Artificial intelligence-based on-duty efficiency comprehensive evaluation method

Abstract

The invention relates to the technical field of duty performance, in particular to a duty performance comprehensive evaluation method based on artificial intelligence. The method comprises the following steps: determining a specific target of performance evaluation, and determining a key index for evaluating the duty performance; on-duty data related to evaluation are collected according to the determined indexes; and analyzing the on-duty data by using an extreme gradient boosting model, and evaluating the on-duty efficiency based on the on-duty scoring rule. According to the design of the invention, by using advanced machine learning algorithms such as an extreme gradient lifting model and the like and combining with a feature selection process, key factors influencing on-duty efficiency can be effectively identified, so that a more accurate and fair evaluation result is provided. Therefore, the work performance of the on-duty personnel can be reflected more truly, the transparency and scientificity of performance management can be promoted, the characteristic that the on-duty data changes along with the time and the environment is considered, and a dynamic adjustment mechanism is introduced, so that the model can better adapt to complex conditions in actual work.

Description

Technical Field

[0001] The present invention relates to the technical field of on-duty performance evaluation, and in particular to an on-duty performance comprehensive evaluation method based on artificial intelligence. Background Art

[0002] Traditional duty performance evaluation often relies on manual judgment and is easily affected by factors such as the evaluator's personal preferences and experience, resulting in large deviations in the evaluation results, lack of objectivity and consistency; previous evaluation methods may only focus on one aspect of performance, such as response speed, while ignoring other important factors, such as task completion quality and customer satisfaction, and cannot fully reflect the actual work efficiency of duty personnel; in the era of big data, a large amount of data is generated during duty, but traditional methods fail to make full use of these data for in-depth analysis, missing the opportunity to discover potential problems and improvement opportunities through data analysis; traditional evaluation methods usually use fixed standards or models, which are difficult to adapt to the ever-changing factors in the duty environment, such as emergencies, seasonal fluctuations, etc., resulting in the evaluation results being out of touch with the actual situation. In summary, a comprehensive evaluation method for duty performance based on artificial intelligence is provided. Summary of the invention

[0003] The purpose of the present invention is to provide a comprehensive evaluation method for duty efficiency based on artificial intelligence to solve the problems of strong subjectivity, single evaluation dimension, low data utilization and lack of dynamic adjustment capability of traditional duty proposed in the above background technology.

[0004] To achieve the above object, the present invention provides a method for comprehensive evaluation of duty efficiency based on artificial intelligence, comprising the following steps: S1. Establish the objectives of performance evaluation and determine the indicators used to evaluate performance; S2. Collect duty data related to the evaluation based on the indicators for evaluating duty effectiveness; S3. Use the extreme gradient boosting model to analyze the duty data, optimize the extreme gradient boosting model for night duty in dim light, and evaluate the duty efficiency based on the duty scoring rules according to the analysis results of the model.

[0005] As a further improvement of the technical solution, in S1, establishing the goal of performance evaluation and determining the indicators for evaluating duty performance include the following steps: S1.1. Clarify the duty response time of the performance evaluation target and set the target; S1.2. Determine the average response time as the key indicator for evaluating on-duty effectiveness.

[0006] As a further improvement of the technical solution, in S2, the duty data includes duty records, duty time, duty frequency, and customer feedback.

[0007] As a further improvement of the technical solution, in S3, according to the analysis results of the model, the duty performance is evaluated based on the duty scoring rules, including the following steps: S3.1. Preprocess the duty data; S3.2, extracting features related to duty response time through duty feature selection model; S3.3, dividing the duty data after feature extraction into a training set and a validation set; S3.4. Use the training set data to train the extreme gradient boosting model and adjust the model parameters by minimizing the objective function; S3.5. Use the trained model to evaluate new duty data and evaluate duty performance based on the duty scoring rules.

[0008] As a further improvement of the technical solution, in S3.2, extracting features related to the duty response time through the duty feature selection model includes the following steps: S3.21, generate new features from duty data; S3.22. Standardize the new features so that the mean of each feature is 0 and the standard deviation is 1; S3.23. Construct a duty feature selection model, consider the skewed distribution of duty response time, optimize the duty feature selection model, and further optimize the duty feature selection model according to the dynamic changes of duty data; S3.24. Use the training data set to train the duty feature selection model; S3.25. After training is completed, check the coefficient of each feature, and the features with non-zero coefficients are considered to be features related to response time.

[0009] As a further improvement of the technical solution, in S3.23, the duty feature selection model is: ; Considering the skewed distribution of duty response time, the duty feature selection model is optimized: ; In view of the dynamic changes of duty data, the duty feature selection model is further optimized: ; in, represents the characteristic coefficient of the duty feature selection model; represents the best characteristic coefficient estimate; represents the best characteristic coefficient estimate after considering the skewed distribution of on-duty response time; It represents the best characteristic coefficient estimate after the dynamic change of the duty data; Indicates the latter When the minimum value is reached The value of Indicates the number of duty data records; Indicates The target variable of the duty data record; Indicates The feature vector of the duty data record; Representation characteristics The corresponding model coefficients; Indicates the number of features; represents the regularization parameter; Indicates the index of duty data record; represents the index of the feature vector; Represents parameters that are dynamically adjusted based on data distribution; Indicates control characteristics Weights under dynamically changing conditions Parameters that affect model complexity; Representation characteristics Weights under dynamically changing conditions.

[0010] As a further improvement of the technical solution, in S3.4, the extreme gradient boosting model is trained using the training set data, and the model parameters are adjusted by minimizing the objective function, including the following steps: S3.41, establishing a response time model error function, and optimizing the response time model error function according to the different heart rate changes of individuals; S3.42. Define the parameters of the extreme gradient boosting model, establish the objective function of the extreme gradient boosting model, optimize the extreme gradient boosting model for remote areas with inconvenient communication, and further optimize the extreme gradient boosting model considering nighttime duty in dim light; S3.43. Train the extreme gradient boosting model using the defined parameters and objective function, and enable early stopping during training.

[0011] As a further improvement of the technical solution, in S3.41, the response time model error function is: ; According to the different changes in individual heart rates, the error function of the response time model is optimized: ; in, Indicates the error in response time; It indicates the error in response time after individual heart rate changes; Indicates The actual response time of each duty data record; The model predicts the The actual response time of each duty data record; Indicates the number of duty data records; Indicates the index of duty data record; represents the number of decision trees; Represents the index of the decision tree; represents the regularization term; Indicates The prediction function of a decision tree; represents the weight associated with heart rate; Indicates the heart rate change of the on-duty personnel during the duty mission.

[0012] As a further improvement of the technical solution, in S3.42, the objective function of the extreme gradient boosting model is: ; For remote areas with poor communication, the extreme gradient boosting model is optimized: ; Considering the dark night duty, the extreme gradient boosting model is further optimized: ; in, Indicates The total objective function at the iteration; It represents the total objective function after duty in remote areas with inconvenient communication; It represents the total objective function after taking into account the night duty with dim light; represents the number of iterations of the extreme gradient boosting model; Represents the slope of the objective function at the current response time prediction value; represents the curvature of the objective function at the current response time prediction value; Indicates At the first iteration, the decision tree Predicted response time based on duty data records; Represents the weight factor for adjusting the impact of areas with inconvenient communications; Indicator function for remote areas; Indicates the weight of adjusting the impact of night duty; Indicator function for night duty; Indicates the index of duty data record; represents the regularization term; Indicates The decision tree model at the iteration.

[0013] As a further improvement of the technical solution, in S3.5, the new duty data is evaluated using the trained model, and the duty efficiency is evaluated based on the duty scoring rule, including the following steps: S3.51. Collect new duty data and make the data format consistent with the training data; S3.52, preprocess and standardize new data; S3.53, generate new features as needed; S3.54, input the standardized new data into the trained model for prediction; S3.55. Based on the response time prediction results of the model and the duty scoring rules, explain the evaluation results of duty effectiveness.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. This AI-based comprehensive evaluation method for on-duty performance uses advanced machine learning algorithms such as the extreme gradient boosting model and combines it with a feature selection process. This method can effectively identify key factors that affect on-duty performance, thereby providing more accurate and fair evaluation results. This not only helps to more realistically reflect the work performance of on-duty personnel, but also promotes the transparency and scientific nature of performance management.

[0015] 2. This AI-based comprehensive evaluation method for duty performance takes into account the characteristics of duty data changing over time and environment, and introduces a dynamic adjustment mechanism (such as special optimization for remote areas and night duty) to enable the model to better adapt to the complex situations in actual work. This not only improves the flexibility and practicality of the model, but also promotes the process of continuous improvement and optimization, helping organizations to promptly discover and solve problems in the duty process, thereby improving the overall service quality and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The figure is a flow chart of the overall method of the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in 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.

[0018] Example: See Figure 1As shown, this embodiment provides a comprehensive evaluation method for duty efficiency based on artificial intelligence, including the following steps: S1. Establish the objectives of performance evaluation and determine the indicators used to evaluate performance; In this embodiment, establishing the goal of performance evaluation and determining the indicators for evaluating duty performance include the following steps: S1.1. Clarify the duty response time of the performance evaluation target and set a target, such as fixing the average response time at 10 minutes; S1.2. Determine the average response time (the average time from receiving the alarm call to the arrival of the first responders at the scene) as the key indicator for evaluating on-duty effectiveness.

[0019] S2. Collect duty data related to the evaluation based on the indicators for evaluating duty effectiveness, including duty records, customer feedback, event processing time, task completion, etc.; In this embodiment, the duty data includes duty records, duty time, duty frequency, and customer feedback.

[0020] S3. Use the extreme gradient boosting model to analyze the duty data. Optimize the extreme gradient boosting model for night duty in dim light. According to the analysis results of the model, the duty efficiency is evaluated based on the duty scoring rules. In this embodiment, the extreme gradient boosting model is to correct the errors of the previous round of models by gradually adding weak learners (base learners) in an iterative manner, that is, by integrating multiple decision trees, the residual is gradually reduced by an additive model, that is, a new decision tree is added each time to fit the residual of the previous round of predictions; the decision tree can well capture the nonlinear relationship in the data, which makes them particularly suitable for processing complex and nonlinear data sets. In the evaluation of duty efficiency, there may be multiple factors interacting and affecting the response time, and the decision tree can effectively discover these complex relationships; by analyzing a large amount of duty data through the extreme gradient boosting model, it is possible to more accurately predict and evaluate the key indicators such as the response time and task completion of the duty personnel, thereby comprehensively reflecting the duty efficiency; the extreme gradient boosting model can select the factors that have the greatest impact on the duty efficiency from many features, helping managers understand which aspects need to be improved, so as to formulate more targeted optimization measures; the extreme gradient boosting model can process dynamically changing data, adapt to the duty conditions in different time periods and different environments, provide real-time performance evaluation, and support dynamic adjustment and optimization strategies; using the extreme gradient boosting model to analyze duty data and evaluate duty efficiency includes the following steps: S3.1. Preprocess the duty data, including cleaning the data and processing missing values ​​and outliers; S3.2. Extract relevant features of duty response time through duty feature selection model, including task type, geographical location, task complexity, resource allocation, and ensure that the model only uses the most relevant features for training; Among them, the duty feature selection model includes data layer, preprocessing layer, model layer and output layer. The goal of the duty feature selection model is to minimize the complexity of the model as much as possible while minimizing the prediction error, so as to avoid overfitting; the duty feature selection model retains the most important features by compressing unimportant feature coefficients to zero, which makes the model more concise, easier to understand and explain, and helps managers to clarify which factors have the greatest impact on duty response time; through feature selection, the model only needs to consider the most relevant features, which reduces the complexity of the model, reduces the risk of overfitting, and improves the generalization ability of the model; reducing the number of features can significantly reduce the computational cost of the model and speed up training and prediction, which is especially important when dealing with large-scale data sets; by excluding irrelevant or redundant features, the model can focus more on key features, thereby improving the accuracy and stability of prediction; extracting features related to duty response time through the duty feature selection model includes the following steps: S3.21. Generate new features from duty data, including extracting duty times, duty duration, and duty frequency features from duty records, and extracting average processing time, maximum processing time, and shortest processing time features from event processing time; S3.22. Standardize the new features so that the mean of each feature is 0 and the standard deviation is 1. This step helps to eliminate the impact of feature scale differences on the model. S3.23, build a duty feature selection model; S3.24. Use the training data set to train the on-duty feature selection model. The model will automatically learn which features have a significant impact on the response time and compress the coefficients of unimportant features to zero. The on-duty feature selection model is a linear model that penalizes larger coefficient values ​​by adding an L1 regularization term to the loss function to achieve the purpose of feature selection. This regularization method can directly compress the coefficients of some unimportant features to zero, effectively achieving feature selection. Furthermore, the duty feature selection model is: ; The duty response time usually presents a skewed distribution, that is, most of the response times are short, while a few of them are long. If this distribution characteristic is not taken into account, the model may overfit the data of short response times and ignore the data of long response times. By optimizing the model, these long-tail effects can be captured more accurately and the accuracy of the overall prediction can be improved. The skewed distribution of data may cause the model to be sensitive to outliers. By optimizing the model, the impact of outliers on the model can be reduced, and the robustness of the model can be improved, so that it can still maintain good performance in the face of extreme values. By considering the skewed distribution, the optimized duty feature selection model can more accurately identify the features that have the greatest impact on the duty response time, which helps to improve the accuracy of feature selection and ensure that the model only uses the most relevant features for training. Considering the skewed distribution of duty response time, the duty feature selection model is optimized: ; Duty data will constantly change due to factors such as time, environment, and equipment status. By dynamically adjusting the importance of features, the model can better adapt to these changes and provide more accurate evaluation results; the importance of features in different time periods and different environments may vary. Through dynamic adjustment, the model can more accurately capture these changes and improve the accuracy of predictions; dynamically adjusting the importance of features can reduce the model's dependence on specific time periods or specific environments, improve the robustness of the model, and enable it to maintain good performance in the face of various changes; through dynamic adjustment, the model can reflect changes in duty data in real time, provide managers with timely feedback, and help them adjust strategies and resource allocation in a timely manner; and further optimize the duty feature selection model in response to the dynamic changes in duty data: ; in, represents the characteristic coefficient of the duty feature selection model; represents the best characteristic coefficient estimate; represents the best characteristic coefficient estimate after considering the skewed distribution of on-duty response time; It represents the best characteristic coefficient estimate after the dynamic change of the duty data; Indicates the latter When the minimum value is reached The value of Indicates the number of duty data records; Indicates The target variable of the duty data record is the duty response time; Indicates The feature vector of the duty data record; Representation characteristics The corresponding model coefficients; Indicates the number of features; Represents the regularization parameter, which is used to control the complexity of the model and prevent overfitting; Indicates the index of duty data record; represents the index of the feature vector; Represents parameters that are dynamically adjusted based on data distribution; Indicates control characteristics Weights under dynamically changing conditions Parameters that affect model complexity; Representation characteristics The weight under dynamically changing conditions is used to adjust the importance of the feature in different time periods or under different environments; S3.25. After training, check the coefficient of each feature. The features with non-zero coefficients are considered to be the features related to response time. S3.3, dividing the duty data after feature extraction into a training set and a validation set; S3.4. Use the training set data to train the extreme gradient boosting model and adjust the model parameters by minimizing the objective function; Among them, by minimizing the objective function, the extreme gradient boosting model can find the optimal model parameters so that the prediction error of the model on the training set is minimized, which helps to improve the prediction accuracy of the model and ensure the reliability of the evaluation results; using the training set data to train the extreme gradient boosting model, and adjusting the model parameters by minimizing the objective function, including the following steps: S3.41, establish the response time model error function; Furthermore, the purpose of establishing the response time model error function is to quantify the difference between the model prediction value and the actual value, so as to guide the optimization and parameter adjustment of the model; the error function quantifies the difference between the model prediction value and the actual value through a mathematical formula, and provides a specific numerical indicator. Through the error function, the performance of the model on the training set and the validation set can be evaluated to ensure that the model has good prediction ability; the response time model error function is: ; If an individual's heart rate is high, it indicates that they may be in a state of high stress or alertness. The model can adjust the predicted response time accordingly to reflect possible behavioral changes in this physiological activation state. Heart rate changes reflect the individual's physiological state and psychological stress, and have a significant impact on response time. By considering heart rate changes, the model can more accurately predict response time and improve the accuracy of predictions. Different individuals have different heart rate changes in different situations (such as tension, fatigue, etc.). By considering these changes, the model can better adapt to data from different individuals and different situations and improve the robustness of the model. By introducing heart rate changes as an additional feature, the model can more comprehensively capture the factors that affect response time and improve the accuracy of predictions. According to the different heart rate changes of individuals, the error function of the response time model is optimized: ; in, Indicates the error in response time; It indicates the error in response time after individual heart rate changes; Indicates The actual response time of each duty data record; The model predicts the The actual response time of each duty data record; Indicates the number of duty data records; Indicates the index of duty data record; represents the number of decision trees; Represents the index of the decision tree; represents the regularization term, which is used to prevent overfitting; Indicates The prediction function of a decision tree , It is The feature vector of the duty data record, It is The number of leaf nodes in a decision tree, It is Decision Tree The output value of a leaf node, It is Decision Tree The area corresponding to the leaf nodes; represents the weight associated with heart rate; Indicates the heart rate change of the on-duty personnel during the duty mission, which is used to adjust the response time prediction; S3.42. Define the parameters of the extreme gradient boosting model, including the objective function, evaluation index, maximum depth of the tree, learning rate, and subsampling ratio, and establish the objective function of the extreme gradient boosting model; Furthermore, by establishing a minimization objective function, we can ensure that the prediction error of the model on the training set is minimized, thereby improving the prediction accuracy of the model; the objective function of the extreme gradient boosting model is: ; ; Data in remote areas may have special distribution and characteristics. By optimizing the model, these special characteristics can be captured more accurately, and the prediction accuracy of the model in remote areas can be improved. Data in remote areas may have more noise and outliers. By optimizing the model, the robustness of the model can be improved so that it can still maintain good performance when facing these data. By introducing specific features and adjusting weights, the optimized model can more accurately predict the response time in remote areas and improve the accuracy of prediction. The optimized model can provide more detailed feature impact analysis to help managers understand which factors have the greatest impact on the performance of duty in remote areas, so as to formulate more targeted improvement measures. For duty in remote areas with inconvenient communications (such as the Qinghai-Tibet Plateau), the extreme gradient boosting model is optimized: ; The data of night duty may have special distribution and characteristics, such as visual impairment caused by dim light and increased fatigue. By optimizing the model, these particularities can be captured more accurately, and the prediction accuracy of the model in the case of night duty can be improved; the data of night duty may have more noise and outliers. By optimizing the model, the robustness of the model can be improved so that it can still maintain good performance when facing these data; the optimized model can better adapt to the specific situation of night duty, provide more personalized evaluation results, and help managers understand the effectiveness of night duty more accurately; the optimized model can better adapt to the data changes of night duty and improve the adaptability and generalization ability of the model; considering the night duty with dim light, the extreme gradient boosting model is further optimized: ; in, Indicates The total objective function at the iteration; It represents the total objective function after duty in remote areas with inconvenient communication; It represents the total objective function after taking into account the night duty with dim light; represents the number of iterations of the extreme gradient boosting model; represents the curvature of the objective function at the current response time prediction value; Indicates At the first iteration, the decision tree Predicted response time based on duty data records; Represents the weight factor for adjusting the impact of areas with inconvenient communications; The indicator function for remote areas takes the value 1 when the duty data record is located in a remote area with inconvenient communication, otherwise it takes the value 0; Indicates the weight of adjusting the impact of night duty; The indicator function represents night duty, which takes the value of 1 when the duty data record is night duty, otherwise it takes the value of 0; Indicates the index of duty data record; Represents the regularization term, which is used to control the complexity of the model and prevent overfitting. The regularization term consists of two parts: the structural complexity of the tree and the sum of the squares of the leaf node weights; Indicates Decision tree model at the iteration; Represents the complexity penalty coefficient of the tree; Indicates the number of leaf nodes; represents the regularization coefficient of the weight; Represents the weight of the leaf node; S3.43. Train the extreme gradient boosting model using the defined parameters and objective function, enabling early stopping during training (training is terminated early if the performance on the validation set does not improve for a given number m of consecutive iterations, which helps prevent overfitting and can save training time). S3.5. Use the trained model to evaluate new duty data and evaluate duty effectiveness (predict duty response time) based on duty scoring rules; Among them, the new duty data is evaluated using the trained model, and the duty performance is evaluated based on the duty scoring rules, including the following steps: S3.51. Collect new duty data and make the data format consistent with the training data. The new data contains the following features: number of duty times, duty duration, duty frequency, etc.; S3.52. Preprocess and standardize new data, including data cleaning, removing missing values, outliers and duplicate values, and ensuring that data quality meets the requirements of the model; S3.53. Generate new features as needed, for example, extract new features from duty records (such as the time interval between the last duty), extract new features from task completion (such as the standard deviation of completion), and extract new features from event processing time (such as the median processing time); S3.54, input the standardized new data into the trained model for prediction; S3.55. Based on the response time prediction results of the model and the duty scoring rules, explain the evaluation results of the duty efficiency. The specific duty scoring rules are: based on the response time of 10 minutes, 1 point will be deducted for every 1 minute the response time exceeds the limit. The efficiency level is: total score 100 points, 90 points and above are excellent, 80-89 points are good, 70-79 points are qualified, and below 70 points require improvement. Any response time exceeding 30 minutes will be marked as abnormal and the cause needs further investigation.

[0021] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention, and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. A comprehensive evaluation method for duty efficiency based on artificial intelligence, characterized in that: The following steps are involved: S1. Establish the objectives of performance evaluation and determine the indicators used to evaluate performance; S2. Collect duty data related to the evaluation based on the indicators for evaluating duty effectiveness; S3. Use the extreme gradient boosting model to analyze the duty data, optimize the extreme gradient boosting model for night duty in dim light, and evaluate the duty efficiency based on the duty scoring rules according to the analysis results of the model.

2. The method for comprehensive evaluation of duty efficiency based on artificial intelligence according to claim 1 is characterized in that: In S1, the goal of performance evaluation is established, and the indicators for evaluating duty performance are determined, including the following steps: S1.

1. Clarify the duty response time of the performance evaluation target and set the target; S1.

2. Determine the average response time as the key indicator for evaluating on-duty effectiveness.

3. The method for comprehensive evaluation of duty efficiency based on artificial intelligence according to claim 2 is characterized in that: In S2, the duty data includes duty records, duty time, duty frequency, and customer feedback.

4. The method for comprehensive evaluation of duty efficiency based on artificial intelligence according to claim 3 is characterized in that: In S3, the extreme gradient boosting model is used to analyze the duty data, and the duty performance is evaluated based on the duty scoring rules according to the analysis results of the model, including the following steps: S3.

1. Preprocess the duty data; S3.2, extracting features related to duty response time through duty feature selection model; S3.3, dividing the duty data after feature extraction into a training set and a validation set; S3.

4. Use the training set data to train the extreme gradient boosting model and adjust the model parameters by minimizing the objective function; S3.

5. Use the trained model to evaluate new duty data and evaluate duty performance based on the duty scoring rules.

5. The method for comprehensive evaluation of duty efficiency based on artificial intelligence according to claim 4 is characterized in that: In S3.2, extracting features related to the duty response time through the duty feature selection model includes the following steps: S3.21, generate new features from duty data; S3.

22. Standardize the new features so that the mean of each feature is 0 and the standard deviation is 1; S3.

23. Construct a duty feature selection model, consider the skewed distribution of duty response time, optimize the duty feature selection model, and further optimize the duty feature selection model according to the dynamic changes of duty data; S3.

24. Use the training data set to train the duty feature selection model; S3.

25. After training is completed, check the coefficient of each feature, and the features with non-zero coefficients are considered to be features related to response time.

6. The method for comprehensive evaluation of duty efficiency based on artificial intelligence according to claim 5 is characterized in that: In S3.23, the duty feature selection model is: ； Considering the skewed distribution of duty response time, the duty feature selection model is optimized: ； In view of the dynamic changes of duty data, the duty feature selection model is further optimized: ； in, represents the characteristic coefficient of the duty feature selection model; represents the best characteristic coefficient estimate; represents the best characteristic coefficient estimate after considering the skewed distribution of on-duty response time; It represents the best characteristic coefficient estimate after the dynamic change of the duty data; Indicates the latter When the minimum value is reached The value of Indicates the number of duty data records; Indicates The target variable of the duty data record; Indicates The feature vector of the duty data record; Representation characteristics The corresponding model coefficients; Indicates the number of features; represents the regularization parameter; Indicates the index of duty data record; represents the index of the feature vector; Represents parameters that are dynamically adjusted based on data distribution; Indicates control characteristics Weights under dynamically changing conditions Parameters that affect model complexity; Representation characteristics Weights under dynamically changing conditions.

7. The method for comprehensive evaluation of duty efficiency based on artificial intelligence according to claim 6 is characterized in that: In S3.4, the extreme gradient boosting model is trained using the training set data, and the model parameters are adjusted by minimizing the objective function, including the following steps: S3.41, establishing a response time model error function, and optimizing the response time model error function according to the different heart rate changes of individuals; S3.

42. Define the parameters of the extreme gradient boosting model, establish the objective function of the extreme gradient boosting model, optimize the extreme gradient boosting model for remote areas with inconvenient communication, and further optimize the extreme gradient boosting model considering nighttime duty in dim light; S3.

43. Train the extreme gradient boosting model using the defined parameters and objective function, and enable early stopping during training.

8. The method for comprehensive evaluation of duty efficiency based on artificial intelligence according to claim 7 is characterized in that: In S3.41, the response time model error function is: ； According to the different changes in individual heart rates, the error function of the response time model is optimized: ； in, Indicates the error in response time; It indicates the error in response time after individual heart rate changes; Indicates The actual response time of each duty data record; The model predicts the The actual response time of each duty data record; Indicates the number of duty data records; Indicates the index of duty data record; represents the number of decision trees; Represents the index of the decision tree; represents the regularization term; Indicates The prediction function of a decision tree; represents the weight associated with heart rate; Indicates the heart rate change of the on-duty personnel during the duty mission.

9. The method for comprehensive evaluation of duty efficiency based on artificial intelligence according to claim 8 is characterized in that: In S3.42, the objective function of the extreme gradient boosting model is: ； For remote areas with poor communication, the extreme gradient boosting model is optimized: ； Considering the dark night duty, the extreme gradient boosting model is further optimized: ； in, Indicates The total objective function at the iteration; It represents the total objective function after duty in remote areas with inconvenient communication; It represents the total objective function after taking into account the night duty with dim light; represents the number of iterations of the extreme gradient boosting model; Represents the slope of the objective function at the current response time prediction value; represents the curvature of the objective function at the current response time prediction value; Indicates At the first iteration, the decision tree Predicted response time based on duty data records; Indicates the weight factor for adjusting the impact of areas with inconvenient communications; Indicator function for remote areas; Indicates the weight of adjusting the impact of night duty; Indicator function for night duty; Indicates the index of duty data record; represents the regularization term; Indicates The decision tree model at the iteration.

10. The method for comprehensive evaluation of duty efficiency based on artificial intelligence according to claim 9 is characterized in that: In S3.5, the new duty data is evaluated using the trained model, and the duty performance is evaluated based on the duty scoring rule, including the following steps: S3.

51. Collect new duty data and make the data format consistent with the training data; S3.52, preprocess and standardize new data; S3.53, generate new features as needed; S3.54, input the standardized new data into the trained model for prediction; S3.

55. Based on the response time prediction results of the model and the duty scoring rules, explain the evaluation results of duty effectiveness.

Images