Human factor reliability quantitative evaluation method and system based on dynamic characteristics

Abstract

The invention provides a human factor reliability quantitative evaluation method and system based on dynamic characteristics, and relates to the technical field of reliability analysis, a task situation model of wind power equipment operation is constructed, a CPC factor corresponding to any task situation in the task situation model is obtained, and a human factor reliability quantitative evaluation result is obtained based on a fuzzy set, a Bayesian network and the CPC factors. Constructing a human factor reliability quantification model; inputting the real-time factors in the task situation into a human factor reliability quantification model, and respectively obtaining reliability factors corresponding to the real-time factors in the task situation and personal state characteristics of wind power equipment management personnel corresponding to the task situation; and based on the facial features and the behavior features, respectively obtaining facial feature points and human body feature points, and according to fatigue indexes corresponding to the facial feature points and the human body feature points, inputting personal working state indexes corresponding to the fatigue indexes into the human factor reliability quantification model to obtain a human factor error probability value. According to the invention, the robustness and accuracy of the human factor reliability quantification model can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of reliability analysis, and in particular to a method and system for quantitatively evaluating human reliability based on dynamic characteristics. Background Art

[0002] With the rapid development of technology, the automation level of complex industrial systems has been continuously enhanced, and the reliability of technical systems in industries such as aviation, aerospace, navigation, nuclear power, and railways has been significantly enhanced. However, this enhancement cannot prevent the occurrence of major disasters. Although the proportion of human participation in the entire complex system has become smaller, due to the physiological and psychological characteristics of human beings, they are highly uncontrollable. Many studies have shown that the failure of human-machine systems is often caused by human factors. More than 70% of the casualties in the world each year are related to human errors, and the number is increasing year by year, especially in high-risk areas such as coal mining, aerospace, and nuclear industry, where the proportion of human-induced accidents is more prominent.

[0003] The Chinese patent with publication number CN104134160A discloses a method for evaluating the human reliability of a power system based on the CREAM model, comprising the following steps: 1) obtaining general performance condition factors and corresponding state values, and calculating the weight coefficients of each factor by means of the analytic hierarchy process; 2) calculating the state values ​​of each general performance condition according to the weight coefficients and the corresponding state values; 3) establishing a fuzzy CREAM model, using the general performance condition state values ​​obtained in step 2) as model input, performing fuzzy calculations, and obtaining the final human error probability value. However, the method provided in the above application lacks a detailed analysis and evaluation of the content and process of the actions performed by the personnel, and only directly gives the human error probability value through the fuzzy CREAM model, and the calculation basis and calculation result are less credible. Therefore, it is very necessary to provide a method and system for quantitative evaluation of human reliability based on dynamic characteristics to improve the robustness and accuracy of the human reliability quantitative model. Summary of the invention

[0004] In view of this, the present invention proposes a human reliability quantitative evaluation method and system based on dynamic characteristics. By using fuzzy sets to handle uncertainty and Bayesian networks to describe the probabilistic relationship between various factors, the combination of the two further improves the robustness and accuracy of the human reliability quantitative model, thereby improving the accuracy of the human error probability value.

[0005] The present invention provides a method for quantitatively evaluating human reliability based on dynamic characteristics, the method comprising:

[0006] Constructing a task scenario model for wind power equipment operation, obtaining a CPC factor corresponding to any task scenario in the task scenario model, and constructing a human reliability quantification model based on fuzzy sets, Bayesian networks and the CPC factor;

[0007] Input the real-time factors in the task scenario into the human reliability quantification model, and respectively obtain the reliability factors corresponding to the real-time factors in the task scenario and the personal status characteristics of the wind power equipment management personnel corresponding to the task scenario, wherein the personal status characteristics include facial features and behavioral features;

[0008] Based on the facial features and behavioral features, facial feature points and body feature points are obtained respectively, and according to the fatigue index corresponding to the facial feature points and the body feature points, the personal work status index corresponding to the fatigue index is input into the human reliability quantification model to obtain the human error probability value.

[0009] On the basis of the above technical solution, preferably, the human reliability quantitative model is constructed based on fuzzy sets, Bayesian networks and the CPC factor, specifically including:

[0010] Analyzing cognitive activities and cognitive functions in the task context according to the CREAM method, and calculating a basic value of the probability of cognitive function failure in the task context based on the cognitive activities and the cognitive functions;

[0011] According to the control mode in the CREAM method and the probability interval corresponding to the control mode, a fuzzy set corresponding to the control mode is constructed, and the uncertain relationship between the CPC factor and the control mode is constructed through a Bayesian network;

[0012] Based on the fuzzy set corresponding to the control mode and the uncertain relationship between the CPC factor and the control mode, a human reliability quantitative model is constructed.

[0013] On the basis of the above technical solution, preferably, the analyzing the cognitive activities and cognitive functions in the task context according to the CREAM method, and calculating the basic value of the probability of cognitive function failure in the task context based on the cognitive activities and the cognitive functions, specifically includes:

[0014] Decomposing the task in the task situation into a plurality of task steps, and determining the cognitive function corresponding to each task step;

[0015] Determine the error categories in cognitive functions corresponding to each sub-step in each task step, and calculate the basic error probability corresponding to each error category, wherein each task step includes multiple sub-steps;

[0016] According to the connection relationship between each sub-step in each task step, the basic value of the probability of cognitive function failure in the task scenario is calculated.

[0017] More preferably, the obtaining of the reliability factor corresponding to the real-time factor in the task scenario and the personal status characteristics of the wind power equipment management personnel corresponding to the task scenario respectively specifically includes:

[0018] Establishing a mapping relationship between the real-time factors and the reliability factors in the task scenario;

[0019] Matching personal status characteristics of wind power equipment management personnel corresponding to the task scenario according to the reliability factor;

[0020] Based on the personal status feature and the human work status evaluation function, a status evaluation value corresponding to the personal status feature is obtained.

[0021] More preferably, the acquiring of facial feature points and body feature points based on the facial features and behavioral features respectively specifically includes:

[0022] Based on the facial feature point recognition model of Dlib and the facial features, obtain the facial body feature points and the spatial position points corresponding to the facial feature points;

[0023] Based on the human posture estimation method of OpenPose and the behavioral characteristics, feature point detection is performed on the human feature points to obtain the human feature points and the spatial position points corresponding to the human feature points;

[0024] Based on the OpenPose posture estimation algorithm, the facial feature points and the human body feature points, a human skeleton graph corresponding to the facial feature points and the human body feature points is constructed, and a structural vector between the facial feature points and the human body feature points is constructed according to the human skeleton graph.

[0025] More preferably, the obtaining of personal status characteristics of wind power equipment management personnel corresponding to the task context specifically includes:

[0026] Based on the Stacking integrated multi-feature behavior recognition model, the reliability analysis is performed on the human body features collected in real time in the task scenario to obtain facial features and behavior features respectively.

[0027] More preferably, the expression of the human working status evaluation function is:

[0028]

[0029] Among them, I c Indicates the real-time working status value, f(action i ) represents the behavior quantization value of the i-th frame, freq represents the device sampling frequency, t represents the sampling duration, I concentration is the result of concentration evaluation, I minIndicates the minimum value of the concentration index, Imin max It represents the difference between the maximum and minimum values ​​of the concentration index, S represents the evaluation result of the individual work status, and I fatigue represents the fatigue index, and w is the corresponding weight of the fatigue index.

[0030] In a second aspect of the present application, a human factor reliability quantitative evaluation system based on dynamic features is provided, wherein the human factor reliability quantitative evaluation system comprises a model building module, a feature acquisition module and a probability analysis module, wherein:

[0031] The model building module is used to build a task scenario model for the operation of wind power equipment, obtain the CPC factor corresponding to any task scenario in the task scenario model, and build a human reliability quantification model based on fuzzy sets, Bayesian networks and the CPC factor;

[0032] The feature acquisition module is used to input the real-time factors in the task scenario into the human reliability quantification model, and respectively obtain the reliability factors corresponding to the real-time factors in the task scenario and the personal status characteristics of the wind power equipment management personnel corresponding to the task scenario, wherein the personal status characteristics include facial features and behavioral features;

[0033] The probability analysis module is used to obtain facial feature points and body feature points respectively based on the facial features and behavioral features, and according to the fatigue index corresponding to the facial feature points and the body feature points, input the personal work status index corresponding to the fatigue index into the human reliability quantification model to obtain the human error probability value.

[0034] In a third aspect of the present application, an electronic device is provided, comprising a processor, a memory, a user interface and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory.

[0035] In a fourth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored, and the computer program is executed by a processor to implement the steps of a human reliability quantitative evaluation method based on dynamic features.

[0036] The human factor reliability quantitative evaluation method and system based on dynamic characteristics provided by the present invention have the following beneficial effects compared with the prior art:

[0037] (1) By constructing a task scenario model for wind power equipment operation, the operating procedures and environmental factors of operators under different tasks are described in detail. At the same time, fuzzy sets are used to deal with uncertainty, and Bayesian networks are used to describe the probabilistic relationship between factors. The combination of the two further improves the robustness and accuracy of the human reliability quantification model. In addition, the reliability factors corresponding to the real-time factors and the facial and behavioral characteristics of the operators are obtained respectively, which can comprehensively evaluate the factors affecting human reliability and improve the comprehensiveness of the human reliability evaluation. In addition, the physiological and posture information of the operators can be captured, and the corresponding fatigue index can be obtained to quantify the fatigue level of the operators and other working conditions, thereby regulating and monitoring the behavior of the staff, reducing the probability of human errors, and improving the accuracy of the probability value of human errors.

[0038] (2) The CREAM method is used to analyze cognitive activities and functions, and the basic value of the probability of cognitive function failure is obtained, which provides an objective data basis for human reliability evaluation. Fuzzy sets are used to deal with the uncertainty of the control mode, and the Bayesian network is used to establish the probabilistic relationship between the CPC factor and the control mode, which effectively solves the uncertainty problem that is difficult to deal with by traditional methods. At the same time, the quantitative model constructed by integrating the above elements not only ensures the scientificity and accuracy of the evaluation results, but also realizes the effective prediction of human errors, thereby enhancing the model's adaptability to uncertain factors.

[0039] (3) The Dlib model is used to accurately locate and obtain the spatial position of facial feature points. The OpenPose technology is used to accurately capture human feature points and their spatial coordinates, realizing multi-dimensional collaborative analysis of facial and human features. A complete human skeleton diagram is constructed to convert discrete feature points into a systematic structural expression. The generated structural vectors between feature points provide richer spatial relationship information, and an association mapping between facial features and human posture is established. At the same time, it supports real-time tracking of facial expressions and human postures, can dynamically capture state changes and behavioral characteristics, and provide a continuous monitoring data stream, thereby enhancing the ability to identify abnormal behaviors and improving the accuracy of human reliability analysis. Through multi-dimensional feature extraction and structured expression, the accuracy and real-time performance of state recognition of wind power equipment management personnel are significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1A schematic diagram of a process flow of a human factor reliability quantitative evaluation method based on dynamic characteristics provided by the present invention;

[0042] Figure 2 A schematic diagram of the structure of the Bayesian network provided by the present invention;

[0043] Figure 3 An example diagram of the model structure of the human body posture algorithm provided by the present invention;

[0044] Figure 4 A schematic diagram of the structure of the human reliability quantitative evaluation system provided by the present invention;

[0045] Figure 5 This is a schematic structural diagram of an electronic device provided by the present invention.

[0046] Explanation of the accompanying symbols: 1. Human reliability quantitative evaluation system; 11. Model building module; 12. Feature acquisition module; 13. Probability analysis module; 2. Electronic device; 21. Processor; 22. Communication bus; 23. User interface; 24. Network interface; 25. Memory. DETAILED DESCRIPTION

[0047] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0048] Before introducing the embodiments of the present invention, some terms and their abbreviations involved in the embodiments of the present invention are first defined and explained.

[0049] CREAM method: Task analysis analyzes the entire task based on the required scenarios and events, and establishes the corresponding event sequence. Evaluate common performance conditions (CPC) CPC is used to reflect the nature of the task from an overall perspective, and the CPC factor is mainly evaluated by professionals in the industry. Determine the scenario control mode The scenario control mode is used in the CREAM method to describe the person's control of the event, which is determined according to the common performance conditions in the specific environment.

[0050] There is a specific relationship between the control mode and CPC. By combining the evaluation results of each CPC factor, the control mode of the staff under the current situation conditions can be determined. For example, the control mode is determined according to the number of CPC factors that increase and decrease the impact on reliability. The relationship can be reflected by a coordinate graph (the horizontal axis represents the number of CPC factors with a reduced impact on reliability, and the vertical axis represents the number of CPC factors with an increased impact on reliability).

[0051] The present invention discloses a human factor reliability quantitative evaluation method based on dynamic characteristics, referring to Figure 1 The steps of the method include steps S1 to S3.

[0052] Step S1, constructing a task scenario model for wind power equipment operation, obtaining CPC factors corresponding to any task scenario in the task scenario model, and constructing a human reliability quantification model based on fuzzy sets, Bayesian networks and CPC factors.

[0053] In this embodiment, the influencing factors of human error are determined based on the actual work of the wind power equipment manager, wherein the influencing factors of human error are determined mainly by referring to the wind power equipment work rules and relevant literature.

[0054] Human factors: In the wind power equipment emergency system, humans are the only active subjects. Human erroneous behaviors can easily lead to human-induced accidents, which may cause equipment failures, organizational management defects, member cooperation problems, etc., which can easily become hidden dangers of accidents.

[0055] Environmental factors: Environmental factors include wind power equipment emergency system facilities and working conditions. Daily maintenance of fire-fighting facilities is critical. If the on-duty personnel do not find the potential hidden dangers of the equipment, a failure may occur during use, thus aggravating the severity of the accident. Working conditions will affect the working status of the on-duty personnel. In addition, extreme weather is more likely to cause fire accidents.

[0056] Organizational management factors: In the wind power equipment emergency system, the on-duty personnel must ensure the safe and efficient operation of the system, which requires the cooperation of team members. If the atmosphere between members is not harmonious, lack of understanding and communication, it will indirectly lead to human-caused accidents. The organizational management defects that may exist in the wind power equipment safety emergency system include: insufficient skill training, insufficient safety education, inadequate equipment maintenance management, improper organizational structure setting, and incomplete allocation of tasks or time (unreasonable allocation of day and night shifts).

[0057] Furthermore, based on the HRA risk analysis method and combined with the actual work of the duty officers, task situation analysis was conducted on the main work content of the wind power equipment manager and task situation analysis was conducted on the actual work content;

[0058] The main contents of task scenario analysis: Through actual investigation, review of relevant rules and regulations, etc., the work content of wind power equipment operators was sorted out from daily situations and emergency situations. Through field observation and interviews of the work behavior of the operators, their status in wind power equipment management was clarified and their behavioral work characteristics were summarized. By referring to the HRA risk analysis method and combining the actual work of the operators, the main work content of wind power equipment operators was summarized into 5 general tasks for analysis: (1) daily maintenance of fire alarm controllers; (2) shift handover; (3) fire emergency handling; (4) false alarm handling; (5) fire alarm controller abnormalities. The above method of decomposing large tasks into multiple sub-steps and analyzing each sub-step to lay the foundation for the subsequent quantification of human errors is called task scenario analysis.

[0059] This step also includes steps S11 to S13.

[0060] Step S11, analyzing cognitive activities and cognitive functions in the task context according to the CREAM method, and calculating a basic value of the probability of cognitive function failure in the task context based on the cognitive activities and cognitive functions.

[0061] In one example, a task in a task context is decomposed into multiple task steps, and the cognitive function corresponding to each task step is determined; the error categories in the cognitive function corresponding to each sub-step in each task step are determined, and the basic error probability corresponding to each error category is calculated, wherein each task step includes multiple sub-steps; based on the connection relationship between each sub-step in each task step, the basic value of the cognitive function failure probability in the task context is calculated.

[0062] Furthermore, the extended method is used to analyze cognitive activities and cognitive functions in the task, and the basic value of the probability of cognitive function failure is corrected according to the CPC factor under the corresponding situational conditions, so as to realize the prediction and analysis of human errors that may occur in the task. The specific steps are as follows:

[0063] Decompose task steps: By consulting materials and consulting relevant experts, organize and clarify specific work tasks, establish a task step sequence, and decompose a task into multiple executable steps.

[0064] Determine the cognitive function that the task steps are most likely to correspond to: The specific relationship between cognitive activities and corresponding cognitive functions is shown in Table 1. Table 1 is a table of the correspondence between cognitive activities and corresponding cognitive functions. After breaking down the work tasks into specific task steps, cognitive activities are used to represent the cognitive profile of the corresponding steps. CREAM has 15 cognitive activities such as inspection, communication, comparison, and adjustment. Based on the correspondence between cognitive activities and cognitive functions, the cognitive function profile in the current work task is obtained.

[0065] Table 1

[0066]

[0067] Furthermore, the error categories are determined: the CREAM method decomposes the cognitive functions that may occur in the operator during the task into four categories: planning, interpretation, observation, and execution. The four categories of cognitive functions also correspond to multiple possible error categories, as shown in Table 2. The most likely error category is determined according to the actual situation, so as to obtain the corresponding basic error probability. Table 2 shows the cognitive functions, corresponding error categories and error probabilities.

[0068] Table 2

[0069]

[0070] Calculate the error probability of sub-steps: After obtaining the basic error probability value of the corresponding sub-step, evaluate the CPC factor according to the basic method to determine whether its impact on reliability is enhanced, insignificant or weakened, and obtain the cognitive function weights corresponding to different impact effects. Multiply the cognitive function weight factors corresponding to the nine CPC factors to obtain the total weight factor of the corresponding sub-step. Multiply the basic error probability of the sub-step by the corresponding total weight factor to obtain the corrected human error probability of the sub-step.

[0071] In one example, the situational environment index β is an improvement on the CREAM method. It reflects the degree of influence of a specific environment on human reliability. The larger the β, the lower the probability of cognitive error of people in the situation. The smaller the β, the higher the probability of cognitive error of people in the situation. CREAM divides the influence level of CPC factors on reliability into three levels: enhancement, no influence, and weakening. In order to facilitate the quantification of the influence of CPC factors, the quantitative effect value is defined as c. When the influence factor level is "enhancement", the c value is 1; when the influence factor level is "no influence", the c value is 0; when the influence factor level is "weakened", the c value is -1. The sum of all the quantitative values ​​of each CPC factor is the β value, and the calculation formula is:

[0072]

[0073] Among them, c i Indicates the quantitative value corresponding to the CPC factor of the i-th category.

[0074] Since the probability range of human cognitive function error is [0.0001, 1], CFP changes exponentially. This idea has been proven in research and application. The corresponding relationship between the situation environment index β and CFP is calculated as follows:

[0075]

[0076] Among them, CFP0 represents the basic error probability of the population to be calculated, and k represents the coefficient to be measured of the scenario environment index β.

[0077] The basic error probability and coefficient k of the whole to be calculated can be determined according to the maximum and minimum values ​​of the human error probability and the situation environment index. The calculation formula is:

[0078]

[0079] For the CPC factor changes, the number of factors that have an "enhanced" and "weakened" impact on reliability is 9, so β max =9,β min =-9. min = -9, the situation control mode is chaos mode, and the maximum error probability is 1, that is, CFP max =1;β max =9, the situation control mode is the strategic mode, and the minimum error probability is 0.0001, that is, CFP min =0.0001. Substituting the corresponding data into the relevant formula for simultaneous calculation, we can finally calculate k=-0.222 and CFP0=0.01.

[0080] Thus, the error probability considering the impact of the situational environment is obtained, and the calculation formula is:

[0081] CFP=CFP0×10 -0.222β

[0082] When the scenario environment index β = 0, it means that the CPC factor will not have an effective impact on reliability. Substituting it into the formula, we get CFP0 = 0.01. At this time, the control mode is tactical mode, which is consistent with the corresponding human error probability range in the table. Because the value of CFP0 has been determined, we only need to determine the scenario environment index β, and then we can get the human error probability in the corresponding subtask or step by calculation.

[0083] Step S12, constructing a fuzzy set corresponding to the control mode according to the control mode and the probability interval corresponding to the control mode in the CREAM method, and constructing an uncertain relationship between the CPC factor and the control mode through a Bayesian network.

[0084] In this embodiment, fuzzy sets are introduced to perform fuzzy membership analysis on the expected effect of CPC factors. Fuzzy sets assign values ​​to each element in the range of [0,1] through membership functions, reflecting the probability of it belonging to a certain set. This paper adopts trapezoidal and triangular membership functions with an input interval of [0,100]. The results are quantified based on expert evaluation, and three fuzzy sets of "enhancement", "unobvious" and "weakening" are established. The calculation formula is:

[0085]

[0086] Among them, x represents the evaluation result of the corresponding CPC factor, which is substituted into the formula to calculate the corresponding membership degree, f i (x) represents the membership quantification value of CPC factor under the corresponding expected effect, where i is 1, 2, and 3, representing the expected effect of “enhanced”, “not obvious”, and “weakened” respectively.

[0087] Step S13, constructing a human reliability quantitative model based on the fuzzy set corresponding to the control mode and the uncertain relationship between the CPC factor and the control mode.

[0088] In this embodiment, a Bayesian network is constructed to indirectly realize the calculation of the probability of human error. Therefore, this paper takes the relationship between CPC factors and control modes as the theoretical basis for constructing a Bayesian network. The Bayesian network is constructed with the adjusted 9 CPC factors as the root node and the 4 scenario control modes as the leaf nodes. The root node refers to the division of CPC factor levels and evaluates the leaf node control mode. In the process of Bayesian network construction, it is necessary to determine the relationship between CPC impact effect and control mode. Figure 3 The number of CPC factors for different expected effects is used as the basis for judgment. By using GeNIe software for Bayesian analysis, the Bayesian network structure diagram constructed is shown in the figure below: Figure 2 shown.

[0089] Furthermore, according to the control modes and their corresponding probability intervals in the CREAM method, a fuzzy set of control modes is established based on fuzzy mathematics theory. The probability interval ranges corresponding to the four control modes are considered, and the probability intervals are logarithmically calculated with a base of 10 to construct a triangular fuzzy function. The calculation formula is:

[0090]

[0091] Where x represents the logarithm of the corresponding error probability with base 10.

[0092] The uncertain relationship between CPC factor and control mode is established through Bayesian network, and the probability interval of control mode is fuzzified by using triangular membership function. There is a lack of specific quantitative calculation of human reliability. The probability value when the membership is 1 is the calibration probability of the corresponding membership function, and the defuzzification operation is performed on this basis. When the membership is 1, the calibration probabilities of the corresponding control modes are 2.24×10 -4 , 0.01, 0.0708, 0.3162, the calculation formula of human error probability HEP of the improved basic method is:

[0093]

[0094] Among them, P i represents the probability of the ith control mode, S i represents the calibration probability corresponding to the i-th control mode.

[0095] Furthermore, the calculation of cognitive function weight is improved, including the following steps:

[0096] Calculate the cognitive weight corresponding to each CPC factor: obtain the membership of each CPC factor through evaluation and calculation, multiply the membership by the corresponding cognitive function weight, and then add them up to get the new weight value. The calculation formula is:

[0097]

[0098] Among them, w ij is the weight of the i-th cognitive function corresponding to the j-th influencing factor obtained by combining the membership degree; w ijk is the weight value of the i-th cognitive function of the j-th influencing factor at the k-th reliability level; m k is the membership degree at the kth reliability level; i represents the four cognitive functions; j represents the nine CPC factors; and k represents the three reliability expectation effects.

[0099] Calculate the total weight of the j-th cognitive function using the following formula:

[0100]

[0101] Among them, w ij is the weight of the i-th cognitive function corresponding to the j-th influencing factor, where i represents the four cognitive functions and j represents the nine CPC factors.

[0102] The CREAM method is used to analyze cognitive activities and functions, and the basic value of the probability of cognitive function failure is obtained, which provides an objective data basis for human reliability evaluation. Fuzzy sets are used to deal with the uncertainty of the control mode, and the Bayesian network is used to establish the probabilistic association between the CPC factor and the control mode, which effectively solves the uncertainty problem that is difficult to deal with by traditional methods. At the same time, the quantitative model constructed by integrating the above elements not only ensures the scientificity and accuracy of the evaluation results, but also realizes the effective prediction of human errors, thereby enhancing the model's adaptability to uncertain factors and providing more comprehensive and reliable technical support for the reliability evaluation of wind power equipment managers.

[0103] Step S2, input the real-time factors in the task scenario into the human reliability quantification model, and obtain the reliability factors corresponding to the real-time factors in the task scenario and the personal status characteristics of the wind power equipment management personnel corresponding to the task scenario, wherein the personal status characteristics include facial features and behavioral characteristics.

[0104] This step also includes steps S21 to S23.

[0105] Step S21, establishing a mapping relationship between the real-time factors and the reliability factors in the task context.

[0106] In one example, a multi-feature behavior recognition model based on Stacking integration is used to perform reliability analysis on human features collected in real time in a task scenario, and obtain facial features and behavioral features respectively.

[0107] In this step, based on the multi-feature behavior recognition model integrated by Stacking, the duration information of different behaviors within a period of time can be collected in real time and mapped with the reliability analysis. The reliability factor is the reliability level concept mentioned above. The influence of the working state on the reliability factor is: enhanced, not obvious, and weakened. For example, if the working state is: sleeping, the reliability of a person will be weakened, that is, the reliability of a person will be weakened when a person is sleeping.

[0108] Step S22, matching the personal status characteristics of the wind power equipment management personnel corresponding to the task context according to the reliability factor.

[0109] In this step, the work status of the duty officer is analyzed by computer vision technology, mainly measuring the facial and behavioral characteristics of the duty officer during the work process, focusing on the fatigue and concentration that are closely related to the real-time work status during the duty process. By quantifying and integrating the monitored behavioral characteristics, it is used as the evaluation quantity of the individual work status, thereby achieving an effective evaluation of the real-time work status of the duty officer. Among them, facial measurements include: yawning, blinking, closing eyes, etc.; behavioral feature measurements include: sleeping, playing with mobile phones, communicating, sitting upright, etc.; feature measurements are evaluations achieved using fuzzy measurement methods.

[0110] Step S23, based on the personal state characteristics and the human work state evaluation function, obtaining the state evaluation value corresponding to the personal state characteristics.

[0111] In this embodiment, based on the normalization results of fatigue index and concentration, a comprehensive evaluation value is obtained in the form of a synthetic average, thereby comprehensively reflecting the real-time status of the duty officer at work. The calculation formula is:

[0112]

[0113]

[0114] The expression of the human working status evaluation function is:

[0115]

[0116] Among them, I cIndicates the real-time working status value, f(action i ) represents the behavior quantization value of the i-th frame, freq represents the device sampling frequency, t represents the sampling duration, I concentration is the result of concentration evaluation, I min Indicates the minimum value of the concentration index, Imin max It represents the difference between the maximum and minimum values ​​of the concentration index, S represents the evaluation result of the individual work status, and I fatigue represents the fatigue index, and w is the corresponding weight of the fatigue index.

[0117] In one example, a comprehensive evaluation model based on fatigue index and concentration is constructed, as shown in Table 3, which is a comparison table of comprehensive evaluation synthesis models.

[0118] Table 3

[0119]

[0120] Step S3, based on facial features and behavioral features, facial feature points and body feature points are obtained respectively, and according to the fatigue index corresponding to the facial feature points and the body feature points, the personal work status index corresponding to the fatigue index is input into the human reliability quantification model to obtain the human error probability value.

[0121] This step also includes steps S31 to S33.

[0122] Step S31, based on the facial feature point recognition model and facial features of Dlib, obtain the facial body feature points and the spatial position points corresponding to the facial feature points.

[0123] In this step, the gradient boosting tree algorithm used by Dlib is a face alignment algorithm based on regression trees. It trains the marked facial shapes by establishing a residual regression tree, stores the residual regressors on each leaf node, and when the input falls on a leaf node, changes the residual to the input for regression. Finally, all residuals are superimposed to complete the detection of facial feature points.

[0124] Step S32: Based on the human posture estimation method and behavior characteristics of OpenPose, feature point detection is performed on human feature points to obtain human feature points and spatial position points corresponding to the human feature points.

[0125] In this step, the model structure of the human posture algorithm is as follows Figure 3As shown in Figure 3, the coordinate information of the human body joints obtained by Openpose can effectively express the global characteristics of human behavior, fully describe the relative position changes of the torso when the human body produces behavior, and can well eliminate the interference of irrelevant information in the image. The spatial position distribution information of the human body key points and the extracted human body corresponding key point information are shown in Table 3. Table 4 is the human body corresponding key point information table.

[0126] Table 4

[0127]

[0128] Step S33, based on the OpenPose posture estimation algorithm, facial feature points and body feature points, construct a human skeleton graph corresponding to the facial feature points and the body feature points, and construct a structural vector between the facial feature points and the body feature points according to the human skeleton graph.

[0129] In this step, the structural vectors between the feature points are constructed in combination with the structural features of the human body, and then the cosine value and vector modulus of the angle between the vectors are calculated using these structural vectors, and the angle features of 18 joint points related to human behavior are extracted from the 25 feature points. Among them, according to the human body structure and the characteristics of the actual application scenarios, the coordinate information of the parts below the waist cannot be collected in most cases due to the obstruction of objects such as desks during work. Therefore, when extracting action features, the upper body will be considered more, and the angle features related to the upper body will be extracted. In order to consider the changes in the overall skeleton state and application scenarios, this paper also appropriately selects the features below the waist, and the naming rules are named according to the three joint points that make up the angle, as shown in Table 5.

[0130] Table 5

[0131]

[0132]

[0133] The selection of distance features is shown in Table 6.

[0134] Table 6

[0135]

[0136] Furthermore, the cross-validation algorithm is used to train the two base classifiers, MLP and RF. The training process is as follows:

[0137] One training subset is used as the validation set, and the other four subsets are used as training sets to be input into the MLP classifier for training. The trained classifier is used to predict the validation set data. The original method is to output the predicted category as the second-layer model;

[0138] Similarly, other training subsets are taken as validation sets in the training set, and the same operation is performed. The corresponding category probability matrix is ​​output, and different training models are obtained through cross-validation. The output vectors of different models are concatenated as the input of the next stage, thereby increasing the generalization ability of the model. At this time, the input data of the second-layer model is n×7, where n is the number of samples in the training set. The MLP model with the highest accuracy is saved for the construction of the integrated model.

[0139] Similarly, RF is used to perform the above two steps to generate the corresponding RF model and category probability matrix. The category probability matrices generated by the two models are concatenated to obtain an n×14 matrix P, where n is the number of samples in the training set and 16 is the number of features of each data. A new training set is constructed using the matrix P and the actual training sample labels. The matrix P is used as input to train LGBM, generate the corresponding model, and finally output the predicted category.

[0140] In terms of feature extraction, the Dlib model is used to accurately locate and obtain the spatial position of facial feature points, and the OpenPose technology is used to accurately capture the human feature points and their spatial coordinates, realizing the multi-dimensional collaborative analysis of facial and human features; in terms of data expression, a complete human skeleton diagram is constructed to transform discrete feature points into a systematic structural expression, and the structural vectors between feature points are generated to provide richer spatial relationship information, and the association mapping between facial features and human postures is established; in terms of monitoring capabilities, it supports real-time tracking of facial expressions and human postures, can dynamically capture state changes and behavioral characteristics, and provide continuous monitoring data streams; in terms of application value, it provides more accurate data support for fatigue status assessment, enhances the ability to identify abnormal behaviors, and improves the accuracy of human reliability analysis. Through multi-dimensional feature extraction and structured expression, this solution significantly improves the accuracy and real-time performance of state recognition of wind power equipment management personnel, provides a more comprehensive and accurate data basis for subsequent fatigue assessment and human reliability analysis, and effectively supports the intelligent upgrade of wind power equipment operation safety management.

[0141] In one example, based on the normalized processing results of fatigue index and concentration, a comprehensive evaluation value is obtained in the form of a synthetic average and determined as a personal work status indicator, including the following steps:

[0142] According to the experimental plan and the determination of eye and mouth fatigue characteristics, the multidimensional fatigue characteristics reflecting the fatigue behavior and process of personnel are analyzed and extracted. By comprehensively considering multiple fatigue characteristics, the mapping relationship between fatigue index is established. Through algorithms such as Dlib and OpenPose, the monitoring video is analyzed and extracted. According to the recognition and analysis of facial features, the actual working state is simulated. Each sample interval is 60 seconds, and the data is saved every 60 seconds and the recording of the next data sample is automatically started. The fatigue data set is established through automatic recording by the program, and the independent variables are the closing time, blinking frequency, yawning time and yawning frequency.

[0143] A multivariate regression method is established on the basis of the univariate regression model to study the relationship between a dependent variable and two or more independent variables and to predict unknown results through the regression model. The fatigue index is predicted using the fatigue data measured experimentally. Considering the accuracy and efficiency of fatigue index prediction, a variety of fatigue characteristics are integrated and a polynomial regression model containing cross terms is used to analyze and fit the quantitative fatigue characteristics obtained in real time. The relationship between each fatigue characteristic and the fatigue index is calculated by the polynomial regression analysis method, which is used to predict the fatigue index of the fatigue data monitored in real time.

[0144] The cross term is to use the relationship between different features as a new feature to participate in model training. The cross term contains the feature set that affects each other. For example, if the input feature is [a, b], the feature set of the second-order polynomial corresponding to the feature is [1, a, b, a 2 ,ab,b 2 】. The cross-term features are used to expand the features of the data set, introduce nonlinearity, and establish direct connections between different features, so as to train the data in a higher dimension and improve the accuracy of the model. The model after introducing multi-scale features is calculated as a multivariate linear regression model. The regression model is shown in the following formula:

[0145] y=w1X1+w2X2+…+w 14 X 14 +w 15 X 15

[0146] Among them, y represents the dependent variable fatigue index, w1, w2, … w 15 represents the regression coefficient, where X1, X2, …X 15 The corresponding indicators are shown in Table 7, where x1 is the quantified value BT corresponding to the eye closing duration, x2 is the quantified value BF corresponding to the blinking frequency, x3 is the quantified value YT corresponding to the yawning duration, and x4 is the quantified value YF corresponding to the yawning frequency.

[0147] Table 7

[0148]

[0149] The third step is to integrate multiple fatigue characteristics and use a polynomial regression model containing cross terms to analyze and fit the quantitative fatigue characteristics. The relationship between each fatigue characteristic and the fatigue index is calculated by the polynomial regression analysis method, which is used to predict the fatigue index of fatigue data.

[0150] The key to the multiple regression equation is to calculate the regression coefficient. This paper uses the least squares method to substitute the experimental data for calculation. The least squares method is a widely used parameter estimation method in the regression model. It has the characteristics of unbiased small and minimum variance. The calculation formula is:

[0151]

[0152] Among them, X represents the matrix corresponding to the independent variable data, and y represents the matrix corresponding to the dependent variable data. Represents the matrix corresponding to the regression coefficients.

[0153] The fourth step is to analyze the rationality of the obtained regression model through goodness of fit and hypothesis testing methods. The calculation formula is:

[0154] SST=SSE+SSR

[0155]

[0156] Among them, SST represents the total deviation sum of squares, SSE represents the residual sum of squares, and SSR represents the regression sum of squares. represents the predicted value of the regression model, represents the sample mean, R 2 Indicates the goodness of fit, ranging from [0,1], R 2 The closer it is to 1, the better the model fit is, and the closer it is to 0, the worse the model fit is.

[0157] The null hypothesis of the regression model is that each regression coefficient has no significant difference from 0. The null hypothesis of the regression model is to indicate that the change of x will not cause a large change of y. The test uses the F statistic, and the calculation formula is:

[0158]

[0159] Among them, n represents the number of samples and k represents the number of independent variables.

[0160] By constructing a task scenario model for the operation of wind power equipment, the operating procedures and environmental factors of operators under different tasks are described in detail. At the same time, fuzzy sets are used to handle uncertainty, and Bayesian networks are used to describe the probabilistic relationship between factors. The combination of the two further improves the robustness and accuracy of the human reliability quantification model. The reliability factors corresponding to the real-time factors and the facial and behavioral characteristics of the operators are obtained respectively, which can comprehensively evaluate the factors affecting human reliability and improve the comprehensiveness of the human reliability evaluation. In addition, the physiological and posture information of the operators can be captured, and the corresponding fatigue index can be obtained to quantify the operator's fatigue level and other working conditions, thereby standardizing and monitoring the behavior of the staff, reducing the probability of human errors, and improving the accuracy of the human error probability value.

[0161] In one example, the membership analysis of CPC factors at different levels is performed based on the evaluation criteria of CPC factors; the analysis results based on CPC factors are used as the prior probability of Bayesian analysis and substituted into the Bayesian network for cognitive control mode analysis; the probability distribution of wind power equipment managers' work scenarios under different control modes is obtained based on Bayesian reasoning, and the defuzzification calculation is performed using the calibration probability method; the CPC factor analysis of the improved basic method is used to predict the error probability using the Bayesian network, and the priority of CPC factor improvement is determined to improve reliability; the quantitative value of each CPC factor effect is calculated based on the membership analysis, and then the corresponding scenario environment index is calculated; based on the calculation of the corresponding cognitive weights of each CPC factor, the total weight of each type of cognitive function is calculated, the probability of human error is predicted, and the quantitative evaluation of human reliability with dynamic characteristics is realized.

[0162] Based on the above method, the embodiment of the present application discloses a human factor reliability quantitative evaluation system based on dynamic characteristics, referring to Figure 4 The human reliability quantitative evaluation system 1 includes a model building module 11, a feature acquisition module 12 and a probability analysis module 13, wherein:

[0163] The model building module 11 is used to build a task scenario model for the operation of wind power equipment, obtain the CPC factor corresponding to any task scenario in the task scenario model, and build a human reliability quantification model based on fuzzy sets, Bayesian networks and CPC factors;

[0164] The feature acquisition module 12 is used to input the real-time factors in the task context into the human reliability quantification model, and obtain the reliability factors corresponding to the real-time factors in the task context and the personal status characteristics of the wind power equipment management personnel corresponding to the task context, wherein the personal status characteristics include facial features and behavioral features;

[0165] The probability analysis module 13 is used to obtain facial feature points and body feature points based on facial features and behavioral features, respectively, and according to the fatigue index corresponding to the facial feature points and the body feature points, input the personal work status index corresponding to the fatigue index into the human reliability quantification model to obtain the probability value of human error.

[0166] In one example, the model building module 11 is used to analyze cognitive activities and cognitive functions in a task context according to the CREAM method, and calculate the basic value of the probability of cognitive function failure in the task context based on the cognitive activities and cognitive functions; construct a fuzzy set corresponding to the control mode according to the control mode in the CREAM method and the probability interval corresponding to the control mode, and construct the uncertain relationship between the CPC factor and the control mode through a Bayesian network; construct a human reliability quantitative model based on the fuzzy set corresponding to the control mode and the uncertain relationship between the CPC factor and the control mode.

[0167] In one example, the model building module 11 is used to decompose the task in the task context into multiple task steps, and determine the cognitive function corresponding to each task step; determine the error category in the cognitive function corresponding to each sub-step in each task step, and calculate the basic error probability corresponding to each error category, wherein each task step includes multiple sub-steps; and calculate the basic value of the probability of cognitive function failure in the task context based on the connection relationship between each sub-step in each task step.

[0168] In one example, the feature acquisition module 12 is used to establish a mapping relationship between real-time factors and reliability factors in a task scenario; match personal status characteristics of wind power equipment managers corresponding to the task scenario according to the reliability factors; and obtain status evaluation values ​​corresponding to the personal status characteristics based on the personal status characteristics and the human work status evaluation function.

[0169] In one example, the probability analysis module 13 is used to obtain facial human feature points and the spatial position points corresponding to the facial feature points based on the facial feature point recognition model and facial features of Dlib; based on the human posture estimation method and behavioral characteristics of OpenPose, feature point detection is performed on the human feature points to obtain the human feature points and the spatial position points corresponding to the human feature points; based on the OpenPose posture estimation algorithm, facial feature points and human feature points, a human skeleton graph corresponding to the facial feature points and the human feature points is constructed, and a structural vector between the facial feature points and the human feature points is constructed according to the human skeleton graph.

[0170] In one example, the personal status characteristics of the wind power equipment management personnel corresponding to the task context are obtained, specifically including:

[0171] Based on the Stacking integrated multi-feature behavior recognition model, the reliability analysis of the human body features collected in real time in the task scenario is performed to obtain facial features and behavioral features respectively.

[0172] In one example, the expression of the human working status evaluation function is:

[0173]

[0174]

[0175] Among them, I c Indicates the real-time working status value, f(action i ) represents the behavior quantization value of the i-th frame, freq represents the device sampling frequency, t represents the sampling duration, I concentration is the result of concentration evaluation, I min Indicates the minimum value of the concentration index, Imin max It represents the difference between the maximum and minimum values ​​of the concentration index, S represents the evaluation result of the individual work status, and I fatigue represents the fatigue index, and w is the corresponding weight of the fatigue index.

[0176] See also Figure 5 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 5 As shown, the electronic device 2 may include: at least one processor 21 , at least one network interface 24 , a user interface 23 , a memory 25 , and at least one communication bus 22 .

[0177] The communication bus 22 is used to realize the connection and communication between these components.

[0178] The user interface 23 may include a display screen (Display) and a camera (Camera), and the optional user interface 23 may also include a standard wired interface and a wireless interface.

[0179] The network interface 24 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0180] Among them, the processor 21 may include one or more processing cores. The processor 21 uses various interfaces and lines to connect various parts in the entire server, and executes various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 25, and calling data stored in the memory 25. Optionally, the processor 21 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 21 can integrate one or more combinations of a central processing unit (Central Processing Unit, CPU), a graphics processor (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 21, and it can be implemented separately through a chip.

[0181] Among them, the memory 25 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 25 includes a non-transitory computer-readable storage medium. The memory 25 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 25 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned method embodiments, etc.; the data storage area may store data involved in the above-mentioned method embodiments, etc. The memory 25 may also be optionally at least one storage device located away from the aforementioned processor 21. As Figure 5 As shown, the memory 25 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a human reliability quantitative evaluation method based on dynamic characteristics.

[0182] exist Figure 5In the electronic device 2 shown, the user interface 23 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 21 can be used to call a human reliability quantitative evaluation method based on dynamic features stored in the memory 25. When executed by one or more processors, the electronic device executes one or more methods in the above-mentioned embodiments.

[0183] A computer-readable storage medium stores instructions, which, when executed by one or more processors, enable the computer to execute one or more methods in the above-mentioned embodiments.

[0184] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for the present application.

[0185] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0186] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0187] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0188] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0189] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned memory includes: various media that can store program codes, such as USB flash drives, mobile hard drives, magnetic disks or optical disks.

[0190] The above are only exemplary embodiments of the present disclosure, and the scope of the present disclosure cannot be limited thereto. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, it will be easy for those skilled in the art to think of other embodiments of the present disclosure. This application is intended to cover any variation, use or adaptive change of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary techniques in the technical field that are not recorded in the present disclosure.

[0191] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A human factor reliability quantitative evaluation method based on dynamic characteristics, characterized in that: The method comprises: Constructing a task scenario model for wind power equipment operation, obtaining a CPC factor corresponding to any task scenario in the task scenario model, and constructing a human reliability quantification model based on fuzzy sets, Bayesian networks and the CPC factor; Input the real-time factors in the task scenario into the human reliability quantification model, and respectively obtain the reliability factors corresponding to the real-time factors in the task scenario and the personal status characteristics of the wind power equipment management personnel corresponding to the task scenario, wherein the personal status characteristics include facial features and behavioral features; Based on the facial features and behavioral features, facial feature points and body feature points are obtained respectively, and according to the fatigue index corresponding to the facial feature points and the body feature points, the personal work status index corresponding to the fatigue index is input into the human reliability quantification model to obtain the human error probability value.

2. A method for quantitatively evaluating human reliability based on dynamic characteristics as claimed in claim 1, characterized in that: The human factor reliability quantitative model is constructed based on fuzzy sets, Bayesian networks and CPC factors, specifically including: Analyzing cognitive activities and cognitive functions in the task context according to the CREAM method, and calculating a basic value of the probability of cognitive function failure in the task context based on the cognitive activities and the cognitive functions; According to the control mode in the CREAM method and the probability interval corresponding to the control mode, a fuzzy set corresponding to the control mode is constructed, and the uncertain relationship between the CPC factor and the control mode is constructed through a Bayesian network; Based on the fuzzy set corresponding to the control mode and the uncertain relationship between the CPC factor and the control mode, a human reliability quantitative model is constructed.

3. A method for quantitatively evaluating human reliability based on dynamic characteristics as claimed in claim 2, characterized in that: The step of analyzing the cognitive activities and cognitive functions in the task context according to the CREAM method, and calculating the basic value of the probability of cognitive function failure in the task context based on the cognitive activities and the cognitive functions, specifically includes: Decomposing the task in the task situation into a plurality of task steps, and determining the cognitive function corresponding to each task step; Determine the error categories in cognitive functions corresponding to each sub-step in each task step, and calculate the basic error probability corresponding to each error category, wherein each task step includes multiple sub-steps; According to the connection relationship between each sub-step in each task step, the basic value of the probability of cognitive function failure in the task scenario is calculated.

4. The method for quantitatively evaluating human reliability based on dynamic characteristics as claimed in claim 1, characterized in that: The obtaining of the reliability factor corresponding to the real-time factor in the task scenario and the personal status characteristics of the wind power equipment management personnel corresponding to the task scenario specifically includes: Establishing a mapping relationship between the real-time factors and the reliability factors in the task scenario; Matching personal status characteristics of wind power equipment management personnel corresponding to the task scenario according to the reliability factor; Based on the personal status feature and the human work status evaluation function, a status evaluation value corresponding to the personal status feature is obtained.

5. The method for quantitatively evaluating human reliability based on dynamic characteristics as claimed in claim 1, characterized in that: Based on the facial features and behavioral characteristics, Obtain facial feature points and body feature points respectively, including: Based on the facial feature point recognition model of Dlib and the facial features, obtain the facial body feature points and the spatial position points corresponding to the facial feature points; Based on the human posture estimation method of OpenPose and the behavioral characteristics, feature point detection is performed on the human feature points to obtain the human feature points and the spatial position points corresponding to the human feature points; Based on the OpenPose posture estimation algorithm, the facial feature points and the human body feature points, a human skeleton graph corresponding to the facial feature points and the human body feature points is constructed, and a structural vector between the facial feature points and the human body feature points is constructed according to the human skeleton graph.

6. The method for quantitatively evaluating human reliability based on dynamic characteristics as claimed in claim 1, characterized in that: The personal status characteristics of the wind power equipment management personnel corresponding to the task scenario are obtained, specifically including: Based on the Stacking integrated multi-feature behavior recognition model, the reliability analysis is performed on the human body features collected in real time in the task scenario to obtain facial features and behavior features respectively.

7. A method for quantitatively evaluating human reliability based on dynamic characteristics as claimed in claim 4, characterized in that: The expression of the human working status evaluation function is: Among them, I c Indicates the real-time working status value, f(action i ) represents the behavior quantization value of the i-th frame, freq represents the device sampling frequency, t represents the sampling duration, I concentration is the result of concentration evaluation, I min Indicates the minimum value of the concentration index, Imin max It represents the difference between the maximum and minimum values ​​of the concentration index, S represents the evaluation result of the individual work status, and I fatigue represents the fatigue index, and w is the corresponding weight of the fatigue index.

8. A human factor reliability quantitative evaluation system based on dynamic characteristics, characterized in that: The human reliability quantitative evaluation system (1) comprises a model building module (11), a feature collection module (12) and a probability analysis module (13), wherein: The model building module (11) is used to build a task scenario model for the operation of wind power equipment, obtain a CPC factor corresponding to any task scenario in the task scenario model, and build a human reliability quantification model based on fuzzy sets, Bayesian networks and the CPC factor; The feature acquisition module (12) is used to input the real-time factors in the task scenario into the human factor reliability quantification model, and respectively obtain the reliability factors corresponding to the real-time factors in the task scenario and the personal status characteristics of the wind power equipment management personnel corresponding to the task scenario, wherein the personal status characteristics include facial features and behavioral features; The probability analysis module (13) is used to obtain facial feature points and human feature points respectively based on the facial features and behavioral features, and according to the fatigue index corresponding to the facial feature points and the human feature points, input the personal work status index corresponding to the fatigue index into the human reliability quantification model to obtain the human error probability value.

9. An electronic device, characterized in that: The electronic device (2) comprises a processor (21), a memory (25), a user interface (23) and a network interface (24), wherein the memory (25) is used to store instructions, the user interface (23) and the network interface (24) are used to communicate with other devices, and the processor (21) is used to execute the instructions stored in the memory (25) so that the electronic device (2) executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images