation coefficients between the different indicators using the standard values, average values, and standard deviations corresponding to the various indicators. Analyze the value parameters corresponding to each indicator based on the correlation coefficients. Accumulate the value parameters of all indicators to obtain a total value. Calculate the ratio between the value parameter of each indicator and the total value. Determine the second weight value of each indicator based on the ratio.

[0039] S503 : The first weight value and the second weight value of each indicator are integrated through a weight analysis algorithm to obtain an actual weight value of the indicator.

[0040] The performance appraisal system based on the security operation and maintenance platform includes a sample data processing unit, an indicator data determination unit, a weight analysis unit and a score output unit;

[0041] After the sample data processing unit obtains different types of sample data, the different types of samples are transferred to a low-volume sample set, an equal-volume sample set, and a high-volume sample set according to the number of samples, and the associated sample data in the low-volume sample set is analyzed using a sample fusion algorithm. After screening the samples to be tested in the high-volume sample set, the sample data in the high-volume sample set and the low-volume sample set are transferred to the equal-volume sample set. After determining the number of decision trees, the data in the equal-volume sample set are randomly sampled to construct a training set and a test set. The decision tree training is completed using the data in the training set, and the data in the test set is input into all decision trees. The parameters of the decision tree are adjusted according to the deviation value between the actual sample result and the predicted result;

[0042] After the indicator data determination unit obtains the real-time data and the corresponding recording time, it transmits it to each decision tree model for analysis to obtain the status corresponding to each type of data. If the status of the data type is a warning, a time interval is set. If no confirmation information is returned within the time interval, the original warning number indicator data is automatically increased by one. If the status of the data type is abnormal, a time interval is set according to the abnormal data type. If the abnormal state is not converted to a normal state within the time interval, the timer corresponding to the current abnormal data starts timing until the state is converted to a normal state. The abnormal processing time of this type is output, and the original type abnormality indicator data is added to the current abnormal processing time to obtain the new type abnormality indicator data;

[0043] The weight analysis unit determines all performance appraisal indicators of the operation and maintenance personnel, uses an expert group to rate the degree of correlation between each indicator and other indicators, and after obtaining the correlation level, uses an influence coefficient generation algorithm to calculate the average influence coefficient between each indicator and other indicators, and infers the first weight value corresponding to each indicator based on the average influence coefficient, determines the initial data corresponding to all performance appraisal indicators, constructs a standard numerical matrix using the initial data, calculates the correlation coefficients between different indicators according to each numerical value in the standard numerical matrix, analyzes the value parameter corresponding to each indicator using the correlation coefficient, determines the second weight value of the indicator through the value parameter, and uses the weight analysis algorithm to fuse the first weight value and the second weight value of each indicator to obtain the actual weight value of each indicator;

[0044] The score output unit uses deep learning technology to convert the indicator data of each operation and maintenance personnel into a corresponding score, calculates the total performance appraisal score of each operation and maintenance personnel based on the indicator score and the actual weight value, and outputs it through a visual interface.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. The present invention collects video monitoring samples, environmental monitoring samples, infrared monitoring samples, communication monitoring samples, and personnel information monitoring samples. During the collection process, in order to prevent the number of different types of samples from being too different, associated sample data is added for types with too few samples, and some data is deleted for types with too many samples. This design can, on the one hand, ensure that the added sample data is associated with the collected sample data, and on the other hand, delete sample data containing noise to prevent interference with subsequent classification results. At the same time, this method uses different types of sample data to train a decision tree, so that the trained decision tree model can accurately analyze the types and states corresponding to the video monitoring data, environmental monitoring data, infrared monitoring data, communication monitoring data, and personnel information monitoring data;

[0047] 2. The present invention monitors the type and status of real-time data, and determines the warning number index data and type abnormality index data of the operation and maintenance personnel according to the response time required to process warning data and abnormal data. This method can reflect the actual problem-solving ability of each operation and maintenance personnel to a great extent. Combined with other indicators, it enriches the content of performance appraisal and increases the rationality of the appraisal. At the same time, a corresponding weight is set for each indicator, so that the total score of the performance appraisal can more accurately reflect the various aspects of the operation and maintenance personnel's capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 Provides an overall method flow chart for an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0050] See also Figure 1 The present invention provides a technical solution: a performance evaluation method based on a security operation and maintenance platform, comprising the following steps:

[0051] S1. Determine associated sample data: After obtaining different types of sample data, transfer the different types of samples to a low-volume sample set, an equal-volume sample set, and a high-volume sample set according to the number of samples. Divide all samples of the same type in the low-volume sample set into groups of six, select three target samples from each group, calculate the data of the center point of the group based on the target sample data in the same group, use a sample fusion algorithm to analyze the center point data to determine the associated sample data corresponding to each target sample, and store the associated sample data in the low-volume sample set.

[0052] S2. Screening samples to be tested: After selecting a sample as a sample to be tested from the high-volume sample set, count all neighboring samples of the sample to be tested, determine the ratio between the number of neighboring samples with different monitoring methods and the total number of neighboring samples, and if the ratio is greater than a threshold, delete the sample data to be tested. Repeat the operation until the number of samples of all types in the high-volume sample set is equal to the preset value, extract sample data from the high-volume sample set and the low-volume sample set and transfer them to the equal-volume sample set. After determining the number of decision trees, randomly sample the data from the equal-volume sample set to construct a training set and a test set. Use the data in the training set to complete the decision tree training, input the data in the test set into all decision trees, use the averaging method to analyze the type and status of the current sample data, and adjust the parameters of the decision tree according to the deviation between the actual sample result and the predicted result;

[0053] S3. Record indicator data: After obtaining real-time data and the corresponding recording time, transmit it to each decision tree model for analysis to obtain the status corresponding to each type of data. If the status of a data type is warning, set a time interval. If no confirmation information is returned within the time interval, the original warning number indicator data is automatically increased by one. If the status of a data type is abnormal, set a time interval according to the abnormal data type. If the abnormal state is not converted to normal state within the time interval, the timer corresponding to the current abnormal data starts timing until the state is converted to normal state. Output the abnormal processing time of this type, and add the original type abnormality indicator data to the current abnormal processing time to obtain the new type abnormality indicator data;

[0054] S4. Analyze the first weight value: Determine all performance evaluation indicators for operation and maintenance personnel, use an expert panel to rate the degree of correlation between each indicator and other indicators, and after obtaining the correlation level, use the influence coefficient generation algorithm to calculate the average influence coefficient between each indicator and other indicators. Based on the average influence coefficient, construct a comprehensive influence matrix, and use the comprehensive influence matrix to deduce the first weight value corresponding to each indicator;

[0055] S5. Calculate the actual weight value: Determine the initial data corresponding to all performance evaluation indicators, use the initial data to construct a standard numerical matrix, calculate the correlation coefficients between different indicators according to the values ​​in the standard numerical matrix, use the correlation coefficients to analyze the value parameters corresponding to each indicator, determine the second weight value of the indicator based on the value parameters, and use the weight analysis algorithm to combine the first weight value and the second weight value of each indicator to obtain the actual weight value of each indicator;

[0056] S6. Output the total performance appraisal score: Use deep learning technology to convert the indicator data of each operation and maintenance personnel into a corresponding score, calculate the total performance appraisal score of each operation and maintenance personnel based on the indicator score and the actual weight value, and output it through a visual interface.

[0057] Step S1 specifically includes the following steps:

[0058] S101. After obtaining sample data of different types, count the number of samples corresponding to the same type. If the number of samples is lower than a preset value, transfer the samples corresponding to the current type to a low-volume sample set. If the number of samples is equal to the preset value, transfer the samples corresponding to the current type to an equal-volume sample set. If the number of samples is higher than the preset value, transfer the samples corresponding to the current type to a high-volume sample set. The types include video monitoring, environmental monitoring, infrared monitoring, communication monitoring, and personnel information monitoring.

[0059] S102, dividing all samples of the same type in the low-volume sample set into groups of six. If the number of remaining samples after grouping is less than three, the remaining samples are added to any group; otherwise, the remaining samples are combined into one group.

[0060] S103. Randomly select three samples from each group. As the target sample, determine the target sample The corresponding value of each monitoring method included in is Calculate the center point of the group based on the corresponding values ​​of different monitoring methods contained in the three target samples in the same group The value of ;

[0061] S104: Combine each target sample data in the same group with the center point data, analyze the combined data using a sample fusion algorithm, obtain associated sample data corresponding to each target sample data, store the associated sample data in a low-volume sample set, and repeat the operation until the number of samples of all types in the low-volume sample set is equal to a preset value. The sample fusion algorithm is specifically as follows:

[0062] ;

[0063] in, Represents associated sample data, Represents the center point data, represents the target sample data, represents a random function;

[0064] Step S2 specifically includes the following steps:

[0065] S201: After selecting an unlabeled sample from a large sample set as a sample to be tested, the similarity between the sample to be tested and other samples of the same type in the sample set is calculated. If the similarity is higher than a preset value, the sample is determined to be a neighboring sample of the sample to be tested;

[0066] S202: Compare the monitoring mode included in the sample to be tested with the monitoring modes of all neighboring samples, count the number of samples with different monitoring modes from the sample to be tested, and calculate the ratio between the number of samples with different monitoring modes and the total number of neighboring samples. If the ratio is greater than a threshold, it is determined that the current sample to be tested has redundant data and the data of the sample to be tested is deleted. Otherwise, it is determined that the current sample to be tested does not have redundant data and the sample to be tested is marked. Repeat the operation until the number of samples of all types in the high-volume sample set is equal to a preset value.

[0067] Step S2 specifically further includes the following steps:

[0068] S203: After extracting different types of sample data from the high-volume sample set and the low-volume sample set, transfer them to the equal-volume sample set, add type labels and status labels to all samples in the sample set, determine the number of decision trees, randomly extract 80% of the data from the equal-volume sample set, store the extracted data in different training sets, and extract 20% of the data from the equal-volume sample set in the test set, with the status including normal, warning, and abnormal.

[0069] S204: After setting a corresponding training set for each decision tree model, complete the decision tree training using the data in the training set, input the data in the test set into all decision trees, use the averaging method to analyze the type and status of the current sample data through the prediction results of all decision trees, and adjust the parameters of the decision tree according to the deviation between the sample label information and the prediction result;

[0070] Step S3 specifically includes the following steps:

[0071] S301. After acquiring real-time data and the corresponding recording time, transfer them to each decision tree model for analysis. Use the averaging method to obtain the state corresponding to each type of data based on the prediction results of all decision tree models. After determining the result output time, transfer the recording time, result output time, data corresponding to each type, and state to the database for storage. The real-time data includes video monitoring data, environmental monitoring data, infrared monitoring data, communication monitoring data, and personnel information monitoring data.

[0072] S302: If the status of a data type is warning, the recording time, result output time, data corresponding to the type, and status are output through a visual interface, and a time interval is set. If no confirmation information is returned within the time interval, the corresponding operation and maintenance personnel indicator data is queried in the database based on the result output time, the original warning number indicator data is incremented by one, and the new warning number indicator data is obtained and transmitted to the database;

[0073] S303. If the state of a data type is abnormal, the recording time, result output time, data corresponding to the type, and state are output through a visual interface, and a corresponding time interval is set according to the abnormal data type. If the abnormal state is not converted to a normal state within the time interval, a corresponding timer is set for the current abnormal data until the abnormal state is converted to a normal state, thereby obtaining the abnormal processing time of the type. According to the result output time, the corresponding operation and maintenance personnel indicator data is queried in the database, and the original corresponding type abnormal indicator data is added to the current abnormal processing time to obtain the new type abnormal indicator data, which is then transmitted to the database.

[0074] Step S4 specifically includes the following steps:

[0075] S401. Determine all performance evaluation indicators for operation and maintenance personnel, and use an expert panel to rate the degree of correlation between each indicator and other indicators. For indicators and indicators The set association level Afterwards, Convert to triangular fuzzy number ,in is a triangular fuzzy number The parameter values ​​are analyzed to obtain three standard coefficients ,in ,The performance evaluation indicators include the number of warnings indicators, video type anomaly indicators, environment type anomaly indicators, infrared type anomaly indicators, communication type anomaly indicators, and personnel type anomaly indicators, task completion rate indicators, normal operation indicators, customer satisfaction indicators, cooperation number indicators, communication ability indicators, knowledge score indicators, and skill operation indicators;

[0076] S402: After determining the standard coefficients corresponding to all indicators, analyze the standard coefficients of each indicator using an influence coefficient generation algorithm to determine the average influence coefficient between each indicator and other indicators. A comprehensive influence matrix is ​​constructed based on the average influence coefficients. The influence coefficient generation algorithm is specifically as follows:

[0077] ;

[0078] ;

[0079] in, Indicates expert Generate indicators and indicators The corresponding influence coefficient between Indicates expert Generate indicators and indicators The first standard coefficient corresponding to Indicates expert Generate indicators and indicators The corresponding second standard coefficient between Indicates expert Generate indicators and indicators The corresponding third standard coefficient between Indicates expert Generate indicators and indicators The corresponding influence coefficient between Indicates expert Generate indicators and indicators The corresponding influence coefficient between Indicator and indicators The average influence coefficient between Represents parameters;

[0080] S403: After accumulating all data in the same row of the comprehensive influence coefficient matrix to obtain the sum of influences corresponding to each row, all data in the same column of the comprehensive influence coefficient matrix are accumulated to obtain the sum of influences corresponding to each column. The degree value corresponding to each indicator is calculated using the sum of influences of each row and each column, and the corresponding first weight value is determined according to the degree value of each indicator.

[0081] Step S5 specifically includes the following steps:

[0082] S501. Determine the initial data corresponding to all performance evaluation indicators, standardize the initial data, use the processed data to construct a standard value matrix, and determine the mean value and standard deviation corresponding to each indicator data according to each value in the standard value matrix;

[0083] S502. Calculate the correlation coefficients between the different indicators using the standard values, average values, and standard deviations corresponding to the various indicators. Analyze the value parameters corresponding to each indicator based on the correlation coefficients. Accumulate the value parameters of all indicators to obtain a total value. Calculate the ratio between the value parameter of each indicator and the total value. Determine the second weight value of each indicator based on the ratio.

[0084] S503: The first weight value and the second weight value of each indicator are integrated by a weight analysis algorithm to obtain the actual weight value of the indicator. The weight analysis algorithm is specifically as follows:

[0085] ;

[0086] in, Indicates the The actual weight of the indicator, Indicates the The first weight value of the indicator, Indicates the The second weight value of the indicator, Indicates the total number of indicators, Represents parameters;

[0087] A performance appraisal system based on a security operation and maintenance platform includes a sample data processing unit 1, an indicator data determination unit 2, a weight analysis unit 3, and a score output unit 4;

[0088] After the sample data processing unit 1 obtains different types of sample data, it transfers the different types of samples to a low-volume sample set, an equal-volume sample set, and a high-volume sample set according to the number of samples, uses a sample fusion algorithm to analyze the associated sample data in the low-volume sample set, and after screening the samples to be tested in the high-volume sample set, transfers the sample data in the high-volume sample set and the low-volume sample set to the equal-volume sample set. After determining the number of decision trees, it randomly extracts data from the equal-volume sample set to construct a training set and a test set. The decision tree training is completed using the data in the training set, and the data in the test set is input into all decision trees. The parameters of the decision tree are adjusted according to the deviation value between the actual sample result and the predicted result;

[0089] After the indicator data determination unit 2 obtains the real-time data and the corresponding recording time, it transmits it to each decision tree model for analysis to obtain the status corresponding to each type of data. If the status of a data type is a warning, a time interval is set. If no confirmation information is returned within the time interval, the original warning number indicator data is automatically increased by one. If the status of a data type is abnormal, a time interval is set according to the abnormal data type. If the abnormal state is not converted to a normal state within the time interval, the timer corresponding to the current abnormal data starts timing until the state is converted to a normal state. The abnormal processing time of this type is output, and the original type abnormality indicator data is added to the current abnormal processing time to obtain the new type abnormality indicator data;

[0090] The weight analysis unit 3 determines all performance appraisal indicators of the operation and maintenance personnel, uses an expert group to rate the degree of correlation between each indicator and other indicators, and after obtaining the correlation level, uses an influence coefficient generation algorithm to calculate the average influence coefficient between each indicator and other indicators, and infers the first weight value corresponding to each indicator based on the average influence coefficient, determines the initial data corresponding to all performance appraisal indicators, constructs a standard numerical matrix using the initial data, calculates the correlation coefficients between different indicators according to each value in the standard numerical matrix, analyzes the value parameter corresponding to each indicator using the correlation coefficient, determines the second weight value of the indicator based on the value parameter, and uses the weight analysis algorithm to fuse the first weight value and the second weight value of each indicator to obtain the actual weight value of each indicator;

[0091] The score output unit 4 uses deep learning technology to convert the indicator data of each operation and maintenance personnel into corresponding scores, calculates the total performance appraisal score of each operation and maintenance personnel based on the indicator score and the actual weight value, and outputs it through a visual interface.

[0092] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0093] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A performance appraisal method based on a security operation and maintenance platform, characterized in that: The method comprises the following steps: S1. Determine associated sample data: After obtaining different types of sample data, transfer the different types of samples to a low-volume sample set, an equal-volume sample set, and a high-volume sample set according to the number of samples. Divide all samples of the same type in the low-volume sample set into groups of six, select three target samples from each group, calculate the data of the center point of the group based on the target sample data in the same group, use a sample fusion algorithm to analyze the center point data to determine the associated sample data corresponding to each target sample, and store the associated sample data in the low-volume sample set. S2. Screening samples to be tested: After selecting a sample as a sample to be tested from the high-volume sample set, count all neighboring samples of the sample to be tested, determine the ratio between the number of neighboring samples with different monitoring methods and the total number of neighboring samples, and if the ratio is greater than a threshold, delete the sample data to be tested. Repeat the operation until the number of samples of all types in the high-volume sample set is equal to the preset value, extract sample data from the high-volume sample set and the low-volume sample set and transfer them to the equal-volume sample set. After determining the number of decision trees, randomly sample the data from the equal-volume sample set to construct a training set and a test set. Use the data in the training set to complete the decision tree training, input the data in the test set into all decision trees, use the averaging method to analyze the type and status of the current sample data, and adjust the parameters of the decision tree according to the deviation between the actual sample result and the predicted result; S3. Record indicator data: After obtaining real-time data and the corresponding recording time, transfer it to each decision tree model for analysis. Use the averaging method to obtain the status corresponding to each type of data based on the prediction results of all decision tree models. After determining the result output time, transfer the recording time, result output time, data and status corresponding to each type to the database for storage. Real-time data includes video monitoring data, environmental monitoring data, infrared monitoring data, communication monitoring data and personnel information monitoring data. If there is a data type with a warning status, the recording time, result output time, data and status corresponding to the type will be output through the visual interface, and a time interval will be set. If no confirmation information is returned within the time interval, the corresponding operation and maintenance person will be queried in the database based on the result output time. The indicator data of the operator is obtained by adding one to the original warning number indicator data, and the new warning number indicator data is transmitted to the database. If the status of the data type is abnormal, the recording time, result output time, data and status corresponding to the type are output through the visual interface, and the corresponding time interval is set according to the abnormal data type. If the abnormal state is not converted to the normal state within the time interval, the corresponding timer is set for the current abnormal data until the abnormal state is converted to the normal state and the timing ends, thereby obtaining the abnormal processing time of the type. The indicator data of the corresponding operation and maintenance personnel is queried in the database according to the result output time, and the original corresponding type abnormal indicator data is added to the current abnormal processing time to obtain the new type abnormal indicator data, which is transmitted to the database; S4. Analyze the first weight value: determine all performance evaluation indicators of operation and maintenance personnel, use the expert group to rate the degree of correlation between each indicator and other indicators, and determine the expert group For indicators and indicators The set association level Afterwards, Convert to triangular fuzzy number ,in , is a triangular fuzzy number The parameter values ​​are analyzed to obtain three standard coefficients ,in The performance evaluation indicators include warning number indicator, video type abnormality indicator, environment type abnormality indicator, infrared type abnormality indicator, communication type abnormality indicator and personnel type abnormality indicator, task completion rate indicator, normal operation indicator, customer satisfaction indicator, cooperation number indicator, communication ability indicator, knowledge score indicator and skill operation indicator. After determining the standard coefficients corresponding to all indicators, the influence coefficient generation algorithm is used to analyze the standard coefficients of each indicator to determine the average influence coefficient between each indicator and other indicators. A comprehensive influence matrix is ​​constructed based on the average influence coefficient. All data in the same row of the comprehensive influence coefficient matrix are accumulated to obtain the sum of the influence corresponding to each row. Then, all data in the same column of the comprehensive influence coefficient matrix are accumulated to obtain the sum of the influence corresponding to each column. The degree value corresponding to each indicator is calculated using the sum of the influence of each row and each column. The corresponding first weight value is determined according to the degree value of each indicator. The influence coefficient generation algorithm is specifically as follows: ； ； in, Indicates expert Generate indicators and indicators The corresponding influence coefficient between Indicates expert Generate indicators and indicators The first standard coefficient corresponding to Indicates expert Generate indicators and indicators The corresponding second standard coefficient between Indicates expert Generate indicators and indicators The corresponding third standard coefficient between Indicates expert Generate indicators and indicators The corresponding influence coefficient between Indicates expert Generate indicators and indicators The corresponding influence coefficient between Indicator and indicators The average influence coefficient between Represents parameters; S5. Calculate the actual weight value: Determine the initial data corresponding to all performance evaluation indicators, use the initial data to construct a standard numerical matrix, calculate the correlation coefficients between different indicators according to the values ​​in the standard numerical matrix, use the correlation coefficients to analyze the value parameters corresponding to each indicator, determine the second weight value of the indicator based on the value parameters, and use the weight analysis algorithm to combine the first weight value and the second weight value of each indicator to obtain the actual weight value of each indicator; S6. Output the total performance appraisal score: Use deep learning technology to convert the indicator data of each operation and maintenance personnel into a corresponding score, calculate the total performance appraisal score of each operation and maintenance personnel based on the indicator score and the actual weight value, and output it through a visual interface.

2. The performance evaluation method based on the security operation and maintenance platform according to claim 1 is characterized by: The step S1 specifically includes the following steps: S101. After obtaining sample data of different types, count the number of samples corresponding to the same type. If the number of samples is lower than a preset value, transfer the samples corresponding to the current type to a low-volume sample set. If the number of samples is equal to the preset value, transfer the samples corresponding to the current type to an equal-volume sample set. If the number of samples is higher than the preset value, transfer the samples corresponding to the current type to a high-volume sample set. The types include video monitoring, environmental monitoring, infrared monitoring, communication monitoring, and personnel information monitoring. S102, dividing all samples of the same type in the low-volume sample set into groups of six. If the number of remaining samples after grouping is less than three, the remaining samples are added to any group; otherwise, the remaining samples are combined into one group. S103. Randomly select three samples from each group. As the target sample, determine the target sample The corresponding value of each monitoring method included in is Calculate the center point of the group based on the corresponding values ​​of different monitoring methods contained in the three target samples in the same group The value of ; S104. Combine each target sample data in the same group with the center point data, analyze the combined data using a sample fusion algorithm to obtain associated sample data corresponding to each target sample data, store the associated sample data in a low-volume sample set, and repeat the operation until the number of samples of all types in the low-volume sample set is equal to a preset value.

3. The performance evaluation method based on the security operation and maintenance platform according to claim 1 is characterized in that: The step S2 specifically includes the following steps: S201: After selecting an unlabeled sample from a large sample set as a sample to be tested, the similarity between the sample to be tested and other samples of the same type in the sample set is calculated. If the similarity is higher than a preset value, the sample is determined to be a neighboring sample of the sample to be tested; S202. Compare the monitoring method contained in the sample to be tested with the monitoring methods of all adjacent samples, count the number of samples with different monitoring methods from the sample to be tested, and calculate the ratio between the number of samples with different monitoring methods and the total number of adjacent samples. If the ratio is greater than a threshold, it is determined that the current sample to be tested has redundant data, and the data of the sample to be tested is deleted. Otherwise, it is determined that the current sample to be tested does not have redundant data, and the sample to be tested is marked. Repeat the operation until the number of samples of all types in the high-volume sample set is equal to the preset value.

4. The performance evaluation method based on the security operation and maintenance platform according to claim 3 is characterized by: The step S2 specifically further includes the following steps: S203: After extracting different types of sample data from the high-volume sample set and the low-volume sample set, transfer them to the equal-volume sample set, add type labels and status labels to all samples in the sample set, determine the number of decision trees, randomly extract 80% of the data from the equal-volume sample set, store the extracted data in different training sets, and store 20% of the data from the equal-volume sample set in the test set; S204. After setting a corresponding training set for each decision tree model, complete the decision tree training using the data in the training set, input the data in the test set into all decision trees, use the averaging method to analyze the type and status of the current sample data through the prediction results of all decision trees, and adjust the parameters of the decision tree according to the deviation value between the sample label information and the prediction result.

5. The performance evaluation method based on the security operation and maintenance platform according to claim 1 is characterized in that: The step S5 specifically includes the following steps: S501. Determine the initial data corresponding to all performance evaluation indicators, standardize the initial data, use the processed data to construct a standard value matrix, and determine the mean value and standard deviation corresponding to each indicator data according to each value in the standard value matrix; S502. Calculate the correlation coefficients between the different indicators using the standard values, average values, and standard deviations corresponding to the various indicators. Analyze the value parameters corresponding to each indicator based on the correlation coefficients. Accumulate the value parameters of all indicators to obtain a total value. Calculate the ratio between the value parameter of each indicator and the total value. Determine the second weight value of each indicator based on the ratio. S503 : The first weight value and the second weight value of each indicator are integrated through a weight analysis algorithm to obtain an actual weight value of the indicator.

6. A performance appraisal system based on a security operation and maintenance platform, characterized in that: The performance appraisal system is applicable to a performance appraisal method based on a security operation and maintenance platform as described in any one of claims 1 to 5, comprising a sample data processing unit (1), an indicator data determination unit (2), a weight analysis unit (3) and a score output unit (4); The sample data processing unit (1) obtains different types of sample data, transfers the different types of samples to a low-volume sample set, an equal-volume sample set, and a high-volume sample set according to the number of samples, analyzes the associated sample data in the low-volume sample set using a sample fusion algorithm, screens the samples to be tested in the high-volume sample set, transfers the sample data in the high-volume sample set and the low-volume sample set to the equal-volume sample set, determines the number of decision trees, randomly extracts data from the equal-volume sample set, thereby constructing a training set and a test set, completes decision tree training using the data in the training set, inputs the data in the test set into all decision trees, and adjusts the parameters of the decision tree according to the deviation value between the actual sample result and the predicted result; The indicator data determination unit (2) obtains the real-time data and the corresponding recording time, and transmits them to each decision tree model for analysis to obtain the status corresponding to each type of data. If the status of the data type is a warning, a time interval is set. If no confirmation information is returned within the time interval, the original warning number indicator data is automatically increased by one. If the status of the data type is abnormal, a time interval is set according to the abnormal data type. If the abnormal state is not converted to a normal state within the time interval, the timer corresponding to the current abnormal data starts timing until the state is converted to a normal state, and the abnormal processing time of the type is output. The original type abnormal indicator data is added to the current abnormal processing time to obtain the new type abnormal indicator data; The weight analysis unit (3) determines all performance appraisal indicators of the operation and maintenance personnel, uses an expert group to rate the degree of correlation between each indicator and other indicators, and after obtaining the correlation level, uses an influence coefficient generation algorithm to calculate the average influence coefficient between each indicator and other indicators, and infers the first weight value corresponding to each indicator based on the average influence coefficient, determines the initial data corresponding to all performance appraisal indicators, constructs a standard numerical matrix based on the initial data, calculates the correlation coefficient between different indicators according to each value in the standard numerical matrix, analyzes the value parameter corresponding to each indicator using the correlation coefficient, determines the second weight value of the indicator based on the value parameter, and uses the weight analysis algorithm to fuse the first weight value and the second weight value of each indicator to obtain the actual weight value of each indicator; The score output unit (4) converts the indicator data of each operation and maintenance personnel into a corresponding score using deep learning technology, calculates the total performance appraisal score of each operation and maintenance personnel based on the indicator score and the actual weight value, and outputs it through a visual interface.

Images