Electric power operator safety literacy evaluation method and device and storage medium

Abstract

The invention discloses an electric power worker safety literacy evaluation method and device and a storage medium, and the method comprises the steps: employing a machine learning algorithm as a prediction model for training based on the safety literacy evaluation index data of an electric power worker in a historical time period, and obtaining an optimal model; and obtaining safety literacy evaluation index data of the electric power operation personnel in a first time period, inputting the safety literacy evaluation index data in the first time period into the optimal model, and automatically generating a safety evaluation result. According to the method, the optimal model is obtained through training based on the machine learning algorithm, and when the safety literacy of the electric power operation personnel is evaluated, the daily safety behavior data of the electric power operation personnel are input into the optimal model, so that the evaluation result of the electric power operation personnel can be obtained; the technical problem of time-consuming and labor-consuming evaluation of the electric power operating personnel in the prior art is solved, and the safety literacy of the electric power operating personnel is conveniently evaluated. The method can be widely applied to the technical field of electric power safety production.

Description

Technical Field

[0001] The present invention relates to the field of electric power safety technology, and in particular to a method, device and storage medium for evaluating the safety literacy of electric power operators. Background Art

[0002] Power production safety is related to the national economy and people's livelihood. The most active factor in power production is people, and the most difficult factor to control is also people. Human factors are the core factor of safety accidents in power companies. Unsafe behaviors of people in operation will lead to various safety incidents and accidents, causing economic and life and property losses. Among the human factors, the lack of safety knowledge and skills and weak safety awareness of power operators account for a considerable proportion, which shows that power operators lack safety literacy. Therefore, it is necessary to evaluate the safety literacy of power operators.

[0003] Among the existing methods for evaluating the safety literacy of power workers, the commonly used method is based on the analytic hierarchy process, that is, the weight of each indicator is calculated, and then the data of each evaluation indicator is obtained through examinations, assessments, or subjective scoring of employees by leaders. Finally, the different evaluation indicator data and indicator weights are weighted and summed to obtain the evaluation results of the safety literacy of power workers. This evaluation method requires manual scoring and evaluation of employees in various aspects, which often relies on subjective judgment, resulting in the evaluation results being easily affected by the personal bias and subjective emotions of the evaluator, and obtaining indicator data is time-consuming and laborious, resulting in the entire evaluation process being cumbersome and time-consuming. In addition, the evaluation method often directly weights and sums, and the uncertainty and ambiguity in the indicators will affect the evaluation results. Summary of the invention

[0004] The present invention aims to solve one of the technical problems in the related art at least to a certain extent. To this end, one purpose of the present invention is to provide a method, device and storage medium for evaluating the safety literacy of power workers, which can conveniently and quickly evaluate the safety literacy of power workers, facilitate managers to quickly find employees with low safety literacy, and promptly carry out relevant safety education measures to reduce the probability of safety accidents.

[0005] The technical solution adopted by the present invention is: In a first aspect, the present invention provides a method for evaluating the safety literacy of power workers, the method comprising: based on the safety literacy evaluation index data of power workers in historical time periods, using a machine learning algorithm as a prediction model for training to obtain an optimal model; obtaining the safety literacy evaluation index data of the power worker in a first time period, inputting the safety literacy evaluation index data of the first time period into the optimal model, and automatically generating a safety evaluation result, which is one of four types: excellent, good, average, and poor.

[0006] Among them, the safety literacy evaluation index data includes 5 dimensions and 20 indicators. The 5 dimensions are: safety knowledge, safety skills, safety awareness, work qualifications and experience, and safety psychology; among them, the safety knowledge dimension includes 4 indicators: safety regulations examination results, production safety knowledge, theoretical training results, and the number of knowledge-related violations; the safety skills dimension includes 4 indicators: practical training results, safety and preventive measures arrangement ability, health status, and the number of skill-related violations; the safety awareness dimension includes 4 indicators: risk prevention and control mastery ability, safety consultation learning time, number of safety consultation learning articles, and number of safety activities participated in; the work qualifications and experience dimension includes 4 indicators: monthly work times, length of service, professional title, and academic qualifications; the safety psychology dimension includes 4 indicators: sense of responsibility, emotional stability, communication skills, and introversion.

[0007] Among them, based on the safety literacy evaluation index data of power workers in the historical time period, the machine learning algorithm is used as the prediction model for training, and the optimal model includes: obtaining the safety literacy evaluation index data of power workers in the historical time period; cleaning and completing the safety literacy evaluation index data, normalizing some data, and converting the qualitative index data into quantitative index data, so as to generate the index set U={u1,u2,u3,...,u 20}; construct a judgment matrix for all indicators in each dimension, calculate the weight vector of all indicators in each dimension, and then obtain the weights of the 20 indicators, denoted as A={a1,a2,a3,...,a 20}; Determine the review set, recorded as V = {v1, v2, v3, v4}, v1 means excellent, v2 means good, v3 means average, v4 means poor; Determine each indicator u i For each review set v i The membership function is calculated for each index u i In each comment set v i The membership degree of all membership degrees is used to construct a fuzzy comprehensive judgment matrix R. , where r nm is the indicator u n For comments v m The membership degree of n is the number of indicator sets U 20, and m is the number of comment sets 4; calculate the final fuzzy comprehensive evaluation result B, B=A*R=[b1,b2,...,b m ], represents the membership of the final result on each comment, and the comment corresponding to the maximum membership is selected as the final evaluation result, and the evaluation result is a category in the comment set V; the indicators that require manual scoring in the indicator set U are removed, and the remaining indicators and the evaluation result B together constitute a data set for training the optimal model.

[0008] Among them, the step of constructing a judgment matrix for the indicator data of each dimension and calculating the weight vector of the indicator data of each dimension, so as to obtain the weights of the 20 indicator data includes: constructing a judgment matrix step, calculating the relative importance of any two indicators of one dimension, and constructing a 4*4 judgment matrix; layer indicator weight vector calculation step, using the square root method to calculate the weight vector of each layer indicator of the judgment matrix ; CI value calculation step, calculate the maximum eigenvalue of the judgment matrix , and based on Calculate the CI value. ; CR value calculation steps, based on the average random consistency index RI and the CI, calculate CR, ; The consistency check passes the judgment step to determine whether CR is less than or equal to 0.1. If so, it means that the consistency check has passed and enters the next step; if not, return to the step of constructing the judgment matrix and rebuild the judgment matrix; the step of assigning weights to each indicator: the weight vector of each layer of indicators calculated in the step of calculating the weight vector of the layer of indicators is used as the weight of each indicator of this dimension.

[0009] Among them, the determination of each indicator u i For each review set v i The membership function of includes: For each standard normalized indicator, the formula of the membership function is: ; ; ; ;in, is the corresponding index value (within 0~100 points), Indicators In the comments The degree of membership under In addition, for indicators that require manual scoring, the degree of membership of the indicator under each comment can be obtained by simply counting the proportion of votes obtained by the indicator in each comment.

[0010] Among them, the types of machine learning algorithms include: BP neural network, decision tree, and support vector machine SVM.

[0011] In a second aspect, the present invention provides a device for evaluating the safety literacy of power workers, which includes an optimal model generation module for training the optimal model based on the safety literacy evaluation index data of the power workers in a historical time period using a machine learning algorithm as a prediction model; a safety evaluation result generation module for obtaining the safety literacy evaluation index data of the power worker in a first time period, inputting the safety literacy evaluation index data of the first time period into the optimal model, and automatically generating a safety evaluation result, which is one of four types: excellent, good, average, and poor.

[0012] The optimal model generation module includes: a safety literacy evaluation index data acquisition unit, which is used to obtain the safety literacy evaluation index data of power workers in the historical time period; a safety literacy evaluation index data processing unit, which is used to clean the safety literacy evaluation index data, complete the data, normalize some data, and convert the qualitative index data into quantitative index data, so as to generate an index set U={u1,u2,u3,...,u 20}; All indicator weight calculation units are used to construct a judgment matrix for all indicators in each dimension, calculate the weight vectors of all indicators in each dimension, and thus obtain the weights of the 20 indicators, denoted as A={a1,a2,a3,...,a 20}; comment set determination unit, used to determine the comment set, recorded as V = {v1, v2, v3, v4}, v1 represents excellent, v2 represents good, v3 represents average, and v4 represents poor; fuzzy comprehensive judgment matrix R calculation unit, used to determine each indicator u i For each review set v i The membership function is calculated for each index u i In each comment set v i The membership degree of all membership degrees is used to construct a fuzzy comprehensive judgment matrix R. , where r nm is the indicator u n For comments v m The membership degree of n is the number of indicator sets U 20, and m is the number of comment sets 4; the final fuzzy comprehensive evaluation result calculation unit is used to calculate the final fuzzy comprehensive evaluation result B, B=A*R=[b1,b2,...,b m ], represents the membership of the final result on each comment, and the comment corresponding to the maximum membership is selected as the final evaluation result, and the evaluation result is a category in the comment set V; the optimal model training unit is used to eliminate the indicators that need manual scoring in the indicator set U, and the remaining indicators and the evaluation result B together constitute a data set for the optimal model training.

[0013] The weight calculation unit for all indicators includes: a judgment matrix construction subunit, which is used to calculate the relative importance of any two indicators in a dimension and construct a 4*4 judgment matrix; a layer indicator weight vector calculation subunit, which is used to calculate the weight vector of each layer indicator of the judgment matrix using the square root method. ; CI value calculation subunit, used to calculate the maximum eigenvalue of the judgment matrix , and based on Calculate the CI value, ; A CR value calculation subunit, used to calculate CR based on the average random consistency index RI and the CI, ; The consistency check passes the judgment subunit, which is used to determine whether CR is less than or equal to 0.1. If so, it means that the consistency check has passed and enters the subunit of assigning weights to each indicator; if not, it returns to the judgment matrix construction subunit to rebuild the judgment matrix; the subunit of assigning weights to each indicator is used to use the weight vectors of each layer of indicators calculated in the subunit of calculating the weight vector of the indicator of this layer as the weights of each indicator of this dimension.

[0014] In a third aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the method as described above.

[0015] The beneficial effects of the present invention are: The present invention trains the data set based on the machine learning algorithm to obtain the optimal model. When evaluating the safety literacy of power workers, it is only necessary to input the daily safety behavior data of the power workers into the optimal model to obtain the evaluation results of the power workers. This overcomes the technical problems of time-consuming and labor-intensive evaluation of power workers in the prior art, and realizes the quick and convenient evaluation of the safety literacy of power workers.

[0016] Furthermore, when the present invention trains the optimal model, the selected data set is based on objective indicators and the evaluation results obtained by using the fuzzy comprehensive evaluation method, so that the trained optimal model can accurately and fully evaluate the safety literacy of power workers. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of an embodiment of a method for evaluating the safety quality of electric power workers of the present invention; Figure 2 yes Figure 1 A schematic flow chart of an embodiment of step S11; Figure 3 yes Figure 2 A flowchart of an embodiment of step S113; Figure 4It is a structural schematic diagram of an embodiment of a device for evaluating the safety quality of electric power workers of the present invention; Figure 5 yes Figure 4 A schematic structural diagram of an embodiment of an optimal model generation module 11; Figure 6 yes Figure 5 A structural diagram of an embodiment of the all-indicator weight calculation unit 113. DETAILED DESCRIPTION

[0018] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application may be combined with each other. Embodiment 1

[0019] See also Figure 1 , Figure 1 1 is a flow chart of an embodiment of the method for evaluating the safety quality of electric power workers of the present invention. Figure 1 As shown, the method comprises the following steps: S11: Based on the safety literacy evaluation index data of power workers in the historical period, a machine learning algorithm is used as a prediction model for training to obtain the optimal model; The types of machine learning algorithms include: BP neural network, decision tree, support vector machine SVM, etc., which are not limited here. The historical time period is the last month, the last quarter, or the last year.

[0020] See also Figure 2 , Figure 2 yes Figure 1 A flow chart of an embodiment of step S11. Figure 2 As shown, step S11 includes the following sub-steps: S111: Obtain the safety literacy evaluation index data of power workers in the historical period; The safety literacy evaluation index data includes 5 dimensions and 20 indicators. The 5 dimensions are: safety knowledge, safety skills, safety awareness, work qualifications and experience, and safety psychology.

[0021] (1) The safety knowledge dimension includes four indicators: safety regulations test scores, production safety knowledge, theoretical training scores, and the number of knowledge-related violations; Among them, the safety and regulations examination score refers to the average score of all safety and regulations examinations that each employee participated in that month; Safety production knowledge refers to the employees' mastery of the company's safety production regulations and standards. The employees' mastery of safety production knowledge can be assessed through examinations; The theoretical training score is the weighted average score of all trainings that employees have participated in that month, and the formula is as follows: ; Among them, y is the final theoretical training score, n is the number of theoretical trainings the employee participated in that month, is the training hours of the ith theoretical training, The test result of the employee's ith theoretical training; Knowledge-related violations refer to violations such as two-ticket documents and information transmission, which are caused by employees' insufficient understanding of relevant regulations and standards.

[0022] (2) The safety skills dimension includes four indicators: practical training results, safety and preventive measures deployment ability, health status, and number of skill-related violations; The practical training score is similar to the theoretical training score mentioned above, and refers to the weighted average score of the employees' practical training in that month according to the training hours; The ability to arrange safety and preventive measures is evaluated and scored by the team leader or leader of the employee, and given four levels: excellent, good, fair, and poor; Health status refers to the employee's annual physical examination results, which mainly determine whether the employee has any occupational contraindication diseases; Skill-related violations refer to violations in the areas of machine protection, personnel behavior, working environment, etc., which are caused by unsafe behaviors of employees.

[0023] The safety awareness dimension includes four indicators: risk prevention and control mastery, safety consultation and learning time, number of safety consultation and learning articles, and number of safety activities participated in; The risk prevention and control ability is evaluated and scored by the team leader or leader of the employee, and given four levels: excellent, good, medium and poor; Safety consultation learning time refers to the total time employees spend browsing safety consultations in that month; The number of safety consulting learning articles refers to the number of safety consulting articles viewed by employees in that month; The number of safety activities participated in refers to the number of times an employee participates in safety activities carried out by the team in that month.

[0024] The work qualifications and experience dimension includes four indicators: monthly work times, length of service, professional title, and education level; Monthly work times refer to the number of work plans that employees have participated in and completed in that month; Length of service, job title, and education background refer to the employee's years of service, job title, and education background as of the current month.

[0025] The safety psychology dimension includes four indicators: responsibility, emotional stability, communication skills, and introversion and extroversion.

[0026] Responsibility refers to the strength of an individual's self-awareness and whether the individual is responsible for his or her words and deeds. This trait is generally considered positive and mature, and is an indispensable part of human life. The more responsible an employee is, the lower the probability of safety accidents.

[0027] Emotional stability refers to the intensity of an individual's emotional response to unbearable stimulation and pressure, including irritability, depression, frustration, etc. It is mainly related to a person's psychological adaptability, emotional regulation, and psychological stress resistance. Employees' emotional instability often leads to safety accidents, so the worse the emotional stability of an employee, the lower his safety literacy.

[0028] Communicative ability refers to an individual's performance in social and interpersonal interactions, and reflects the individual's ability in social and interpersonal interactions. Gentleness, affinity, compassion, tolerance and concern for others, etc., often show strong communicative ability.

[0029] Introversion and extroversion refer to interpersonal communication and the tendency to gain happiness and satisfaction through socializing. Its traits are divided into extroversion and introversion.

[0030] Optionally, the indicators of the safety psychology dimension can be obtained by questionnaires, such as the 16PF scale, SCL-90, etc. The indicator update frequency is about half a year, so when obtaining employee safety psychology indicators every month, only the data of the latest six months is needed, and employees do not need to fill out the scale every month.

[0031] It should be noted here that if an employee does not participate in the safety regulations examination and safety training in the current month, his / her safety regulations examination score, production safety knowledge, theoretical training score and practical training score will reuse the previous month's score.

[0032] S112: Clean and complete the safety literacy evaluation index data, normalize some of the data, and convert the qualitative index data into quantitative index data, thereby generating an index set U={u1,u2,u3,...,u 20}; When the data collected for a certain indicator in step S111 is incomplete, it is supplemented as the average value of the data set.

[0033] For indicators such as the number of safety activity participations and the number of monthly work, although the final result is a quantitative value, the value range is not 0-100, so the indicator needs to be normalized to 0-100. The maximum and minimum values ​​of the above indicators are normalized to the range of 0-100 using the following formula.

[0034] ; in, is the initial value of the indicator; is the maximum value in the data sample; is the minimum value; y is the normalized result value.

[0035] For health status, professional title, education background, and number of traffic violations, the scoring rules in Tables 1 to 4 can be used to convert qualitative indicators into quantitative indicators. It should be noted here that the scoring rules are not fixed and can be configured later based on actual conditions.

[0036]

[0037]

[0038]

[0039] S113: Construct a judgment matrix for all indicators of each dimension, calculate the weight vector of all indicators of each dimension, and obtain the weights of the 20 indicators, denoted as A={a1,a2,a3,...,a 20}; See also Figure 3 , Figure 3 FIG. 4 is a flow chart of an embodiment of step S113. Figure 3 As shown, step S113 includes the following sub-steps: S1131: Constructing a judgment matrix step: Calculate the relative importance of any two indicators in a dimension and construct a 4*4 judgment matrix; The judgment matrix is ​​composed of experts judging the importance of two indicators at the same level, and scoring is performed using the comparison scale in the table below, where 1 means that the two indicators are equally important, 3 means that one indicator is more important than the other, and the higher the score, the more important one indicator is relative to the other, and the reciprocal score means that one indicator is less important than the other.

[0040]

[0041] The following are examples of the elements of the judgment matrix of the five dimensions of safety knowledge, safety skills, safety awareness, work qualifications and experience, and safety psychology:

[0042]

[0043]

[0044] S1132: Layer index weight vector calculation step: Use the square root method to calculate the weight vector of each layer index of the judgment matrix ; (1) First, calculate the product M of each row of the judgment matrix i ;

[0045] Among them, represents the element in the i-th row and j-th column of the judgment matrix, and n is the dimension of the judgment matrix.

[0046] (2) Calculation of M i The nth root of

[0047] (3) Yes Perform normalization;

[0048] (4) ,in, is the weight vector of each layer indicator of the judgment matrix.

[0049] According to Tables 6 to 10, the weight vector of the security knowledge dimension calculated in Table 6 is [0.529, 0.082, 0.079, 0.319]; The weight vector of the safety skill dimension calculated in Table 7 is [0.1532, 0.3153, 0.4730, 0.0586]; The weight vector of the security awareness dimension calculated in Table 8 is [0.0609, 0.2484, 0.2908, 0.3999]; The weight vectors of the work qualifications and experience dimensions calculated in Table 9 are [0.5937, 0.2150, 0.1337, 0.0540]; The weight vector of the safety psychology dimension calculated in Table 10 is [0.0504, 0.1617, 0.2903, 0.4976].

[0050] S1133: CI value calculation step: Calculate the maximum eigenvalue of the judgment matrix , and based on Calculate the CI value. ; S1134: CR value calculation step: CR is calculated based on the average random consistency index RI and the CI. ;

[0051] In Table 11, the RI value is obtained according to the order of the judgment matrix. For example, in the judgment matrix of the above example, because the order of the matrix is ​​4, according to Table 11, the RI value obtained is 0.89.

[0052] S1135: consistency check pass judgment step: judge whether CR is less than or equal to 0.1. If so, it means that the consistency check has passed, and proceed to the next step S1136; if not, return to the step of constructing the judgment matrix and rebuild the judgment matrix; S1136: Step of assigning weights to each indicator: The weight vector of each layer indicator calculated in step S1132 is used as the weight of each indicator of this dimension.

[0053] According to Tables 6-10, if the CR values ​​calculated by the above five judgment matrices all pass the consistency check, the set A={0.529, 0.082, 0.079, 0.319, 0.1532, 0.3153, 0.4730, 0.0586, 0.0609, 02484, 0.2908, 0.3999, 0.5937, 0.2150, 0.1337, 0.0540, 0.0504, 0.1617, 0.2903,}.

[0054] S114: Determine a review set, denoted as V = {v1, v2, v3, v4}, where v1 represents excellent, v2 represents good, v3 represents average, and v4 represents poor; S115: Determine each indicator u i For each review set v i The membership function is calculated for each index u i In each comment set v i 's membership degrees, construct a fuzzy comprehensive judgment matrix R with all the membership degrees; In step S115, , where r nm is the indicator u n For comments v m The membership degree of , n is the number of indicator sets U (20), m is the number of comment sets (4); For each standard normalized indicator, the membership function formula is: ; ; ; ; in, is the corresponding index value (within 0~100 points), Indicators In the comments The degree of membership under It should be noted here that for indicators that require manual scoring, such as the ability to deploy safety and preventive measures and the ability to grasp risk prevention and control, it is only necessary to count the proportion of votes received by the indicator in each comment to obtain the degree of affiliation of the indicator under each comment.

[0055] Substitute the value of each indicator into the above four formulas to calculate the 4 memberships of each indicator under the four comments. Because there are 20 indicator values, we finally get 20*4=80 memberships. Based on the values ​​of 80 memberships, we generate the matrix R. Each row of R represents the 4 memberships of an indicator under the four comments.

[0056] S116: Calculate the final fuzzy comprehensive evaluation result B, B=A*R=[b1,b2,...,b m ], represents the membership of the final result on each comment, and the comment corresponding to the maximum membership is selected as the final evaluation result. The evaluation result is a category in the comment set V; In step S116, according to the above example, since A is a 1*20 matrix and R is a 20*4 matrix, A*R is subtracted to obtain a 1*4 matrix B. If the maximum value in the matrix B is the first, the final evaluation result is excellent; if the maximum value in the matrix B is the second, the final evaluation result is good; if the maximum value in the matrix B is the third, the final evaluation result is fair; if the maximum value in the matrix B is the fourth, the final evaluation result is poor.

[0057] S117: Eliminate the indicators that require manual scoring in the indicator set U, and use the remaining indicators and the evaluation result B to form a data set for training the optimal model.

[0058] In step S117, the indicators that require manual scoring and evaluation, such as the ability to deploy safety and preventive measures and the ability to control risk, are eliminated, and the remaining indicators and the evaluation result B together constitute a data set for training the optimal model.

[0059] The method of obtaining the optimal model by training the data set is not the focus of the present invention and will not be elaborated here.

[0060] S12: Obtain the safety literacy evaluation index data of the power operator in the first time period, input the safety literacy evaluation index data of the first time period into the optimal model, and automatically generate a safety evaluation result.

[0061] Among them, the safety evaluation result is one of the four types: excellent, good, average, and poor.

[0062] The first time period can be this month, this quarter or this year. The method for obtaining the safety literacy evaluation index data for the first time period is the same as step S111, and the safety literacy evaluation index data for the first time period is processed using the method of step S112, and then the processed data is input into the optimal model to automatically generate the safety evaluation results. It should be noted here that the safety literacy evaluation index data for the first time period is the objective data among the 20 indicators, excluding the indicators that require manual scoring.

[0063] In this embodiment, the data set for training the optimal model is based on multiple indicator data for evaluating the safety literacy of power workers, including five dimensions: safety knowledge, safety skills, safety awareness, work qualifications and experience, and safety psychology. It can evaluate the safety literacy of employees in an all-round and multi-dimensional manner, making the evaluation results more credible.

[0064] Furthermore, the evaluation results obtained by the fuzzy comprehensive evaluation method based on the data set for training the optimal model can reduce the impact of the fuzziness of the evaluation indicators on the evaluation results, making the evaluation results more credible.

[0065] Finally, since the safety literacy evaluation index data of the first time period is objective data, there is no need for subjective scoring index data, so the evaluation results of the power operator can be quickly obtained. Embodiment 2

[0066] See also Figure 4 , Figure 4 Schematic diagram of the structure of an embodiment of the safety literacy evaluation device for electric power workers of the present invention. Figure 4 As shown, the device includes an optimal model generating module 11 and a safety evaluation result generating module 12.

[0067] The optimal model generation module 11 is used to obtain the optimal model by training the safety literacy evaluation index data of power workers in the historical time period using a machine learning algorithm as a prediction model.

[0068] The safety evaluation result generation module 12 is used to obtain the safety literacy evaluation index data of the power operator in the first time period, input the safety literacy evaluation index data of the first time period into the optimal model, and automatically generate a safety evaluation result, which is one of the four types: excellent, good, general, and poor.

[0069] See also Figure 5 , Figure 5 FIG. 1 is a schematic diagram of the structure of an embodiment of the optimal model generation module 11. Figure 5As shown, the optimal model generation module 11 includes a safety literacy evaluation index data acquisition unit 111, a safety literacy evaluation index data processing unit 112, an all index weight calculation unit 113, a comment set determination unit 114, a fuzzy comprehensive judgment matrix R calculation unit 115, a final fuzzy comprehensive evaluation result calculation unit 116, and an optimal model training unit 117.

[0070] The safety literacy evaluation index data acquisition unit 111 is used to acquire the safety literacy evaluation index data of the power operation personnel in the historical time period; The safety literacy evaluation index data processing unit 112 is used to clean the safety literacy evaluation index data, complete the data, normalize some data, and convert the qualitative index data into quantitative index data, so as to generate an index set U={u1,u2,u3,...,u 20}; The all-indicator weight calculation unit 113 is used to construct a judgment matrix for all indicators in each dimension, calculate the weight vectors of all indicators in each dimension, and thus obtain the weights of the 20 indicators, which are recorded as A={a1,a2,a3,...,a 20}; The comment set determination unit 114 is used to determine a comment set, which is denoted as V={v1, v2, v3, v4}, where v1 represents excellent, v2 represents good, v3 represents average, and v4 represents poor; The fuzzy comprehensive judgment matrix R calculation unit 115 is used to determine each indicator u i For each review set v i The membership function is calculated for each index u i In each comment set v i The membership degree of all membership degrees is used to construct a fuzzy comprehensive judgment matrix R. , where r nm is the indicator u n For comments v m The membership degree of , n is the number of indicator sets U (20), m is the number of comment sets (4); The final fuzzy comprehensive evaluation result calculation unit 116 is used to calculate the final fuzzy comprehensive evaluation result B, B=A*R=[b1,b2,...,b m ], represents the membership of the final result on each comment, and the comment corresponding to the maximum membership is selected as the final evaluation result. The evaluation result is a category in the comment set V; The optimal model training unit 117 is used to remove the indicators that need manual scoring from the indicator set U, and use the remaining indicators and the evaluation result B to form a data set for the optimal model training.

[0071] See also Figure 6 , Figure 6 is a schematic diagram of the structure of an embodiment of all indicator weight calculation units 113. Figure 6 As shown, all indicator weight calculation units 113 include a judgment matrix construction subunit 1131, a layer indicator weight vector calculation subunit 1132, a CI value calculation subunit 1133, a CR value calculation subunit 1134, a consistency check pass judgment subunit 1135, and each indicator weight assignment subunit 1136.

[0072] The judgment matrix construction subunit 1131 is used to calculate the relative importance of any two indicators in a dimension and construct a 4*4 judgment matrix; The layer index weight vector calculation subunit 1132 is used to calculate the weight vector of each layer index of the judgment matrix using the square root method. ; The CI value calculation subunit 1133 is used to calculate the maximum eigenvalue of the judgment matrix. , and based on Calculate the CI value, ; The CR value calculation subunit 1134 is used to calculate the CR based on the average random consistency index RI and the CI. ; The consistency check passes the judgment subunit 1135, which is used to judge whether CR is less than or equal to 0.1. If so, it means that the consistency check has passed, and then enters the indicator weight assignment subunit 1136; if not, it returns to the judgment matrix construction subunit 1131 to reconstruct the judgment matrix; The indicator weight assignment subunit 1136 is used to use the weight vectors of the indicators of each layer calculated in the indicator weight vector calculation subunit of this layer as the weights of the indicators of this dimension.

[0073] Specifically, the working methods of the above modules have been described in detail in Implementation 1 and will not be repeated here. Embodiment 3

[0074] The present invention further provides a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the method described in the first embodiment.

[0075] The above is a specific description of the preferred implementation of the present invention, but the invention is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A method for evaluating the safety literacy of power workers, characterized in that: include: Based on the safety literacy evaluation index data of power workers in historical time periods, a machine learning algorithm is used as a prediction model for training to obtain the optimal model; The safety literacy evaluation index data of the power workers in the first time period is obtained, the safety literacy evaluation index data of the first time period is input into the optimal model, and a safety evaluation result is automatically generated. The safety evaluation result is one of the four types: excellent, good, general, and poor.

2. The method according to claim 1, characterized in that The safety literacy evaluation index data includes 5 dimensions and 20 indicators. The 5 dimensions are: safety knowledge, safety skills, safety awareness, work qualifications and experience, and safety psychology; The safety knowledge dimension includes four indicators: safety test scores, production safety knowledge, theoretical training scores, and the number of knowledge-related violations; The safety skills dimension includes four indicators: practical training results, safety and preventive measures deployment ability, health status, and number of skill-related violations; The safety awareness dimension includes four indicators: risk prevention and control mastery, safety consultation and learning time, number of safety consultation and learning articles, and number of safety activities participated in; The work qualifications and experience dimension includes four indicators: monthly work times, length of service, professional title, and education level; The safety psychology dimension includes four indicators: responsibility, emotional stability, communication skills, and introversion and extroversion.

3. The method according to claim 2, characterized in that The safety literacy evaluation index data of power workers in the historical time period is used to train the machine learning algorithm as a prediction model, and the optimal model obtained includes: Obtain the safety literacy evaluation index data of power workers in historical time periods; The safety literacy evaluation index data is cleaned and completed, some data is normalized, and the qualitative index data is converted into quantitative index data, so as to generate an index set U={u1,u2,u3,...,u 20 }; A judgment matrix is ​​constructed for all indicators of each dimension, and the weight vectors of all indicators of each dimension are calculated to obtain the weights of the 20 indicators, which are recorded as A={a1,a2,a3,...,a 20 }; Determine the review set, denoted as V = {v1, v2, v3, v4}, where v1 represents excellent, v2 represents good, v3 represents average, and v4 represents poor; Determine each indicator u i For each review set v i The membership function is calculated for each index u i In each review set v i The membership degree of all membership degrees is used to construct a fuzzy comprehensive judgment matrix R. , where r nm is the indicator u n For comments v m The membership degree of , n is the number of indicator sets U (20), m is the number of comment sets (4); Calculate the final fuzzy comprehensive evaluation result B, B=A*R=[b1,b2,...,b m ], represents the membership of the final result on each comment, and the comment corresponding to the maximum membership is selected as the final evaluation result, and the evaluation result is a category in the comment set V; The indicators that require manual scoring are eliminated from the indicator set U, and the remaining indicators and the evaluation result B together constitute a data set for the optimal model training.

4. The method according to claim 3, characterized in that: The step of constructing a judgment matrix for the indicator data of each dimension and calculating the weight vector of the indicator data of each dimension to obtain the weights of the 20 indicator data includes: Construct a judgment matrix step, calculate the relative importance of any two indicators in a dimension, and build a 4*4 judgment matrix; The step of calculating the weight vector of each layer indicator is to use the square root method to calculate the weight vector of each layer indicator of the judgment matrix. ; CI value calculation step, calculating the maximum eigenvalue of the judgment matrix , and based on Calculate the CI value. ; CR value calculation step, based on the average random consistency index RI and the CI, calculate CR, ; The consistency check passes the judgment step to determine whether CR is less than or equal to 0.

1. If so, it means that the consistency check has passed and the next step is entered; if not, it returns to the step of constructing the judgment matrix and rebuilds the judgment matrix; The step of assigning weights to each indicator: using the weight vectors of each layer indicator calculated in the step of calculating the layer indicator weight vector as the weights of each indicator of the dimension.

5. The method according to claim 3, characterized in that: The determination of each indicator u i For each review set v i The membership functions include: For each standard normalized indicator, the membership function formula is: ； ； ； ； in, is the corresponding index value (within 0~100 points), Indicator In the comments The degree of membership under In addition, for indicators that require manual scoring, the degree of membership of the indicator under each comment can be obtained by simply counting the proportion of votes obtained in each comment of the indicator.

6. The method according to any one of claims 1 to 5, characterized in that: The types of machine learning algorithms include: BP neural network, decision tree, and support vector machine SVM.

7. A device for evaluating the safety literacy of electric power workers, characterized in that: include: The optimal model generation module is used to train the prediction model using a machine learning algorithm based on the safety literacy evaluation index data of power workers in the historical period to obtain the optimal model; The safety evaluation result generation module is used to obtain the safety literacy evaluation index data of the power operation personnel in the first time period, input the safety literacy evaluation index data of the first time period into the optimal model, and automatically generate a safety evaluation result, which is one of the four types: excellent, good, general, and poor.

8. The device according to claim 7, characterized in that The optimal model generation module comprises: A safety literacy evaluation index data acquisition unit is used to acquire safety literacy evaluation index data of power operators in a historical period of time; The safety literacy evaluation index data processing unit is used to clean the safety literacy evaluation index data, complete the data, normalize some data, and convert the qualitative index data into quantitative index data, so as to generate an index set U={u1,u2,u3,...,u 20 }; The weight calculation unit of all indicators is used to construct a judgment matrix for all indicators of each dimension, calculate the weight vector of all indicators of each dimension, and thus obtain the weights of the 20 indicators, which are recorded as A={a1,a2,a3,...,a 20 }; A review set determination unit, used to determine a review set, denoted as V = {v1, v2, v3, v4}, where v1 represents excellent, v2 represents good, v3 represents average, and v4 represents poor; Fuzzy comprehensive judgment matrix R calculation unit is used to determine each indicator u i For each review set v i The membership function is calculated for each index u i In each review set v i The membership degree of all membership degrees is used to construct a fuzzy comprehensive judgment matrix R. , where r nm is the indicator u n For comments v m The membership degree of , n is the number of indicator sets U (20), m is the number of comment sets (4); The final fuzzy comprehensive evaluation result calculation unit is used to calculate the final fuzzy comprehensive evaluation result B, B=A*R=[b1,b2,...,b m ], represents the membership of the final result on each comment, and the comment corresponding to the maximum membership is selected as the final evaluation result, and the evaluation result is a category in the comment set V; The optimal model training unit is used to eliminate the indicators that need manual scoring in the indicator set U, and use the remaining indicators and the evaluation result B to form a data set for the optimal model training.

9. The device according to claim 8, characterized in that The all-indicator weight calculation unit includes: The judgment matrix construction subunit is used to calculate the relative importance of any two indicators in a dimension and construct a 4*4 judgment matrix; The layer index weight vector calculation subunit is used to calculate the weight vector of each layer index of the judgment matrix using the square root method ; The CI value calculation subunit is used to calculate the maximum eigenvalue of the judgment matrix , and based on Calculate the CI value. ; The CR value calculation subunit is used to calculate the CR based on the average random consistency index RI and the CI. ; The consistency check passes the judgment subunit, which is used to determine whether CR is less than or equal to 0.

1. If so, it means that the consistency check has passed and enters the weight assignment subunit of each indicator; if not, it returns to the judgment matrix construction subunit to rebuild the judgment matrix; Each indicator weight assignment subunit is used to use the weight vector of each layer indicator calculated in the layer indicator weight vector calculation subunit as the weight of each indicator of the dimension.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the method according to any one of claims 1 to 6.

Images