Operation plan risk assessment method and device in combination with personnel ability, terminal equipment and storage medium
By combining work plan data and multi-dimensional personnel data, and using a risk assessment model to identify risk labels and determine dynamic levels, the problem of risk prediction being out of sync with actual on-site conditions due to neglecting personnel capabilities in existing technologies has been solved, thus achieving accurate risk assessment and early warning.
Patent Information
- Application Number
- CN202511706903.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-20
- Publication Date
- 2026-03-06
AI Technical Summary
The existing risk assessment methods for work plans ignore the dynamic differences in personnel capabilities, resulting in a disconnect between risk prediction and actual on-site conditions.
By acquiring work plan data and multidimensional personnel data, and using a trained risk assessment model, combined with personnel capabilities, risk labels are identified and key risk points are judged, risk levels are dynamically determined, and early warning signals are issued when necessary.
It enables accurate assessment of risks in work plans, ensuring that the risk level matches the actual situation on site, and avoiding the problem of risk prediction being out of sync with reality.
Smart Images
Figure CN121615917A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of risk assessment, and more particularly to a method, apparatus, terminal equipment, and storage medium for risk assessment of work plans that incorporates personnel capabilities. Background Technology
[0002] Current risk assessments of work plans often focus on evaluating objective factors such as the work environment, equipment condition, and work process. These factors are used as the main basis for risk assessment, and a series of fixed indicators and models are set to calculate the risk level.
[0003] However, this risk assessment method ignores the dynamic differences in personnel capabilities, resulting in the risk level failing to accurately reflect the actual risk level during individual operations, causing risk prediction to become disconnected from the actual situation on site. Summary of the Invention
[0004] This invention provides a method, apparatus, terminal device, and storage medium for risk assessment of work plans that incorporates personnel capabilities. This addresses the problem that existing risk assessment methods ignore dynamic differences in personnel capabilities, leading to a disconnect between risk prediction and actual on-site conditions, and achieves accurate risk assessment.
[0005] One embodiment of the present invention provides a method for risk assessment of work plans that incorporates personnel capabilities, comprising:
[0006] Obtain the risk level values of the work plan data in the work plan to be evaluated and the multi-dimensional personnel data of the staff;
[0007] Input the risk level value and multidimensional personnel data into the trained risk assessment model so that the risk assessment model can determine the risk label of the work plan to be assessed based on the risk level value and multidimensional personnel data;
[0008] Based on the risk labels, identify several key risk points corresponding to the work plan to be evaluated;
[0009] Determine whether all key risk points exist in a scenario consistent with reality. If so, if the total number of key risk points falls within the first preset range, mark the risk level of the work plan to be evaluated as Level 1 risk; if the total number of key risk points falls within the second preset range, mark the risk level of the work plan to be evaluated as Level 2 risk; if the total number of key risk points falls within the third preset range, mark the risk level of the work plan to be evaluated as Level 3 risk. Otherwise, mark the risk level of the work plan to be evaluated as Level 3 risk. The severity of the work plan risk increases in the order of Level 1, Level 2, and Level 3 risk. The first preset range is smaller than the second preset range, and the second preset range is smaller than the third preset range.
[0010] A warning signal will be issued when the risk level is Level 2 or Level 3.
[0011] Furthermore, after obtaining the risk level values of the work plan data in the work plan to be evaluated and the multidimensional personnel data of the staff, it also includes:
[0012] After filling in missing values and correcting outliers in the multidimensional personnel data, the preprocessed multidimensional personnel data is obtained after manual review and confirmation.
[0013] Furthermore, after obtaining the preprocessed multidimensional personnel data, it also includes:
[0014] The preprocessed multidimensional personnel data is divided into qualitative indicators and primary quantitative indicators;
[0015] Convert qualitative indicators into secondary quantitative indicators;
[0016] The first and second quantitative indicators are normalized to obtain normalized quantitative indicators;
[0017] Identify extreme and non-extreme values in normalized quantitative indicators;
[0018] Perform a logarithmic transformation on each extreme value to obtain the corresponding transformed data;
[0019] By integrating non-extreme values and all transformed data, standardized multidimensional personnel data is obtained.
[0020] Furthermore, the transformed data includes: first transformed data and second transformed data;
[0021] Perform a logarithmic transformation on each extreme value to obtain the corresponding transformed data, including:
[0022] For each extreme value, determine whether the extreme value is greater than a preset threshold;
[0023] If so, take the absolute value of the extreme values and calculate the natural logarithm to obtain the first transformed data;
[0024] Otherwise, the natural logarithm is calculated on the sum of the absolute values of the extreme values and the preset constant factors to obtain the second transformed data.
[0025] Furthermore, the preset constant factor is determined in the following way:
[0026] Extreme values exceeding a preset threshold are considered as out-of-limit data points;
[0027] Calculate the average of the first derivatives of all out-of-limit data points;
[0028] Extreme values that are not greater than a preset threshold are taken as data points that do not exceed the limit;
[0029] Based on the data points that do not exceed the limits and the constant factors to be confirmed, construct the first derivative function relationship;
[0030] Determine the upper and lower limits of the objective inequality based on the average value of the first derivative function;
[0031] Construct the objective inequality based on the first derivative function relationship, upper limit value, and lower limit value;
[0032] The integer with the largest constant factor in the first-order derivative relation that satisfies the objective inequality is used as the preset constant factor.
[0033] Furthermore, the risk assessment model is obtained in the following way:
[0034] Acquire several training data sets; each training data set includes: the risk level value of the work plan sample data in a work plan sample, the multidimensional personnel sample data of the staff and their corresponding real risk labels;
[0035] Several training data are input into the risk assessment model to be trained, so that the risk assessment model takes the risk level value of the work plan sample data and the multidimensional personnel sample data of the staff as output, and the predicted risk label of the work plan sample as output for iterative training. During the training process, the loss function value is calculated based on the predicted risk label and the actual risk label, and the model parameters of the risk assessment model are adjusted according to the loss function value until the loss function value converges, and the trained risk assessment model is obtained.
[0036] Based on the above method embodiments, the present invention provides corresponding device embodiments, including: an operation data acquisition module, a risk label identification module, a key risk point identification module, a risk level dynamic determination module, and an early warning module;
[0037] The work data acquisition module is used to acquire the risk level value of the work plan data in the work plan to be evaluated and the multi-dimensional personnel data of the staff;
[0038] The risk label recognition module is used to input risk level values and multidimensional personnel data into a trained risk assessment model, so that the risk assessment model can determine the risk label of the work plan to be assessed based on the risk level values and multidimensional personnel data.
[0039] The key risk point identification module is used to identify several key risk points corresponding to the work plan to be evaluated based on risk labels;
[0040] The dynamic risk level determination module is used to determine whether all key risk points exist in accordance with the actual scenario. If so, the risk level of the work plan to be evaluated is marked as Level 1 risk if the total number of key risk points is within the first preset range, Level 2 risk if the total number of key risk points is within the second preset range, and Level 3 risk if the total number of key risk points is within the third preset range. Otherwise, the risk level of the work plan to be evaluated is marked as Level 3 risk. The severity of the work plan risk increases in the order of Level 1, Level 2, and Level 3 risk. The first preset range is smaller than the second preset range, and the second preset range is smaller than the third preset range.
[0041] The early warning module is used to issue early warning signals when the risk level is level two or level three.
[0042] Furthermore, the work plan risk assessment device that incorporates personnel capabilities also includes: a work data preprocessing module;
[0043] The task data preprocessing module is used to fill in missing values and correct outliers in the multidimensional personnel data. After manual review and confirmation, the preprocessed multidimensional personnel data is obtained.
[0044] Based on the above method embodiments, the present invention provides a corresponding terminal device embodiment, including: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the steps of the work plan risk assessment method combining personnel capabilities as described in the present invention.
[0045] Based on the above method embodiments, the present invention provides a corresponding computer-readable storage medium embodiment, including: a stored computer program that, when the computer program is running, controls the device where the computer-readable storage medium is located to perform the steps of the work plan risk assessment method combining personnel capabilities as described in the present invention.
[0046] Compared with the prior art, the beneficial effects of this embodiment are as follows:
[0047] This invention acquires the risk level values of the work plan data and multi-dimensional personnel data of the staff in the work plan to be evaluated. The risk level values and multi-dimensional personnel data are input into a trained risk assessment model to determine the risk label of the work plan to be evaluated. This effectively combines the objective risks of the work itself with the dynamic differences in personnel capabilities, overcoming the shortcomings of traditional assessments that only focus on the work level and ignore personnel factors. Further, based on the risk labels, several key risk points corresponding to the work plan to be evaluated are identified. It is determined whether all key risk points exist in a scenario that matches the actual situation, thus transforming the abstract risk labels into concrete and perceptible risk scenarios. If the total number of key risk points is within a first preset range, the risk level of the work plan to be evaluated is marked as Level 1 risk; if the total number of key risk points is within a second preset range, it is marked as Level 2 risk; if the total number of key risk points is within a third preset range, it is marked as Level 3 risk; otherwise, the risk level of the work plan to be evaluated is marked as Level 3 risk, avoiding the problem of risk prediction being out of sync with the actual situation on site. Finally, a warning signal is issued when the risk level is Level 2 or Level 3 risk.
[0048] In summary, this invention solves the problem of existing risk assessment methods ignoring dynamic differences in personnel capabilities, leading to a disconnect between risk prediction and actual on-site conditions, by integrating operational risk data with multidimensional personnel data, analyzing the data through a model, and calibrating risk levels on-site. This achieves accurate risk assessment. Attached Figure Description
[0049] Figure 1 This is a flowchart illustrating a work plan risk assessment method that incorporates personnel capabilities, provided in an embodiment of the present invention.
[0050] Figure 2 This is a schematic diagram of the structure of a work plan risk assessment device that combines personnel capabilities, provided in an embodiment of the present invention. Detailed Implementation
[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0052] In the description of this invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated.
[0053] like Figure 1As shown, in order to address the problem that existing risk assessment methods ignore the dynamic differences in personnel capabilities, leading to a disconnect between risk prediction and actual on-site conditions, an embodiment of the present invention provides a work plan risk assessment method that incorporates personnel capabilities. This method includes at least the following steps:
[0054] Step S1: Obtain the risk level values of the work plan data and the multidimensional personnel data of the staff in the work plan to be evaluated;
[0055] For step S1, firstly, collect the work plan data in the work plan to be evaluated and the multi-dimensional personnel data of each worker.
[0056] The above-mentioned work plan data includes: the number of major violations, the number of general violations, the safety operation compliance rate, the number of safety hazard reports, the special operation certificate holding rate of the staff participating in the current work plan, the emergency drill participation rate, and the safety skills assessment score.
[0057] The aforementioned multidimensional personnel data includes: basic personnel attribute data, safety behavior data, qualification and competence data, and historical performance data. Basic personnel attribute data consists of tagged information about the workers, including: name, employee ID, department / team, job title, date of employment, age, job type, and certification status. Safety behavior data is a direct record of the workers' dynamic performance, including: the number of safety inspection violations in the past year, violation types, violation rectification completion rate, and the duration of safety training participation in the past six months. Qualification and competence data is the core basis for determining whether workers are competent for their jobs, including: internal skill ratings, external certification levels, and skill competition awards. Historical performance data is a quantitative basis for assessing personnel risk inertia, including: the completion rate of work tasks in the past year, quality pass rate, number of rework / repairs in the past year, number of near misses in the past three years, minor accidents, and determination of responsibility for major accidents.
[0058] It should be noted that the aforementioned staff specifically refers to the personnel directly involved in the execution of the work plan to be evaluated.
[0059] Obtain the risk level value X from the work plan data. b b = 1, 2, ..., 7. Specifically, the risk level values include: the risk level value of the number of major violations in the past year X1, the risk level value of the number of general violations X2, the risk level value of the safe operation compliance rate X3, the risk level value of the number of safety hazard reports X4, the risk level value of the special operation certificate holding rate of the staff participating in the current operation plan X5, the risk level value of the emergency drill participation rate X6, and the risk level value of the safety skills assessment score X7.
[0060] Based on the risk level value X in the work plan data b Construct the basic matrix
[0061] It should be noted that the risk level values are assigned based on their importance. The higher the risk level value, the greater the importance of the indicator data, and vice versa.
[0062] This invention collects four dimensions of data on personnel's basic attributes, safety behaviors, qualifications and capabilities, and historical performance, while also linking them to the risk level value of the work plan. This multi-dimensional data covers all elements and avoids the one-sidedness of a single indicator.
[0063] In a preferred embodiment, after obtaining the risk level values of the work plan data in the work plan to be evaluated and the multidimensional personnel data of the staff, the method further includes:
[0064] After filling in missing values and correcting outliers in the multidimensional personnel data, the preprocessed multidimensional personnel data is obtained after manual review and confirmation.
[0065] In one embodiment of the present invention, for multidimensional personnel data with missing values, the missing values are filled with the mean of the corresponding operational personnel data, outliers are corrected by natural logarithmic transformation, and confirmed by manual review, so as to finally obtain the preprocessed multidimensional personnel data.
[0066] For example, a certain round of maintenance workshops needs to standardize the "safety credit score" of operators, which involves multi-dimensional data. Among them, "number of major violations in the past year" is a key indicator, which has missing values and outliers. The missing values are filled and the outliers are corrected in the following ways:
[0067] First, input the number of major violations in the past year for 5 workers: [0, 1, NaN, 3, 2], where NaN represents missing values;
[0068] Calculate the mean: The mean of non-missing values is (0+1+3+2) / 4 = 1.5;
[0069] Fill missing values: Fill the missing value of the 3rd person with 1.5;
[0070] Manual review: Confirmed that the third person has no recent major violations and the information provided is appropriate;
[0071] Output data: [0, 1, 1.5, 3, 2].
[0072] Secondly, regarding the handling of outliers, it is important to understand that outliers can significantly distort the statistical characteristics of data, such as the mean, variance, and skewness. After logarithmic transformation, outliers are compressed, the mean and variance are closer to the true level, and the data distribution tends to be symmetrical.
[0073] Specifically, data exceeding twice the average number of major violations by employees in the same position within the past year is considered outlier. In this example, the average is 1.5, so values exceeding 3 are outliers. In the example above, the data after missing value imputation is: [0, 1, 1.5, 3, 2], with no outliers exceeding 3. Assuming the adjusted data is [0, 1, 1.5, 5, 2], where 5 > 3, it is an outlier. At this point:
[0074] Identifying extreme values: The fourth person's "number of major violations in the past year" is 5, which is an extreme value;
[0075] Logarithmic transformation: Using the natural logarithm transformation, f(x) = ln(x), the transformed value is ln(5)≈1.609;
[0076] Standardization complete: Record the transformed value as 1.609, marking the data as standardized.
[0077] Output data: [0, 1, 1.5, 1.609, 2]. At this point, the data for the 4th person has changed.
[0078] Through the above operations of missing value imputation, outlier correction, and manual review, the multidimensional personnel data is transformed from a state of missing and outlier data into complete and accurate high-quality data, avoiding the loss of analysis dimensions due to missing values and the deviation of analysis results due to outliers.
[0079] In a preferred embodiment, after obtaining the preprocessed multidimensional personnel data, the method further includes:
[0080] The preprocessed multidimensional personnel data is divided into qualitative indicators and primary quantitative indicators;
[0081] Convert qualitative indicators into secondary quantitative indicators;
[0082] The first and second quantitative indicators are normalized to obtain normalized quantitative indicators;
[0083] Identify extreme and non-extreme values in normalized quantitative indicators;
[0084] Perform a logarithmic transformation on each extreme value to obtain the corresponding transformed data;
[0085] By integrating non-extreme values and all transformed data, standardized multidimensional personnel data is obtained.
[0086] In a preferred embodiment, the transformed data includes: first transformed data and second transformed data;
[0087] Perform a logarithmic transformation on each extreme value to obtain the corresponding transformed data, including:
[0088] For each extreme value, determine whether the extreme value is greater than a preset threshold;
[0089] If so, take the absolute value of the extreme values and calculate the natural logarithm to obtain the first transformed data;
[0090] Otherwise, the natural logarithm is calculated on the sum of the absolute values of the extreme values and the preset constant factors to obtain the second transformed data.
[0091] In one embodiment of the present invention, the preprocessed multidimensional personnel data is divided into qualitative indicators and a first quantitative indicator. Specifically, the qualitative indicators include: violation type (including: operational / protective / management), internal skill rating (including: Level 1 technician / Level 2 technician / Level 3 technician), external certification level (including: Ministry of Human Resources and Social Security vocational skill level), skill competition award records, minor accident situation (including: minor injury / equipment damage), and determination of responsibility for major accidents. The first quantitative indicator includes: the number of safety inspection violations in the past year (including: the number of times safety helmets were not worn and the number of times seat belts were not worn), the violation rectification completion rate, the duration of safety training participation in the past six months, the work task completion rate in the past year (e.g., on-time completion rate ≥ 95%), the quality pass rate (e.g., the first-time inspection pass rate ≥ 98%), the number of rework / repairs, and the number of near misses in the past three years (e.g., the number of operational errors that almost caused accidents).
[0092] It should be noted that in multidimensional personnel data, since the basic attribute data of personnel is the labeled information of the workers, dividing the multidimensional personnel data into qualitative and quantitative indicators is actually dividing the safety behavior data, qualification and ability data and historical performance data into qualitative and quantitative indicators.
[0093] Next, the qualitative indicators are converted into a second quantitative indicator, for example:
[0094] Input qualitative indicators: "Internal skills rating" (divided into three levels: "Level 1 Technician", "Level 2 Technician", and "Level 3 Technician");
[0095] Conversion rules: Level 1 technician = 5 points; Level 2 technician = 3 points; Level 3 technician = 1 point;
[0096] The 5 workers: ["Level 1 Technician", "Level 2 Technician", "Level 3 Technician", "Level 1 Technician", "Level 2 Technician"], are converted to [5, 3, 1, 5, 3].
[0097] By classifying multidimensional personnel data as described above and further quantifying qualitative indicators, all the originally disparate multidimensional personnel data are transformed into quantifiable numerical forms, providing a unified data dimension foundation.
[0098] To eliminate the difference in magnitude, the first and second quantitative indicators are normalized, mapping quantitative indicators with different dimensions and value ranges to a specific interval to obtain normalized quantitative indicators.
[0099] Set an extreme threshold, compare the normalized quantitative index with the set extreme threshold, and determine the extreme value that exceeds the threshold range as an abnormal deviation from the main distribution, and the value that is within the threshold range as a non-extreme value within the normal fluctuation range.
[0100] For identified extreme values, since these extreme values may mask group patterns and distort statistical indicators, a nonlinear transformation using the natural logarithm function is employed. The natural logarithm function has the characteristics of compressing large values and preserving small values, which can mitigate the distorting effect of extreme values on data analysis while retaining the relative relationships between data, making the distribution shape closer to a normal distribution. Specifically:
[0101] Let the extreme value be denoted as G. h The preset threshold is set to 1;
[0102] When G h When the value is greater than 1, the natural logarithm is calculated by taking the absolute value of the extreme values, i.e., ln|G h |, obtain the first transformed data; when G h When ≤1, since the extreme values are all greater than 0 in the scenario of this invention, the extreme values are not greater than the preset threshold, that is, 0 < G. h ≤1, in this case, calculate the natural logarithm of the sum of the absolute value of the extreme value and the preset constant factor c, i.e., ln(|G h |+c), to obtain the second transformed data, the overall formula is as follows:
[0103]
[0104] Among them, P h Let h represent the h-th transformed data, where h is the index of the extreme value, h = 1, 2, ..., H, where H represents the total number of extreme values, and c represents the preset constant factor, and c ∈ [1, +∞].
[0105] Finally, the non-extreme values and all transformed data are integrated to obtain standardized multidimensional personnel data, forming a complete dataset, ensuring that all data retains the original semantics while meeting the normative requirements of numerical distribution.
[0106] In this embodiment, the preset constant factor can be a pre-set value, or it can be determined in the following ways:
[0107] Extreme values exceeding a preset threshold are considered as out-of-limit data points;
[0108] Calculate the average of the first derivatives of all out-of-limit data points;
[0109] Extreme values that are not greater than a preset threshold are taken as data points that do not exceed the limit;
[0110] Based on the data points that do not exceed the limits and the constant factors to be confirmed, construct the first derivative function relationship;
[0111] Determine the upper and lower limits of the objective inequality based on the average value of the first derivative function;
[0112] Construct the objective inequality based on the first derivative function relationship, upper limit value, and lower limit value;
[0113] The integer with the largest constant factor in the first-order derivative relation that satisfies the objective inequality is used as the preset constant factor.
[0114] Specifically, extreme values exceeding a preset threshold are defined as out-of-limit data points G. m Extreme values that are not greater than a preset threshold are defined as data points G that have not exceeded the limit. n Where m is the index of the data point that exceeds the limit, and m = 1, 2, ..., M, M is the total number of data points that exceed the limit, and n is the index of the data point that does not exceed the limit, and n = 1, 2, ..., N, N is the total number of data points that do not exceed the limit, satisfying M + N = H.
[0115] For the m-th out-of-limit data point G m Its transformation function is:
[0116] P m =ln|G m |;
[0117] Among them, P m This represents the transformed data for the m-th out-of-limit data point.
[0118] Calculate P m The first derivative function is obtained by finding the value of the first derivative function P′. m Calculate all first derivative values P′ m The average value of the first derivative function is obtained by averaging the values of the first derivative function.
[0119] For the nth data point G that does not exceed the limit n Its transformation function is:
[0120]
[0121] Among them, P n This represents the transformed data of the nth data point that did not exceed the limit. This represents a constant factor that needs to be confirmed, and there are several options.
[0122] For each possible value P can be calculated nThe corresponding first derivative value P n ′ .
[0123] Construct the objective inequality:
[0124]
[0125] The first derivative relation P that satisfies the objective inequality n ′ The integer of the largest constant factor is used as the preset constant factor to balance the compression intensity and the preservation of data details.
[0126] By dynamically determining the constant factor as described above, the distribution characteristics of different datasets can be adapted to avoid undercompression or overcompression problems caused by fixed values.
[0127] For example:
[0128] Taking the transformed value of "internal skills rating" as an example:
[0129] The original data: "Internal skill rating" was converted into quantitative values [5, 3, 1, 5, 3], where the mean was (5+3+1+5+3) / 5 = 3.4;
[0130] At this point, all values are >1, and all values are out-of-limit data points. Take the m-th out-of-limit data point. At this point, m = 1, 2, 3, 4, 5.
[0131] Calculate the first derivative: Based on the m-th data point, for the transformation function P m =ln|G m |, by taking the first derivative, we obtain the corresponding first derivative function value P′. m ;
[0132] Taking the transformed value of "completion rate of rectification of violations" as an example:
[0133] The original data is: "Completion rate of rectification of violations" is [85%, 90%, 95%, 100%, 50%], where the mean is (85%+90%+95%+100%+50%) / 5 = 84%. Therefore, the extreme value is defined as ≤84%. Only 50%≤84% is taken as the nth data point that does not exceed the limit. At this time, n = 1, 2, 3, 4.
[0134] Calculate the first derivative: Based on the nth data point, perform a transformation on the function... Taking the first derivative, we obtain the corresponding first derivative value P. n ′ .
[0135] Step S2: Input the risk level value and multidimensional personnel data into the trained risk assessment model so that the risk assessment model can determine the risk label of the work plan to be assessed based on the risk level value and multidimensional personnel data;
[0136] For step S2, the risk level value and multidimensional personnel data obtained in step S1 are input into the trained risk assessment model. The pre-trained risk assessment model has established a complex mapping relationship between risk level, personnel characteristics and actual risk results through learning from historical data. Through the analysis and calculation within the model, a risk label for the work plan to be assessed is finally generated.
[0137] Preferably, the risk label can be a binary classification or a multi-level classification. The specific classification dimensions can be set according to the actual requirements of safety management and operation scenarios, and the model training can be completed based on the corresponding labeled data.
[0138] In a preferred embodiment, the risk assessment model is obtained in the following manner:
[0139] Acquire several training data sets; each training data set includes: the risk level value of the work plan sample data in a work plan sample, the multidimensional personnel sample data of the staff and their corresponding real risk labels;
[0140] Several training data are input into the risk assessment model to be trained, so that the risk assessment model takes the risk level value of the work plan sample data and the multidimensional personnel sample data of the staff as output, and the predicted risk label of the work plan sample as output for iterative training. During the training process, the loss function value is calculated based on the predicted risk label and the actual risk label, and the model parameters of the risk assessment model are adjusted according to the loss function value until the loss function value converges, and the trained risk assessment model is obtained.
[0141] In one embodiment of the present invention, firstly, a large amount of labeled training data is collected. Each training data includes the risk level value of the work plan sample data in the work plan sample, the multidimensional personnel sample data of the staff, and the real risk label determined by experience or historical results.
[0142] Subsequently, these training data are input into the risk assessment model to be trained. During the model training phase, a nonlinear mapping relationship from job risk attributes and personnel characteristics to risk labels is established through a neural network. During the training process, the model generates predicted risk labels based on the current parameters and calculates the loss function, such as cross-entropy loss or mean squared error, by comparing it with the real labels. Then, the backpropagation algorithm and optimizer (such as Adam) are used to dynamically adjust the network weights and gradually minimize the prediction error.
[0143] Continue iterative optimization until the loss function converges or the number of iterations meets the preset number of iterations, and finally generate a trained risk assessment model.
[0144] Step S3: Identify several key risk points corresponding to the work plan to be evaluated based on the risk labels;
[0145] For step S3, based on the specific type of the risk label, the preset key risk point mapping data table in the work plan to be evaluated is called, and several corresponding key risk points are determined by matching the risk label with the corresponding entries in the data table.
[0146] It should be noted that, due to the different work scenarios, operating procedures, safety control priorities, and personnel capability requirements of different work plans, the preset key risk point mapping data tables contained in different work plans are different. Therefore, even for the same risk label, the corresponding key risk points may be different in different work plans.
[0147] Example: A wheelset repair workshop needs to conduct a risk assessment for a "high-altitude wheelset disassembly operation plan," and the identified key risk points are R1-R3:
[0148] R1: Equipment aging (used for more than 15 years);
[0149] R2: The workers do not have a high-altitude work permit;
[0150] R3: No isolation in cross-operation areas.
[0151] Step S4: Determine whether all key risk points exist and match the actual scenario. If so, if the total number of key risk points is within the first preset range, mark the risk level of the work plan to be evaluated as Level 1 risk; if the total number of key risk points is within the second preset range, mark the risk level of the work plan to be evaluated as Level 2 risk; if the total number of key risk points is within the third preset range, mark the risk level of the work plan to be evaluated as Level 3 risk. Otherwise, mark the risk level of the work plan to be evaluated as Level 3 risk. The severity of the work plan risk increases in the order of Level 1 risk, Level 2 risk, and Level 3 risk. The first preset range is smaller than the second preset range, and the second preset range is smaller than the third preset range.
[0152] For step S4, to ensure that the identification results of key risk points fully match the actual risk situation at the work site and to guarantee the accuracy and reliability of risk level determination, each key risk point is manually checked to determine whether it meets the actual scenario. Based on this, and combined with the preset range to which the total number of key risk points belongs, the risk level of the work plan is graded and marked. Specifically, if all key risk points meet the actual scenario and the total number of key risk points is within the first preset range, it is marked as Level 1 risk, i.e., the lowest risk severity; if it is within the second preset range, it is marked as Level 2 risk; if it is within the third preset range, it is marked as Level 3 risk, i.e., the highest risk severity; if there are key risk points that do not meet the actual scenario, they are all marked as Level 3 risk.
[0153] For example:
[0154] R1 verification: If the crane has been in use for 16 years (more than 15 years) and this is consistent with the preset description, then it conforms to the actual scenario.
[0155] R2 verification: If the person in charge of the operation holds a valid high-altitude operation certificate, which is inconsistent with the preset description of "no certificate", then it does not conform to the actual scenario.
[0156] R3 verification: A spacing of 4 meters (<5 meters) is consistent with the preset description "no isolation", which is in line with the actual scenario;
[0157] The first preset range is set to [0,1], the second preset range is set to (1,2], and the third preset range is set to (2,+∞).
[0158] Level 1 Risk: The number of key risk points is [0,1], all of which are consistent with actual scenarios;
[0159] Level 2 Risk: The number of key risk points is (1,2], which is consistent with the actual scenario;
[0160] Level 3 risk: The number of key risk points is (2, +∞), all of which conform to the actual scenario, or any risk point does not conform to the actual scenario (e.g., the equipment is not aged but is marked as aged).
[0161] After manually verifying each of the key risk points, it was confirmed that risk points R1 and R3 (2 in total) conformed to the actual scenario, while R2 (1) did not. Therefore, the key risk level of this work plan was marked as Level 3 risk.
[0162] Step S5: Issue an early warning signal when the risk level is Level 2 or Level 3.
[0163] For step S5, if the risk level is level 2 or level 3, a warning signal is issued. In this embodiment, level 1 risk is a blue signal, level 2 risk is an orange signal, and level 3 risk is a red signal.
[0164] like Figure 2 As shown, based on the above method embodiments, corresponding apparatus embodiments are provided;
[0165] One embodiment of the present invention provides a work plan risk assessment device that combines personnel capabilities, including: a work data acquisition module, a risk label identification module, a key risk point identification module, a risk level dynamic determination module, and an early warning module;
[0166] The work data acquisition module is used to acquire the risk level value of the work plan data in the work plan to be evaluated and the multi-dimensional personnel data of the staff;
[0167] The risk label recognition module is used to input risk level values and multidimensional personnel data into a trained risk assessment model, so that the risk assessment model can determine the risk label of the work plan to be assessed based on the risk level values and multidimensional personnel data.
[0168] The key risk point identification module is used to identify several key risk points corresponding to the work plan to be evaluated based on risk labels;
[0169] The dynamic risk level determination module is used to determine whether all key risk points exist in accordance with the actual scenario. If so, the risk level of the work plan to be evaluated is marked as Level 1 risk if the total number of key risk points is within the first preset range, Level 2 risk if the total number of key risk points is within the second preset range, and Level 3 risk if the total number of key risk points is within the third preset range. Otherwise, the risk level of the work plan to be evaluated is marked as Level 3 risk. The severity of the work plan risk increases in the order of Level 1, Level 2, and Level 3 risk. The first preset range is smaller than the second preset range, and the second preset range is smaller than the third preset range.
[0170] The early warning module is used to issue early warning signals when the risk level is level two or level three.
[0171] In a preferred embodiment, the work plan risk assessment device that incorporates personnel capabilities further includes: a work data preprocessing module;
[0172] The task data preprocessing module is used to fill in missing values and correct outliers in the multidimensional personnel data. After manual review and confirmation, the preprocessed multidimensional personnel data is obtained.
[0173] It is understood that the above-described device embodiments correspond to the method embodiments of the present invention, and can implement the work plan risk assessment method combining personnel capabilities provided by any of the above-described method embodiments of the present invention.
[0174] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can specifically be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.
[0175] Based on the above embodiments of the work plan risk assessment method combining personnel capabilities, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the work plan risk assessment method combining personnel capabilities according to any embodiment of the present invention.
[0176] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the terminal device.
[0177] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.
[0178] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting various parts of the terminal device via various interfaces and lines.
[0179] Based on the above-described method embodiments, another embodiment is provided: another embodiment of the present invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the work plan risk assessment method combining personnel capabilities as described in any of the above-described method embodiments of the present invention.
[0180] The modules / units integrated into the work plan risk assessment device / terminal equipment that combine personnel capabilities, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.
[0181] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.
Claims
1. A work plan risk assessment method that combines human capabilities, characterized by, The method comprises the following steps: obtaining a risk level value of job plan data in a job plan to be evaluated and multidimensional personnel data of a worker; inputting the risk level value and the multidimensional personnel data into a trained risk assessment model to enable the risk assessment model to determine a risk label of the job plan to be evaluated according to the risk level value and the multidimensional personnel data; identifying a plurality of key risk points corresponding to the job plan to be evaluated according to the risk label; determining whether all the key risk points exist in an actual scene; if yes, marking the risk level of the job plan to be evaluated as a first-level risk when the total number of key risk points belongs to a first preset range, as a second-level risk when the total number of key risk points belongs to a second preset range, and as a third-level risk when the total number of key risk points belongs to a third preset range; otherwise, marking the risk level of the job plan to be evaluated as a third-level risk; wherein the risk severity of the job plan increases in the order of the first-level risk, the second-level risk and the third-level risk; the first preset range is smaller than the second preset range, and the second preset range is smaller than the third preset range; in the case of the second-level risk or the third-level risk, issuing a warning signal.
2. The work plan risk assessment method incorporating human capabilities according to claim 1, wherein, After obtaining the risk level value of the job plan data in the job plan to be evaluated and the multidimensional personnel data of the worker, the method further comprises the following steps: after filling in missing values and correcting abnormal values of the multidimensional personnel data, obtaining preprocessed multidimensional personnel data through manual review and confirmation.
3. The work plan risk assessment method incorporating human capabilities according to claim 2, wherein, After obtaining the preprocessed multidimensional personnel data, the method further comprises the following steps: dividing the preprocessed multidimensional personnel data into qualitative indicators and first quantitative indicators; converting the qualitative indicators into second quantitative indicators; performing normalization processing on the first quantitative indicators and the second quantitative indicators to obtain normalized quantitative indicators; identifying extreme values and non-extreme values in the normalized quantitative indicators; performing logarithmic transformation on each extreme value to obtain corresponding transformed data; integrating the non-extreme values and all the transformed data to obtain standardized multidimensional personnel data.
4. The work plan risk assessment method incorporating human capabilities according to claim 3, wherein, The transformed data comprises first transformed data and second transformed data. The step of performing logarithmic transformation on each extreme value to obtain corresponding transformed data comprises the following steps: for each extreme value, determining whether the extreme value is greater than a preset threshold value; if yes, calculating the natural logarithm of the absolute value of the extreme value to obtain the first transformed data; otherwise, calculating the natural logarithm of the sum of the absolute value of the extreme value and a preset constant factor to obtain the second transformed data.
5. The work plan risk assessment method incorporating human capabilities according to claim 4, wherein, The preset constant factor is determined by the following method: regarding the extreme values greater than the preset threshold value as over-limit data points; calculating the average value of the first-order derivative function of all the over-limit data points; regarding the extreme values not greater than the preset threshold value as non-over-limit data points; constructing a first-order derivative function relationship according to the non-over-limit data points and a constant factor to be confirmed; determining the upper limit value and the lower limit value of a target inequality according to the average value of the first-order derivative function; constructing the target inequality according to the first-order derivative function relationship, the upper limit value and the lower limit value; regarding the integer of the maximum constant factor in the first-order derivative function relationship satisfying the target inequality as the preset constant factor.
6. The work plan risk assessment method incorporating human capabilities according to claim 1, wherein, The risk assessment model is obtained by: obtaining a plurality of training data; each training data includes: a risk level value of job plan sample data in a job plan sample, multidimensional personnel sample data of a worker, and a corresponding true risk label; inputting the plurality of training data into a risk assessment model to be trained, so that the risk assessment model outputs the risk level value of the job plan sample data in the job plan sample and the multidimensional personnel sample data of the worker, iteratively trains the risk assessment model with the predicted risk label of the job plan sample as the output, in the training process, calculates a loss function value according to the predicted risk label and the true risk label, and adjusts the model parameters of the risk assessment model according to the loss function value until the loss function value converges, to obtain a trained risk assessment model.
7. A work plan risk assessment device that combines human capabilities, characterized by, It comprises: a job data acquisition module, a risk label identification module, a key risk point identification module, a risk level dynamic determination module, and a warning module; the job data acquisition module is configured to obtain a risk level value of job plan data in a job plan to be evaluated and multidimensional personnel data of a worker; the risk label identification module is configured to input the risk level value and the multidimensional personnel data into the trained risk assessment model, so that the risk assessment model determines the risk label of the job plan to be evaluated according to the risk level value and the multidimensional personnel data; the key risk point identification module is configured to identify a plurality of key risk points corresponding to the job plan to be evaluated according to the risk label; the risk level dynamic determination module is configured to determine whether all key risk points exist in an actual scene; if yes, in a case where the total number of key risk points belongs to a first preset range, marking the risk level of the job plan to be evaluated as a first risk level, in a case where the total number of key risk points belongs to a second preset range, marking the risk level of the job plan to be evaluated as a second risk level, and in a case where the total number of key risk points belongs to a third preset range, marking the risk level of the job plan to be evaluated as a third risk level; otherwise, marking the risk level of the job plan to be evaluated as a third risk level; wherein the severity of the job plan risk increases in the order of the first risk level, the second risk level, and the third risk level; the first preset range is smaller than the second preset range, and the second preset range is smaller than the third preset range; the warning module is configured to issue a warning signal in a case where the risk level is a second risk level or a third risk level.
8. The work plan risk assessment device for integrating human capabilities according to Claim 7, wherein It further comprises: a job data preprocessing module; the job data preprocessing module is configured to perform missing value filling and abnormal value correction on the multidimensional personnel data, and obtain preprocessed multidimensional personnel data after manual review and confirmation.
9. A terminal device, comprising: It comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and the processor implements the job plan risk assessment method combined with personnel ability according to any one of claims 1-6 when executing the computer program.
10. A computer-readable storage medium, characterized in that, The computer readable storage medium comprises a stored computer program, wherein the computer readable storage medium controls the device where the computer readable storage medium is located to execute the work plan risk assessment method combined with personnel capacity when the computer program runs.