Application method and system of metallurgical safety enterprise personal safety portrait model, medium and equipment

Abstract

The invention provides an application method and system of a metallurgical safety enterprise personal safety portrait model, a medium and equipment. The application method comprises the following steps: step 1, collecting safety related information data of personnel in a metallurgical enterprise; step 2, performing data cleaning on the acquired information data through a preset data cleaning rule; 3, establishing a personal safety portrait rule based on big data analysis; step 4, applying a preset portrait label rule to mark one or more labels on the person; and 5, acquiring concerned crowd information through the label, and providing a basis for safety management of an enterprise. According to the invention, key minority is caught through portrait label rules, key crowds are found accurately, key education and key staring are implemented, and the effectiveness and pertinence of field violation correction are improved.

Description

Technical Field

[0001] The present invention relates to the field of production safety technology, and in particular to an application method, system, medium and equipment for a personal safety portrait model of a metallurgical safety enterprise. Background Art

[0002] In the field of production safety, especially in the metallurgical industry, enterprises face many challenges, including but not limited to frequent safety accidents, insufficient safety awareness of employees, improper allocation of safety management resources, etc. Traditional safety management methods often rely on manual inspection and experience judgment, lack of refined and personalized management, and it is difficult to effectively identify and prevent potential safety risks.

[0003] Existing safety management technologies mostly focus on macro-level risk assessment and accident prevention, but fail to fully consider the differences in safety behaviors of individual employees. For example, traditional safety training and assessments often use uniform standards, ignoring employees' job characteristics, personal experience, and learning needs, resulting in uneven training results. In addition, most existing safety management systems rely on static data and cannot reflect changes in employees' safety status in real time, causing safety management measures to lag behind actual needs.

[0004] Patent application document CN116167618A discloses a method for safety profiling of on-site power grid workers, including obtaining worker information through work tickets associated with work plans; conducting safety profiling of all workers; preventing accidents for high-risk workers through safety profiling scoring; realizing data-based and index-based safety evaluation of workers; worker safety profiling helps to support targeted control of on-site operations and improve control efficiency; worker safety profiling can support the early warning function of workers, timely identify personnel who do not meet the operation requirements, and prevent violations of operation rules and accidents. The safety profiling method for on-site power grid workers has been more carefully designed and optimized for the characteristics of power grid operations, while the personal safety profiling model of metallurgical safety enterprises focuses more on specific risk factors in the metallurgical industry. Summary of the invention

[0005] In view of the defects in the prior art, the purpose of the present invention is to provide an application method, system, medium and equipment for a personal safety portrait model of a metallurgical safety enterprise.

[0006] The application method of the personal safety portrait model of a metallurgical safety enterprise provided by the present invention comprises:

[0007] Step 1: Collect safety-related information data of personnel in metallurgical enterprises;

[0008] Step 2: Clean the collected information data using preset data cleaning rules, including data integrity check, data consistency verification, duplicate data identification, error and outlier detection, data type and range verification, reference data verification, text data standardization, and privacy and sensitive information processing;

[0009] Step 3: Establish personal security profile rules based on big data analysis, where the personal security profile rules include basic personnel information, behavior data, and assessment data;

[0010] Step 4: Use the preset portrait labeling rules to label one or more labels for individuals, which cover basic personnel information, safety certification information, safety education and training, risk identification and control, hidden danger investigation and management, and safety behavior performance;

[0011] Step 5: Obtain information about people of concern through tags to provide a basis for the company's security management.

[0012] Preferably, the data cleaning process includes data review, rule definition, tool selection, data preprocessing, data verification, data conversion, data audit, documentation and continuous monitoring.

[0013] Preferably, the big data analysis process includes data collection, data storage, data cleaning and preprocessing, data integration, data modeling, data analysis, data visualization, deployment and application, monitoring and optimization, and security and compliance checks.

[0014] Preferably, the data modeling includes:

[0015] In logistic regression, the model is of the form:

[0016]

[0017] Among them, P(y=1|X) represents the probability that the dependent variable y takes the value of 1 under the condition of a given independent variable X; β0 is the intercept term; β1, β2...β n are regression coefficients; X1, X2, ...X n is the independent variable; n is the number of independent variables;

[0018] The parameters β are estimated by maximizing the log-likelihood function and the weights are optimized using the gradient descent algorithm:

[0019]

[0020] Where N is the number of samples; y i is the actual label value of the i-th sample;

[0021] After the model training is completed, its classification effect is evaluated through cross-validation and ROC curve to reveal the patterns in the data;

[0022] For clustering problems, K-means clustering is used. The goal is to find the best clustering division of the data by minimizing the Euclidean distance from each sample x point to the center of the cluster to which it belongs:

[0023]

[0024] Among them, k is the number of categories, S i represents the i-th class sample set, μ i is the center of the i-th cluster, and convergence is achieved by iteratively updating the center point and redistributing samples to clusters;

[0025] In the regression analysis, linear regression is used and the expression of the model is:

[0026] y=β0+β1X1+β2X2+…β n X n +ε

[0027] Where ε is the error term; the coefficients are estimated by minimizing the residual sum of squares using the least squares method.

[0028]

[0029] in, is the predicted label value of the i-th sample;

[0030] To avoid overfitting, a regularization method is used to add an L2 regularization term to the loss function:

[0031]

[0032] Among them, λ is the regularization parameter;

[0033] The complexity of the model is controlled by regularization, the generalization ability of the model is improved, and the mean square error is used to evaluate the fitting effect of the model;

[0034] For association rule mining, the Apriori model is used to mine frequent item sets and association rules from large-scale data sets through support and confidence indicators. By mining rules with high confidence and support, potential patterns and trends between data items are discovered.

[0035] Preferably, the data analysis includes: using a big data processing framework to perform distributed processing on large-scale data sets. First, massive data is stored on multiple computing nodes through the Hadoop distributed file system, and then the MapReduce programming model is used to perform distributed processing and calculation of the data. For complex machine learning tasks, the SparkMLlib library is used to implement the training and prediction of distributed machine learning models. The entire process includes data sharding, parallel computing, task scheduling, and result merging, thereby achieving efficient processing and analysis of large-scale data sets.

[0036] The application system of the personal safety portrait model of a metallurgical safety enterprise provided by the present invention comprises:

[0037] Module M1: Collecting safety-related information data of personnel in metallurgical enterprises;

[0038] Module M2: The collected information data is cleaned according to the preset data cleaning rules, including data integrity check, data consistency verification, duplicate data identification, error and outlier detection, data type and range verification, reference data verification, text data standardization, and privacy and sensitive information processing;

[0039] Module M3: Establishing personal security profile rules based on big data analysis, the personal security profile rules include basic personnel information, behavior data and assessment data;

[0040] Module M4: Use the preset portrait labeling rules to label one or more labels for individuals, which cover basic information of personnel, safety certification information, safety education and training, risk identification and control, hidden danger investigation and management, and safety behavior performance;

[0041] Module M5: Obtain information on people of concern through tags to provide a basis for the company's security management.

[0042] Preferably, the data cleaning process includes data review, rule definition, tool selection, data preprocessing, data verification, data conversion, data audit, documentation and continuous monitoring.

[0043] Preferably, the big data analysis process includes data collection, data storage, data cleaning and preprocessing, data integration, data modeling, data analysis, data visualization, deployment and application, monitoring and optimization, and security and compliance checks.

[0044] Preferably, the data modeling includes:

[0045] In logistic regression, the model is of the form:

[0046]

[0047] Among them, P(y=1|X) represents the probability that the dependent variable y takes the value of 1 under the condition of a given independent variable X; β0 is the intercept term; β1, β2...β n are regression coefficients; X1, X2, ...X n is the independent variable; n is the number of independent variables;

[0048] The parameters β are estimated by maximizing the log-likelihood function and the weights are optimized using the gradient descent algorithm:

[0049]

[0050] Where N is the number of samples; y i is the actual label value of the i-th sample;

[0051] After the model training is completed, its classification effect is evaluated through cross-validation and ROC curve to reveal the patterns in the data;

[0052] For clustering problems, K-means clustering is used. The goal is to find the best clustering division of the data by minimizing the Euclidean distance from each sample x point to the center of the cluster to which it belongs:

[0053]

[0054] Among them, k is the number of categories, S i represents the i-th class sample set, μ i is the center of the i-th cluster, and convergence is achieved by iteratively updating the center point and redistributing samples to clusters;

[0055] In the regression analysis, linear regression is used and the expression of the model is:

[0056] y=β0+β1X1+β2X2+…β n X n +ε

[0057] Where ε is the error term; the coefficients are estimated by minimizing the residual sum of squares using the least squares method.

[0058]

[0059] in, is the predicted label value of the i-th sample;

[0060] To avoid overfitting, a regularization method is used to add an L2 regularization term to the loss function:

[0061]

[0062] Among them, λ is the regularization parameter;

[0063] The complexity of the model is controlled by regularization, the generalization ability of the model is improved, and the mean square error is used to evaluate the fitting effect of the model;

[0064] For association rule mining, the Apriori model is used to mine frequent item sets and association rules from large-scale data sets through support and confidence indicators. By mining rules with high confidence and support, potential patterns and trends between data items are discovered.

[0065] Preferably, the data analysis includes: using a big data processing framework to perform distributed processing on large-scale data sets. First, massive data is stored on multiple computing nodes through the Hadoop distributed file system, and then the MapReduce programming model is used to perform distributed processing and calculation of the data. For complex machine learning tasks, the SparkMLlib library is used to implement the training and prediction of distributed machine learning models. The entire process includes data sharding, parallel computing, task scheduling, and result merging, thereby achieving efficient processing and analysis of large-scale data sets.

[0066] According to the computer-readable storage medium storing a computer program provided by the present invention, when the computer program is executed by a processor, the steps of the application method of the personal safety portrait model of the metallurgical safety enterprise are implemented.

[0067] The electronic device provided according to the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the steps of the application method of the personal safety portrait model of the metallurgical safety enterprise are implemented.

[0068] Compared with the prior art, the present invention has the following beneficial effects:

[0069] (1) Through personal safety portrait rules and portrait label rules, the present invention can quickly and accurately obtain the safety level and status of personnel in the enterprise, accurately locate high-risk groups, customize improvement plans, and improve the effectiveness and pertinence of on-site rectification;

[0070] (2) The present invention can solve problems such as frequent safety accidents, insufficient safety awareness among employees, and improper allocation of safety management resources. The present invention also provides refined and personalized management, fully considering the differences in safety behaviors of individual employees, and can effectively identify and prevent potential safety risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0072] Figure 1A flow chart of the application method of the personal safety portrait model for metallurgical safety enterprises;

[0073] Figure 2 It is the interface of the smart security management information system;

[0074] Figure 3 It is the interface of the smart security management information system. DETAILED DESCRIPTION

[0075] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0076] Example 1

[0077] like Figure 1 The present invention provides an application method of a personal safety portrait model for a metallurgical safety enterprise, comprising:

[0078] (1) Data collection: Collect information related to personnel safety, such as: employee job changes (new employment / new resumption of work), three-level safety education, daily safety training and examinations, certification status, (repeated) violations, accident injury / accident responsibility history, occupational health exposure, physical examination records, allocation and distribution of labor protection supplies, job risks, etc.

[0079] (2) Data cleaning: The collected data is cleaned and converted into data sets that require attention.

[0080] The data cleaning rules are as follows:

[0081] Data integrity check: ensure that all required fields have values ​​and are not missing. For non-required fields, mark them or set default values.

[0082] Data consistency verification: Check whether the data format is consistent, such as date format, document number format, etc., to ensure that the data follows the predefined coding standards or classifications;

[0083] Duplicate data identification: Identify and identify duplicate records, such as identical personnel information, and merge or delete duplicates to avoid data redundancy;

[0084] Error and outlier detection: Identify obvious errors, such as illogical ages, and detect outliers, which may be input errors or true extreme cases;

[0085] Data type and range check: Ensure that the data type is correct. For example, age should be an integer. Check whether the value is within a reasonable range.

[0086] Reference data verification: Verify whether all foreign keys and reference data are valid, such as organization codes, job codes, etc., and modify or mark invalid references;

[0087] Text data standardization: standardize the format of text data, such as all uppercase or all lowercase, and remove extra spaces, special characters, and punctuation marks;

[0088] Privacy and sensitive information handling: Ensure compliance with data protection regulations and desensitize or encrypt sensitive information, such as ID numbers.

[0089] The data cleaning process is as follows:

[0090] Data review: Take a preliminary look at the data set to understand the structure and content of the data, and identify obvious problems and potential data quality issues in the data set;

[0091] Rule definition: Define data cleaning rules based on business needs and data usage scenarios to determine which fields need to be cleaned and how to clean them.

[0092] Tool selection: Choose appropriate data cleaning tools or programming languages, such as Python, R, SQL, etc., prepare data cleaning scripts or use data cleaning software.

[0093] Data preprocessing: cleaning missing values, error values, outliers and inconsistent data, standardizing data formats and text data;

[0094] Data verification: Check the results after data cleaning to ensure the accuracy and consistency of the data and verify whether the data cleaning rules are correctly executed;

[0095] Data conversion: Perform necessary conversions on the cleaned data, such as data type conversion, data aggregation, etc., to ensure that the data remains complete and accurate after conversion;

[0096] Data review: Business experts review the cleaned data to ensure that the data cleaning results meet business requirements and data quality standards;

[0097] Documentation: Record the process and rules of data cleaning, including data comparison before and after cleaning, and write documents for the data cleaning process to facilitate subsequent audits and reviews;

[0098] Continuous monitoring: Establish a data quality monitoring mechanism, regularly check data quality, and continuously optimize data cleaning rules and processes based on data usage feedback.

[0099] (3) Data modeling: Establish profiling rules and use big data technology for analysis.

[0100] The rules for portraits are as follows:

[0101] Basic information of personnel:

[0102] Job length of service: Ensure that job length of service data is within a reasonable range and has no logical errors (such as working dates in the future).

[0103] Gender: Standardize gender data to ensure that it contains only a limited number of valid values ​​(such as male and female).

[0104] Position: Ensure consistency and accuracy of position classification, refer to the established position classification.

[0105] Qualifications: Ensure consistency and accuracy of qualification classifications, refer to established qualification classifications.

[0106] Certificate: Ensure that the employee's certificate is within the specified certificate information range and there are no logical errors (such as the certificate issuance date is in the future, the certificate is within the validity period, etc.)

[0107] Behavioral data:

[0108] Safety knowledge: infer the safety knowledge of personnel through their learning, examination and assessment, knowledge evaluation, and safety certification;

[0109] Safety skills: Infer the safety skills of personnel through their current job experience, hidden danger inspection, violation, and risk identification;

[0110] Safety behavior: Infer the safety behavior of personnel through whether they have violated regulations, whether accidents have occurred, safety integrity, negative safety events, etc.

[0111] Evaluation data:

[0112] Safety assessment: Safety assessment is used to infer the initial safety awareness, safety attitude, and safety skills scores of personnel.

[0113] The specific process of big data analysis is as follows:

[0114] Data collection: Collect data from different sources, including structured data (such as tables in databases) and unstructured data (such as text, images, and videos), and use ETL (Extract, Transform, Load) tools or data collection APIs to extract data;

[0115] Data storage: Store the collected data in a system suitable for big data processing, such as a distributed file system (HDFS), a NoSQL database (such as Cassandra, MongoDB), a data warehouse, etc.

[0116] Data cleaning and preprocessing: Clean the data to remove errors, inconsistencies and duplicate information, and preprocess the data, including data conversion, normalization, denoising, handling missing values, etc.;

[0117] Data integration: Integrate data from different sources to form a unified data view, involving operations such as data fusion, data federation, and data replication;

[0118] Data modeling: According to the analysis objectives, select appropriate data models and algorithms, such as classification, clustering, regression, association rule mining, etc., and use machine learning algorithms to model the data to discover patterns and trends in the data. For classification problems, commonly used models include logistic regression, support vector machines, and decision trees.

[0119] For example, in logistic regression, the model is of the form:

[0120]

[0121] The parameters β are estimated by maximizing the log-likelihood function and the weights are optimized using the gradient descent algorithm:

[0122]

[0123] After the model training is completed, its classification effect is evaluated through cross-validation, ROC curve, etc., to reveal the patterns in the data.

[0124] For clustering problems, commonly used models include K-means clustering and hierarchical clustering. Taking K-means clustering as an example, the goal is to find the best clustering division of the data by minimizing the Euclidean distance from each sample point to the center of the cluster to which it belongs:

[0125]

[0126] Among them, μ i is the center of the i-th cluster, and convergence is achieved by iteratively updating the center point and redistributing samples to clusters.

[0127] In regression analysis, commonly used models include linear regression and ridge regression. Taking linear regression as an example, the expression of the model is:

[0128] y=β0+β1X1+β2X2+…β n X n +ε

[0129] Among them, β is the regression coefficient and ε is the error term. The coefficient is estimated by minimizing the residual sum of squares using the least squares method:

[0130]

[0131] To avoid overfitting, regularization methods such as ridge regression can be used to add an L2 regularization term to the loss function:

[0132]

[0133] The complexity of the model is controlled by regularization, the generalization ability of the model is improved, and the mean square error is used to evaluate the fitting effect of the model.

[0134] For association rule mining, commonly used models include the Apriori model and the FP-Growth model. Taking the Apriori model as an example, the model mines frequent item sets and association rules from large-scale data sets through indicators such as support and confidence. The calculation formula for support is:

[0135]

[0136] The confidence calculation formula is:

[0137]

[0138] Discover potential patterns and trends among data items by mining rules with high confidence and support.

[0139] Data analysis: Use statistical analysis and data mining techniques to conduct in-depth analysis of the data, including data preprocessing, feature extraction, model building and evaluation. First, clean and preprocess the raw data, handle missing values, outliers, and standardize data to ensure data integrity and consistency. Then, use feature selection and dimensionality reduction techniques, such as principal component analysis and factor analysis, to extract key variables, reduce data dimensions, and improve the generalization ability of the model. On this basis, select appropriate statistical models and mining algorithms based on data features, such as linear regression, decision trees, cluster analysis, etc., conduct in-depth analysis and pattern mining, and optimize model parameters through cross-validation, grid search and other methods, and finally select the best model for prediction and classification.

[0140] At the same time, big data processing frameworks such as Apache Hadoop and Apache Spark are used to perform distributed processing on large-scale data sets. First, the massive data is stored on multiple computing nodes through the Hadoop distributed file system, and then the MapReduce programming model is used to perform distributed processing and calculation of the data. For complex machine learning tasks, the Spark MLlib library can be used to implement the training and prediction of distributed machine learning models. The entire process includes steps such as data sharding, parallel computing, task scheduling, and result merging, thereby achieving efficient processing and analysis of large-scale data sets.

[0141] Data visualization: Visualize the analysis results through charts, dashboards, etc. to better understand and convey information, using data visualization tools such as Tableau, Grafana, Kibana, etc.;

[0142] Deployment and application: deploy analytical models and results into production environments and develop application programming interfaces;

[0143] Monitoring and optimization: Monitor the performance of analytical models to ensure their accuracy and reliability, and continuously optimize analytical models and processes;

[0144] Security and Compliance: Ensure that the data analysis process complies with data protection laws and company policies, implement data security controls and encryption, and protect data security.

[0145] (4) Build a profile: assign one or more “labels” to an individual.

[0146] The rules for portrait tags are as follows:

[0147] Basic information of personnel:

[0148] High education: Label users according to their education level, such as bachelor degree or above;

[0149] Young people: Label users according to their age range, such as under 35 years old;

[0150] Job length of service: Label the user according to their job length of service, such as 2 years of service, 3 years of service, etc.

[0151] Risky positions: Labeled according to the inherent risk level of the hazardous sources identified or controlled by the user's position risk, such as Class A risky positions.

[0152] Safety certification information

[0153] Certification: Depending on whether the user holds the corresponding certificate, the corresponding label will be added, such as registered safety engineer, etc.

[0154] Safety education and training

[0155] Safety learning: According to the user's safety learning situation, the corresponding label is added, such as "excellent safety learning", "excellent safety learning", etc.

[0156] Risk identification and control

[0157] Risk identification: According to the user's risk identification situation, the corresponding label is added to meet the composite range, such as "excellent hazard source control", "excellent hazard source control", etc.

[0158] Hidden danger investigation and treatment

[0159] Hidden danger investigation: According to the user's hidden danger investigation results, corresponding labels are added, such as "Hidden danger investigation is excellent", "TOP 10 hidden dangers found", etc.

[0160] Safety behavior performance

[0161] Safety behavior: Label the user based on their safety behavior, such as “zero violations”, “zero accidents”, etc.

[0162] (5) Use of portraits: User portraits will eventually label different user groups, allowing relevant analysts to immediately understand the groups they want to focus on. By focusing on the key minority, identifying key groups, and implementing key education and key surveillance, the effectiveness and pertinence of on-site rectification can be improved.

[0163] By building a safety management information system, we can provide a source of basic data and support the display of personnel safety portraits, such as Figure 2 and Figure 3 .

[0164] Example: Baosteel Co., Ltd. smart security system

[0165] My points are calculated based on five dimensions: safety certification, safety education and training, risk identification and control, hidden danger investigation and management, and safety behavior performance.

[0166] Safety assessment: Obtain safety assessment data, and obtain initial scores for safety awareness, safety attitude, and safety skills through the assessment. Dynamic scores are updated through safety points.

[0167] Risk identification and control: The risk identification and control score is calculated according to the model based on the number of hazards found, the number of hazards checked and whether they are under control, etc.

[0168] Hidden danger inspection: The hidden danger inspection score is calculated according to the model based on the number of hidden dangers found, the number of violations found, the total score of the hidden danger inspection quality, the number of hidden dangers rectified, the number of major hidden dangers found, the number of hidden dangers rectified beyond the deadline, and the number of hidden dangers not rectified beyond the deadline.

[0169] Safety behavior performance: The score of safety behavior performance is calculated according to the model based on the number of high-risk maintenance projects responsible, the number of dangerous operation projects responsible, the number of violations found, the number of people responsible for accidents, etc.

[0170] Example 2

[0171] The present invention also provides an application system for the personal safety portrait model of a metallurgical safety enterprise. The application system for the personal safety portrait model of a metallurgical safety enterprise can be realized by executing the process steps of the application method for the personal safety portrait model of a metallurgical safety enterprise, that is, technical personnel in this field can understand the application method for the personal safety portrait model of a metallurgical safety enterprise as a preferred implementation mode of the application system for the personal safety portrait model of a metallurgical safety enterprise.

[0172] The application system of the personal safety portrait model of a metallurgical safety enterprise provided by the present invention comprises: module M1: collecting safety-related information data of personnel in a metallurgical enterprise; module M2: performing data cleaning on the collected information data through preset data cleaning rules, including data integrity check, data consistency verification, duplicate data identification, error and outlier detection, data type and range verification, reference data verification, text data standardization, and privacy and sensitive information processing; module M3: establishing personal safety portrait rules based on big data analysis, the personal safety portrait rules including basic personnel information, behavior data, and assessment data; module M4: applying preset portrait labeling rules to label one or more labels for individuals, the labels covering basic personnel information, safety certification information, safety education and training, risk identification and control, hidden danger investigation and management, and safety behavior performance; module M5: obtaining information on people of concern through labels to provide a basis for the enterprise's safety management.

[0173] The data cleaning process includes data review, rule definition, tool selection, data preprocessing, data verification, data conversion, data audit, documentation and continuous monitoring.

[0174] The big data analysis process includes data collection, data storage, data cleaning and preprocessing, data integration, data modeling, data analysis, data visualization, deployment and application, monitoring and optimization, and security and compliance checks.

[0175] The data modeling includes:

[0176] In logistic regression, the model is of the form:

[0177]

[0178] Among them, P(y=1|X) represents the probability that the dependent variable y takes the value of 1 under the condition of a given independent variable X; β0 is the intercept term; β1, β2...β n are regression coefficients; X1, X2, ...X n is the independent variable; n is the number of independent variables;

[0179] The parameters β are estimated by maximizing the log-likelihood function and the weights are optimized using the gradient descent algorithm:

[0180]

[0181] Where N is the number of samples; y i is the actual label value of the i-th sample;

[0182] After the model training is completed, its classification effect is evaluated through cross-validation and ROC curve to reveal the patterns in the data;

[0183] For clustering problems, K-means clustering is used. The goal is to find the best clustering division of the data by minimizing the Euclidean distance from each sample x point to the center of the cluster to which it belongs:

[0184]

[0185] Among them, k is the number of categories, S i represents the i-th class sample set, μ i is the center of the i-th cluster, and convergence is achieved by iteratively updating the center point and redistributing samples to clusters;

[0186] In the regression analysis, linear regression is used and the expression of the model is:

[0187] y=β0+β1X1+β2X2+…β n X n +ε

[0188] Where ε is the error term; the coefficients are estimated by minimizing the residual sum of squares using the least squares method.

[0189]

[0190] in, is the predicted label value of the i-th sample;

[0191] To avoid overfitting, a regularization method is used to add an L2 regularization term to the loss function:

[0192]

[0193] Among them, λ is the regularization parameter;

[0194] The complexity of the model is controlled by regularization, the generalization ability of the model is improved, and the mean square error is used to evaluate the fitting effect of the model;

[0195] For association rule mining, the Apriori model is used to mine frequent item sets and association rules from large-scale data sets through support and confidence indicators. By mining rules with high confidence and support, potential patterns and trends between data items are discovered.

[0196] The data analysis includes: using a big data processing framework to perform distributed processing on large-scale data sets. First, massive data is stored on multiple computing nodes through the Hadoop distributed file system, and then the MapReduce programming model is used to perform distributed processing and calculation of the data. For complex machine learning tasks, the Spark MLlib library is used to implement the training and prediction of distributed machine learning models. The entire process includes data sharding, parallel computing, task scheduling, and result merging, thereby achieving efficient processing and analysis of large-scale data sets.

[0197] Those skilled in the art know that, in addition to implementing the system, device and its various modules provided by the present invention in a purely computer-readable program code, it is entirely possible to implement the same program in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers and embedded microcontrollers by logically programming the method steps. Therefore, the system, device and its various modules provided by the present invention can be considered as a hardware component, and the modules included therein for implementing various programs can also be considered as structures within the hardware component; the modules for implementing various functions can also be considered as both software programs for implementing the method and structures within the hardware component.

[0198] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. An application method of a personal safety portrait model for a metallurgical safety enterprise, characterized in that: include: Step 1: Collect safety-related information data of personnel in metallurgical enterprises; Step 2: Clean the collected information data using preset data cleaning rules, including data integrity check, data consistency verification, duplicate data identification, error and outlier detection, data type and range verification, reference data verification, text data standardization, and privacy and sensitive information processing; Step 3: Establish personal security profile rules based on big data analysis, where the personal security profile rules include basic personnel information, behavior data, and assessment data; Step 4: Use the preset portrait labeling rules to label one or more labels for individuals, which cover basic personnel information, safety certification information, safety education and training, risk identification and control, hidden danger investigation and management, and safety behavior performance; Step 5: Obtain information about people of concern through tags to provide a basis for the company's security management.

2. The application method of the personal safety portrait model of metallurgical safety enterprises according to claim 1 is characterized in that: The data cleaning process includes data review, rule definition, tool selection, data preprocessing, data verification, data conversion, data audit, documentation, and continuous monitoring; The big data analysis process includes data collection, data storage, data cleaning and preprocessing, data integration, data modeling, data analysis, data visualization, deployment and application, monitoring and optimization, and security and compliance checks.

3. The application method of the personal safety portrait model of metallurgical safety enterprises according to claim 2 is characterized in that: The data modeling includes: In logistic regression, the model is of the form: Among them, P(y=1|X) represents the probability that the dependent variable y takes the value of 1 under the condition of a given independent variable X; β0 is the intercept term; β1, β2...β n are regression coefficients; X1, X2, ...X n is the independent variable; n is the number of independent variables; The parameters β are estimated by maximizing the log-likelihood function and the weights are optimized using the gradient descent algorithm: Where N is the number of samples; y i is the actual label value of the i-th sample; After the model training is completed, its classification effect is evaluated through cross-validation and ROC curve to reveal the patterns in the data; For clustering problems, K-means clustering is used. The goal is to find the best clustering division of the data by minimizing the Euclidean distance from each sample x point to the center of the cluster to which it belongs: Among them, k is the number of categories, S i represents the i-th class sample set, μ i is the center of the i-th cluster, and convergence is achieved by iteratively updating the center point and redistributing samples to clusters; In the regression analysis, linear regression is used and the expression of the model is: y=β0+β1X1+β2X2+…βnX,+ε Where ε is the error term; the coefficients are estimated by minimizing the residual sum of squares using the least squares method. in, is the predicted label value of the i-th sample; To avoid overfitting, a regularization method is used to add an L2 regularization term to the loss function: Among them, λ is the regularization parameter; The complexity of the model is controlled by regularization, the generalization ability of the model is improved, and the mean square error is used to evaluate the fitting effect of the model; For association rule mining, the Apriori model is used to mine frequent item sets and association rules from large-scale data sets through support and confidence indicators. By mining rules with high confidence and support, potential patterns and trends between data items are discovered.

4. The application method of the personal safety portrait model of metallurgical safety enterprises according to claim 2 is characterized in that: The data analysis includes: using a big data processing framework to perform distributed processing on large-scale data sets. First, massive data is stored on multiple computing nodes through the Hadoop distributed file system, and then the MapReduce programming model is used to perform distributed processing and calculation of the data. For complex machine learning tasks, the Spark MLlib library is used to implement the training and prediction of distributed machine learning models. The entire process includes data sharding, parallel computing, task scheduling, and result merging, thereby achieving efficient processing and analysis of large-scale data sets.

5. An application system of a personal safety portrait model for metallurgical safety enterprises, characterized in that: include: Module M1: Collecting safety-related information data of personnel in metallurgical enterprises; Module M2: The collected information data is cleaned according to the preset data cleaning rules, including data integrity check, data consistency verification, duplicate data identification, error and outlier detection, data type and range verification, reference data verification, text data standardization, and privacy and sensitive information processing; Module M3: Establishing personal security profile rules based on big data analysis, the personal security profile rules include basic personnel information, behavior data and assessment data; Module M4: Use the preset portrait labeling rules to label one or more labels for individuals, which cover basic information of personnel, safety certification information, safety education and training, risk identification and control, hidden danger investigation and management, and safety behavior performance; Module M5: Obtain information on people of concern through tags to provide a basis for the company's security management.

6. The application system of the personal safety portrait model of metallurgical safety enterprises according to claim 5 is characterized in that: The data cleaning process includes data review, rule definition, tool selection, data preprocessing, data verification, data conversion, data audit, documentation, and continuous monitoring; The big data analysis process includes data collection, data storage, data cleaning and preprocessing, data integration, data modeling, data analysis, data visualization, deployment and application, monitoring and optimization, and security and compliance checks.

7. The application system of the personal safety portrait model of metallurgical safety enterprises according to claim 6 is characterized in that: The data modeling includes: In logistic regression, the model is of the form: Among them, P(y=1|X) represents the probability that the dependent variable y takes the value of 1 under the condition of a given independent variable X; β0 is the intercept term; β1, β2...β n are regression coefficients; X1, X2, ...X n is the independent variable; n is the number of independent variables; The parameters β are estimated by maximizing the log-likelihood function and the weights are optimized using the gradient descent algorithm: Where N is the number of samples; y i is the actual label value of the i-th sample; After the model training is completed, its classification effect is evaluated through cross-validation and ROC curve to reveal the patterns in the data; For clustering problems, K-means clustering is used. The goal is to find the best clustering division of the data by minimizing the Euclidean distance from each sample x point to the center of the cluster to which it belongs: Among them, k is the number of categories, S i represents the i-th class sample set, μ i is the center of the i-th cluster, and convergence is achieved by iteratively updating the center point and redistributing samples to clusters; In the regression analysis, linear regression is used and the expression of the model is: y=β0+β1X1+β2X2+…βnX,+ε Where ε is the error term; the coefficients are estimated by minimizing the residual sum of squares using the least squares method. in, is the predicted label value of the i-th sample; To avoid overfitting, a regularization method is used to add an L2 regularization term to the loss function: Among them, λ is the regularization parameter; The complexity of the model is controlled by regularization, the generalization ability of the model is improved, and the mean square error is used to evaluate the fitting effect of the model; For association rule mining, the Apriori model is used to mine frequent item sets and association rules from large-scale data sets through support and confidence indicators. By mining rules with high confidence and support, potential patterns and trends between data items are discovered.

8. The application system of the personal safety portrait model of metallurgical safety enterprises according to claim 6 is characterized in that: The data analysis includes: using a big data processing framework to perform distributed processing on large-scale data sets. First, massive data is stored on multiple computing nodes through the Hadoop distributed file system, and then the MapReduce programming model is used to perform distributed processing and calculation of the data. For complex machine learning tasks, the Spark MLlib library is used to implement the training and prediction of distributed machine learning models. The entire process includes data sharding, parallel computing, task scheduling, and result merging, thereby achieving efficient processing and analysis of large-scale data sets.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for applying the personal safety portrait model of a metallurgical safety enterprise described in any one of claims 1 to 4 are implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer program is executed by a processor, the steps of the method for applying the personal safety portrait model of a metallurgical safety enterprise described in any one of claims 1 to 4 are implemented.

Images