Occupational hazard comprehensive risk assessment and early warning method based on multi-source heterogeneous data

Abstract

The invention relates to an occupational hazard comprehensive risk assessment and early warning method based on multi-source heterogeneous data, and belongs to the field of occupational health and occupational disease control. The method specifically comprises the following steps: constructing a pneumoconiosis risk index system and an early warning model, a radioactive risk index system and an evaluation model, and a noise hazard index system and an evaluation model of an operator; then, in combination with other physical factor index systems and evaluation models, a multi-source heterogeneous comprehensive evaluation or early warning model is established; occupational environment monitoring data, occupational health monitoring data, occupational disease data and occupational health supervision data are collected or collected; dividing the resource data into dust, radioactivity, industrial noise and other physical factors, and constructing a multi-source heterogeneous data occupational hazard prevention and control decision support big data center; and then a multi-dimensional indexing technology for occupational hazard data is designed by using a data distributed acquisition and storage technology, and an extensible occupational hazard early warning support platform with high performance and high fault tolerance is constructed.

Description

Technical Field

[0001] The present invention belongs to the field of occupational health and occupational disease control, and relates to a comprehensive risk assessment and early warning method for occupational hazards based on multi-source heterogeneous data. Background Art

[0002] Occupational hazards refer to factors that may harm the health or life of workers in the working environment. Occupational hazards include dust, radioactivity, industrial noise and other physical factors. In order to protect the health and life of workers, employers need to formulate occupational hazard prevention measures and management plans.

[0003] High-risk dust operations refer to operations in which workers of an employer are exposed to asbestos or dust with a free silica content of more than 10% during their occupational activities, or operations in which they are exposed to other dusts and have a high risk of developing pneumoconiosis. Pneumoconiosis is one of the statutory occupational diseases. It is a systemic disease characterized by diffuse fibrosis of the lung tissue caused by long-term inhalation of productive dust by workers during their occupational activities and its deposition in the lungs. High-risk dust is a serious hazard to workers. "Rapid, mass, and high-incidence" pneumoconiosis is basically caused by high-risk dust, with high mortality and disability rates. By monitoring and early warning of high-risk dust risks, pneumoconiosis caused by high-risk dust is controlled, which basically solves the problem of high incidence of pneumoconiosis.

[0004] Radioactive incidents refer to emergencies caused by leakage, accidents, etc. of nuclear radiation or other radioactive materials. Such incidents are highly unpredictable and dangerous, and have a significant impact on life, property and the environment. In order to protect the safety of the public and reduce possible losses, it is particularly important to formulate and implement a scientific and complete radioactive incident emergency plan.

[0005] Industrial noise refers to the noise generated by mechanical vibration, friction, impact and air flow disturbance during the production process of the factory. For example, the noise generated by the air compressor, blower and boiler exhaust venting in the chemical plant is air flow noise caused by air vibration. The noise generated by ball mills, crushers and looms is mechanical noise caused by mechanical vibration or friction and impact of solid parts. Since the sources of industrial noise are numerous and scattered, the noise types are relatively complex, and the continuous sound sources of production are also difficult to identify, making it quite difficult to control.

[0006] Therefore, there is an urgent need for a method that can comprehensively evaluate and warn of multiple occupational hazards. Summary of the invention

[0007] In view of this, the purpose of the present invention is to provide a comprehensive risk assessment and early warning method for occupational hazards based on multi-source heterogeneous data, so as to achieve effective early warning of potential hazard risks, thereby reducing the occupational hazards brought to workers by the job.

[0008] In order to achieve the above object, the present invention provides the following technical solutions:

[0009] A comprehensive risk assessment and early warning method for occupational hazards based on multi-source heterogeneous data specifically includes the following steps:

[0010] S1: Collect and analyze occupational disease data such as pneumoconiosis, radiation hazards and noise hazards, and establish a risk indicator system and early warning model for pneumoconiosis among workers, a radioactive risk indicator system and early warning model, and a noise hazard indicator system and early warning model;

[0011] S2: Combine the three models established in step S1 with other physical factor indicator systems and evaluation models to establish a multi-source heterogeneous comprehensive evaluation or early warning model;

[0012] S3: Collect or collect occupational environment monitoring data, occupational health monitoring data, occupational disease data, and occupational health supervision data; divide resource data into four types of occupational hazard databases, including dust, radioactivity, industrial noise, and other physical factors (temperature, air pressure, electromagnetic field, etc.); then use different computing frameworks to construct the four types of occupational hazard databases into a multi-source heterogeneous data occupational hazard prevention and control decision support big data center;

[0013] S4: Based on the occupational hazard prevention and control decision-making big data center built in step S3, use data distributed collection and distributed storage technology to design a multidimensional indexing technology for occupational disease hazard data; then, collect real-time data of occupational hazard sites through online monitoring equipment, combine it with static data, and establish an occupational hazard early warning support platform based on big data and cloud computing;

[0014] S5: Integrate the occupational hazard early warning support platform based on big data and cloud computing established in step S4 with the multi-source heterogeneous comprehensive assessment or early warning model established in step S3 to form a comprehensive risk assessment and early warning system for occupational hazards.

[0015] Furthermore, in step S1, the specific steps of establishing the risk indicator system and early warning model for pneumoconiosis of workers are as follows:

[0016] S101: Collect no less than 1,000 files of employees who have worked in dusty workplaces for more than 5 years, and extract test data on pneumoconiosis hazard factors and medical examination reports;

[0017] S102: Investigate or collect the relationship between the cumulative exposure to respirable dust and the incidence of pneumoconiosis, as well as the results of the pathogenic mechanism of pneumoconiosis at home and abroad. Use the Delphi method and hierarchical analysis method, combined with the survey data of relevant on-site engineers and workers, preliminarily screen out the corresponding factors of the pneumoconiosis risk index system, and conduct weight analysis on the main indicators and indicators at each level, mainly covering the following contents: Dust hazard properties: types of dust (coal dust, silica dust, cement dust, barite dust), free silica content, dust dispersion, workplace concentration, contact time; on-site production conditions: dust properties (composition), process, dust source control, dust prevention measures, dust reduction measures, equipment age, equipment loss, protective facility loss, protective facility effect (dust removal effect), real-time monitoring concentration; personal protective measures; contact time: dust exposure years, average daily dust exposure operation time; dust exposure operation conditions: labor intensity, breathing volume; on-site management, etc.; individual factors: personal living habits, etc.;

[0018] S103: Through theoretical calculation and statistical analysis, the collected data on employee occupational health examinations, occupational disease diagnosis data and occupational disease hazard factor detection data in the workplace are classified and sorted; mathematical statistical analysis is performed on relevant indicators based on the data, and relevant connections are preliminarily explored; the preliminarily formed indicator system is revised to form an indicator system; experts are organized to demonstrate the indicator system, and based on the demonstration results, the indicator system is further revised to form a respiratory dust hazard warning indicator system;

[0019] S104: Apply regression analysis and simulation mathematics to conduct mathematical analysis on the collected data, and form a risk value for respiratory dust hazards by weighting the dust hazard weighted value, the cumulative exposure to respirable dust, and the weighted value of personal factors. Establish a pneumoconiosis early warning model to estimate the harm of respirable dust to the human body, and issue early warnings for specific types of work, so as to take corresponding measures to reduce the risk of disease or prolong the years of disease onset;

[0020] S105: Organize experts to demonstrate and explore the established model, discuss and revise it based on the expert feedback, and form an indicator system and early warning model for the risk of pneumoconiosis in workers that can effectively and scientifically predict the risk of pneumoconiosis in workers; combine the early warning model with the site to verify the scientificity and accuracy of the indicator system and early warning model for the risk of pneumoconiosis in workers.

[0021] Furthermore, in step S1, the specific steps of establishing a radioactive risk indicator system and early warning model are as follows:

[0022] S111: Collect measured data on radon concentration in mine tunnels and data on human hazard warning lines to form a radioactive hazard database; then use the segmented interpolation simulation method to study the prediction model of temperature and humidity in mine tunnels, and combine it with the radon concentration detection method to form a relationship model between radon concentration and temperature and humidity; iterate and apply the relationship model multiple times to improve the accuracy of radon concentration prediction and early warning;

[0023] S112: On this basis, with the help of radioactive occupational disease data, analytical methods are adopted, combined with relevant on-site investigation data, a radioactive risk indicator system is established, and weight analysis is conducted on the main indicators and indicators at all levels, mainly involving radon gas concentration, dose, duration, body factors, etc.; then radioactive occupational disease diagnosis data, hazard factor data, health examination data, radon gas and human hazard dose-response relationship, radon gas concentration, contact time, body factors, etc. and disease contribution rate (nervous system, respiratory system), etc. are collected to establish a radioactive risk indicator system and early warning model; the early warning model is combined with the site to verify the scientific nature of the early warning model.

[0024] Furthermore, in step S1, the specific steps of establishing the noise hazard index system and early warning model are as follows:

[0025] S121: Analyze the acoustic signals collected by each microphone array in the spherical harmonic domain to obtain the local sound field expansion coefficient corresponding to the array; analyze the influence of factors such as environmental noise on the local sound field expansion coefficient, and quantitatively estimate the environmental noise in the spherical harmonic domain; on this basis, use the local sound field expansion coefficient and local environmental noise information, and use the least squares method and other methods to synthesize the global sound field expansion coefficient and its corresponding environmental noise data to support sound field reconstruction and prediction; based on the global sound field expansion coefficient, study the sound source localization method in the spherical harmonic domain, so as to achieve accurate measurement of typical industrial equipment noise;

[0026] S122: On this basis, with the help of noise occupational disease data, using methods such as hierarchical analysis, combined with factors such as relevant on-site engineers and workers’ survey data, the corresponding factors of the noise hazard index system are established, and weight analysis is conducted on the main indicators and indicators at each level, mainly involving intensity, frequency, contact time, etc.; then regression analysis and other methods are used to collect noise occupational disease diagnosis data, hazard factor data, health examination data, the relationship model between noise and human health hazards (with the help of ISO 26639 standard issued by the International Organization for Standardization (ISO), "Environmental Noise and Health" guidelines and "Noise Occupational Disease Hazard Risk Management Guidelines" issued by the World Health Organization (WHO), noise intensity, frequency and contact time The disease contribution rate (hearing, nervous system, reproductive system, cardiovascular system), etc., to establish a noise hazard system and early warning model; combine the early warning model with the site to verify the scientific nature of the early warning model.

[0027] Further, in step S3, the specific steps of constructing a big data center for occupational hazard prevention and control decision support of multi-source heterogeneous data are as follows:

[0028] S301: Collect or collect data on occupational environment monitoring, occupational diseases, occupational health supervision and occupational disease prevention and control, and process and integrate multi-source heterogeneous data such as imaging data and online monitoring data according to the mode of distributed collection and cleaning, centralized integration and storage, and global application. Divide resource data into four types of occupational hazard databases: dust (chemical properties, physical properties, dust concentration in the workplace, cumulative exposure time, dust prevention measures, etc.), radioactivity (radiation type, dose, duration, body factors, etc.), industrial noise (intensity, frequency, contact time, etc.) and other physical factors (temperature, air pressure, electromagnetic field, etc.), design a data resource classification storage plan, and establish a theme-based occupational hazard warning information classification database;

[0029] S302: After the data is collected, integrated and stored, different computing frameworks need to be used according to the different characteristics of the data and tasks. It is planned to use a combination of MapReduce, Storm and Spark computing frameworks to process the data.

[0030] Furthermore, in step S302, the MapReduce computing framework is used to process massive historical data and offline data. MapReduce can be used to analyze low-latency and low-timeliness data, such as weekly and monthly statistical tasks; the Storm streaming computing framework is used to process and convert the continuously collected data streams of the data access layer, and perform other real-time statistics and analysis; the Spark memory computing framework is used for real-time analysis and iterative computing. Spark's distributed memory computing model can analyze and model massive data online.

[0031] Furthermore, in step S302, when performing data mining analysis and fault diagnosis based on the four types of occupational hazard databases, the traditional stand-alone mode obviously cannot meet the performance and timeliness requirements, so it is necessary to use Mahout based on MapReduce and Spark MLlib based on Spark. Through the built-in computing models and various machine learning algorithms of Mahout and MLlib, business personnel can easily perform various fault diagnosis modeling based on classification, clustering and time series association.

[0032] Further, in step S4, an occupational hazard early warning support platform based on big data and cloud computing is established, and the specific steps are as follows:

[0033] S401: Based on the occupational hazard prevention and control decision-making big data center, using data distributed collection and distributed storage technology, parallel database technology and Spark technology, adopting hierarchical design and structured development methods, designing multidimensional indexing technology for occupational disease hazard data, and developing an occupational hazard early warning support platform with scalability, high performance and high fault tolerance;

[0034] S402: Collect real-time data of occupational hazard sites through online monitoring equipment, combine it with static data, and establish an occupational hazard early warning support platform based on big data and cloud computing.

[0035] The beneficial effects of the present invention are as follows: the present invention constructs a big data center for occupational hazard prevention and control decision support of multi-source heterogeneous data, and then utilizes data distributed collection and distributed storage technology to design a multidimensional indexing technology for occupational disease hazard data, and designs a scalable, high-performance, and highly fault-tolerant occupational hazard early warning support platform. The model can effectively warn of potential hazard risks, thereby reducing the occupational hazards brought by the types of work to the operators.

[0036] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0038] Figure 1 It is an overall flow chart of the comprehensive risk assessment and early warning method of occupational hazards based on multi-source heterogeneous data of the present invention;

[0039] Figure 2 A flowchart for the construction of a risk indicator system and early warning model for pneumoconiosis among workers;

[0040] Figure 3 A flow chart for the construction of a comprehensive assessment model and early warning model for the radioactive risk indicator system;

[0041] Figure 4 A flowchart for constructing a noise hazard index system and early warning model for typical industrial equipment;

[0042] Figure 5 A flowchart for the construction of a big data center for occupational hazard prevention and control decision-making;

[0043] Figure 6This is a flowchart for building an occupational hazard early warning support platform based on big data and cloud computing. DETAILED DESCRIPTION

[0044] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0045] See also Figure 1 to Figure 6 ,The present invention provides a comprehensive risk assessment and early warning method for occupational hazards based on multi-source heterogeneous data, such as Figure 1 As shown in the figure, the overall idea of ​​the method is to collect occupational disease data such as pneumoconiosis, radioactive hazards, and noise hazards, complete the research on the early warning index system and early warning of pneumoconiosis for workers, comprehensive assessment and early warning of radioactive risks, and comprehensive assessment and early warning of noise for typical industrial equipment, and obtain the risk index system and early warning model of pneumoconiosis for workers, the radioactive risk index system and evaluation model, and the noise hazard index system and evaluation model respectively; then combine the index system and evaluation model of other physical factors to establish a multi-source heterogeneous comprehensive evaluation or early warning model; collect or collect occupational environment monitoring data, occupational health monitoring data, occupational disease data, and occupational health supervision data; divide resource data into four types of occupational hazard databases, including dust, radioactivity, industrial noise, and other physical factors (temperature, air pressure, electromagnetic field, etc.); on this basis, combine the databases of dust, radioactivity, noise, and other physical factors to build a multi-source heterogeneous data occupational hazard prevention and control decision support big data center; then use data distributed acquisition and distributed storage technology to design a multi-dimensional indexing technology for occupational disease hazard data, and develop an occupational hazard early warning support platform with scalability, high performance, and high fault tolerance.

[0046] like Figure 2 As shown in the figure, the process of constructing the risk indicator system and early warning model for pneumoconiosis in workers is as follows:

[0047] First, we collected files of no less than 1,000 employees who had worked in dusty workplaces for more than five years, and extracted detection data on pneumoconiosis hazard factors and medical examination reports.

[0048] Further investigation or collection of results on the relationship between cumulative exposure to respirable dust and pneumoconiosis, the pathogenic mechanism of pneumoconiosis at home and abroad, etc., the Delphi method, hierarchical analysis method and other methods were used, combined with factors such as the survey data of relevant on-site engineers and workers, to preliminarily screen out the corresponding factors of the pneumoconiosis risk index system, and conduct weight analysis on the main indicators and indicators at each level, mainly covering the following contents: Nature of dust hazards: types of dust (coal dust, silica dust, cement dust, barite dust), free silica content, dust dispersion, workplace concentration, contact time; on-site production conditions: dust properties (composition), process, dust source control, dust prevention measures, dust reduction measures, equipment age, equipment loss, protective facility loss, protective facility effect (dust removal effect), real-time monitoring concentration; personal protection measures; contact time: length of service in dust exposure, average daily dust exposure operation time; dust exposure operation conditions: labor intensity, breathing volume; on-site management, etc.; individual factors: personal living habits, etc.

[0049] Next, the collected data on occupational health examinations of employees, occupational disease diagnosis data, and occupational disease hazard factor detection data in the workplace were classified and sorted through theoretical calculations and statistical analysis. According to the data, mathematical statistical analysis was conducted on relevant indicators to preliminarily explore the relevant connections. The preliminary indicator system was revised to form an indicator system; experts were organized to demonstrate the indicator system, and based on the demonstration results, the indicator system was further revised to form a respiratory dust hazard warning indicator system.

[0050] Then, regression analysis, simulation mathematics and other methods were used to conduct mathematical analysis on the collected data. The respiratory dust hazard risk value was formed by comprehensive weighting of the dust hazard weighted value, the cumulative exposure to respirable dust, and the weighted value of personal factors. A pneumoconiosis early warning model was established to estimate the harm of respirable dust to the human body, issue early warnings for specific types of work, and take corresponding measures to reduce the risk of disease or prolong the years of disease onset.

[0051] Finally, experts were organized to discuss and explore the established model, and based on the feedback from experts, the model was discussed and revised to form an effective and scientific prediction model. The early warning model was combined with the site to verify the scientificity and accuracy of the risk indicator system for pneumoconiosis among workers and the early warning model.

[0052] like Figure 3 As shown in the figure, the process of constructing a comprehensive assessment model and early warning model for the radioactive risk indicator system is as follows:

[0053] First, the measured data of radon concentration in mine tunnels and data on warning lines for human hazards are collected to form a radioactive hazard database. Then, the segmented interpolation simulation method is used to study the temperature and humidity prediction model of mine tunnels. Combined with the radon concentration detection method, a relationship model between radon concentration and temperature and humidity is formed. The relationship model is iterated and applied multiple times to improve the accuracy of radon concentration prediction and early warning.

[0054] On this basis, with the help of radioactive occupational disease data, analytical methods are adopted, combined with relevant on-site investigation data, a radioactive risk indicator system is established, and weight analysis is conducted on the main indicators and indicators at all levels, mainly involving radon gas concentration, dose, duration, body factors, etc.; then radioactive occupational disease diagnosis data, hazard factor data, health examination data, radon gas and human hazard dose-response relationship, radon gas concentration, contact time, body factors, etc. and disease contribution rate (nervous system, respiratory system) are collected to establish a radioactive risk indicator system and early warning model; the early warning model is combined with the site to verify the scientific nature of the early warning model.

[0055] like Figure 4 As shown in the figure, the process of constructing the noise hazard index system and early warning model for typical industrial equipment is as follows:

[0056] The acoustic signals collected by each microphone array are analyzed in the spherical harmonic domain to obtain the local sound field expansion coefficient corresponding to the array; the influence of factors such as environmental noise on the local sound field expansion coefficient is analyzed, and the environmental noise is quantitatively estimated in the spherical harmonic domain. On this basis, the local sound field expansion coefficient and local environmental noise information are used to synthesize the global sound field expansion coefficient and its corresponding environmental noise data using methods such as the least squares method to support sound field reconstruction and prediction. Based on the global sound field expansion coefficient, the sound source localization method in the spherical harmonic domain is studied. This enables accurate measurement of typical industrial equipment noise.

[0057] On this basis, with the help of noise occupational disease data, hierarchical analysis and other methods, combined with relevant on-site engineers and workers survey data and other factors, the corresponding factors of the noise hazard index system are established, and the main indicators and indicators at each level are weighted, mainly involving intensity, frequency, contact time, etc.; then regression analysis and other methods are used to collect noise occupational disease diagnosis data, hazard factor data, health examination data, the relationship model between noise and human health hazards (with the help of ISO 26639 standard issued by the International Organization for Standardization (ISO), "Environmental Noise and Health" guidelines and "Noise Occupational Disease Hazard Risk Management Guidelines" issued by the World Health Organization (WHO), noise intensity, frequency, The contribution rate of illness (auditory, nervous system, reproductive system, cardiovascular system), etc. of contact time, etc., to establish a noise hazard system and early warning model; the early warning model is combined with the site to verify the scientific nature of the early warning model.

[0058] like Figure 5 As shown in the figure, the process of building a big data center for occupational hazard prevention and control decision support with multi-source heterogeneous data is as follows:

[0059] First, data on occupational environment monitoring, occupational diseases, occupational health supervision, occupational disease prevention and control are collected or collected. According to the mode of distributed collection and cleaning, centralized integration and storage, and global application, multi-source heterogeneous data such as imaging data and online monitoring data are processed and integrated for occupational hazard data. Resource data are divided into four types of occupational hazard databases: dust (chemical properties, physical properties, dust concentration in the workplace, cumulative exposure time, dust prevention measures, etc.), radioactivity (radiation type, dose, duration, body factors, etc.), industrial noise (intensity, frequency, contact time, etc.), and other physical factors (temperature, air pressure, electromagnetic field, etc.). A data resource classification and storage plan is designed to establish a theme-based occupational hazard warning information classification database.

[0060] Next, after the data is collected, integrated and stored, different computing frameworks are needed for different characteristics of data and tasks. We plan to use a combination of three computing frameworks: MapReduce, Storm and Spark. The MapReduce computing framework is good at processing massive historical data and offline data. Through MapReduce, we can analyze low-latency and low-timeliness data, such as weekly and monthly statistical tasks. The Storm streaming computing framework can process and convert the continuous collection of data streams at the data access layer, and perform other real-time statistics and analysis. The Spark memory computing framework is used for real-time analysis and iterative computing. Spark's distributed memory computing model can analyze and model massive data online.

[0061] Finally, when conducting data mining analysis and fault diagnosis based on the four occupational hazard databases, the traditional stand-alone mode obviously cannot meet the performance and timeliness requirements, so it is necessary to use Mahout based on MapReduce and Spark MLlib based on Spark. Through the built-in computing models and various machine learning algorithms of Mahout and MLlib, business personnel can easily perform various fault diagnosis modeling based on classification, clustering and time series association.

[0062] like Figure 6 As shown in the figure, the process of building an occupational hazard early warning support platform based on big data and cloud computing is as follows:

[0063] Based on the occupational hazard prevention and control decision-making big data center, using data distributed collection and distributed storage technology, parallel database technology, Spark technology, adopting layered design and structured development methods, designing multidimensional indexing technology for occupational disease hazard data, and developing a scalable, high-performance, and highly fault-tolerant occupational hazard early warning support platform; then, through online monitoring equipment, collect real-time data of occupational hazard sites, combine it with static data, and establish an occupational hazard early warning support platform based on big data and cloud computing; finally, integrate the obtained occupational hazard early warning indicator system and early warning model of dust, radioactivity, noise, etc. to form a comprehensive risk assessment and early warning technology system for occupational hazards.

[0064] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A comprehensive risk assessment and early warning method for occupational hazards based on multi-source heterogeneous data, characterized in that: The method specifically comprises the following steps: S1: Collect and analyze data on pneumoconiosis, radioactive hazards and noise hazards, and establish a risk indicator system and early warning model for pneumoconiosis among workers, a radioactive risk indicator system and early warning model, and a noise hazard indicator system and early warning model; S2: Combine the three models established in step S1 with other physical factor indicator systems and evaluation models to establish a multi-source heterogeneous comprehensive evaluation or early warning model; S3: Collect or collect occupational environment monitoring data, occupational health monitoring data, occupational disease data, and occupational health supervision data; divide resource data into four types of occupational hazard databases: dust, radioactivity, industrial noise, and other physical factors; then use different computing frameworks to construct the four types of occupational hazard databases into a multi-source heterogeneous data occupational hazard prevention and control decision support big data center; S4: Based on the occupational hazard prevention and control decision-making big data center built in step S3, use data distributed collection and distributed storage technology to design a multidimensional indexing technology for occupational disease hazard data; then, collect real-time data of occupational hazard sites through online monitoring equipment, combine it with static data, and establish an occupational hazard early warning support platform based on big data and cloud computing; S5: Integrate the occupational hazard early warning support platform based on big data and cloud computing established in step S4 with the multi-source heterogeneous comprehensive assessment or early warning model established in step S3 to form a comprehensive risk assessment and early warning system for occupational hazards.

2. The comprehensive risk assessment and early warning method for occupational hazards according to claim 1 is characterized in that: In step S1, the specific steps of establishing the risk indicator system and early warning model for pneumoconiosis of workers are as follows: S101: Collect employee files in dusty workplaces and extract pneumoconiosis hazard factor detection data and medical examination reports; S102: Investigate or collect information on the relationship between cumulative exposure to respirable dust and the incidence of pneumoconiosis, as well as the pathogenic mechanism of pneumoconiosis. Use the Delphi method and analytic hierarchy process, combined with survey data from relevant on-site engineers and workers, to preliminarily screen out the corresponding factors of the pneumoconiosis risk index system, and conduct weight analysis on the main indicators and indicators at each level; S103: Through theoretical calculation and statistical analysis, the collected data on employee occupational health examinations, occupational disease diagnosis data and workplace occupational disease hazard factor detection data are classified and sorted; mathematical statistical analysis of relevant indicators is carried out according to the data, and relevant connections are preliminarily explored; Revise the initially formed indicator system to form an indicator system; organize experts to demonstrate the indicator system, and further revise the indicator system based on the demonstration results to form a respirable dust hazard warning indicator system; S104: Apply regression analysis and simulation mathematical methods to conduct mathematical analysis on the collected data, and form a risk value for respiratory dust hazard by weighting the dust hazard weighted value, the cumulative exposure to respirable dust, and the weighted value of personal factors. Establish a pneumoconiosis early warning model to estimate the harm of respirable dust to the human body, and issue early warnings for specific types of work, so as to take corresponding measures to reduce the risk of disease or prolong the years of disease onset; S105: Organize experts to demonstrate and explore the established model, discuss and revise it based on the experts’ feedback, and form a risk indicator system and early warning model for pneumoconiosis in workers.

3. The comprehensive risk assessment and early warning method for occupational hazards according to claim 1 is characterized in that: In step S1, the specific steps of establishing the radioactive risk indicator system and early warning model are: S111: Collect measured data on radon concentration in mine tunnels and data on human hazard warning lines to form a radioactive hazard database; then use the segmented interpolation simulation method to study the prediction model of temperature and humidity in mine tunnels, and combine it with the radon concentration detection method to form a relationship model between radon concentration and temperature and humidity; iterate and apply the relationship model multiple times to improve the accuracy of radon concentration prediction and early warning; S112: With the help of radiation occupational disease data, analytical methods are adopted, combined with relevant on-site investigation information, a radiation risk indicator system is established, and weight analysis is performed on the main indicators and indicators at all levels; then radiation occupational disease diagnosis data, hazard factor data, health examination data, radon gas and human hazard dose-response relationship, radon gas concentration, contact time, body factors and disease contribution rate are collected to establish a radiation risk indicator system and early warning model.

4. The comprehensive risk assessment and early warning method for occupational hazards according to claim 1 is characterized in that: In step S1, the specific steps of establishing the noise hazard index system and early warning model are: S121: Analyze the acoustic signals collected by each microphone array in the spherical harmonic domain to obtain the local sound field expansion coefficient corresponding to the array; analyze the influence of environmental noise on the local sound field expansion coefficient, and quantitatively estimate the environmental noise in the spherical harmonic domain; on this basis, use the local sound field expansion coefficient and local environmental noise information to synthesize the global sound field expansion coefficient and its corresponding environmental noise data using the least squares method to support sound field reconstruction and prediction; Based on the global sound field expansion coefficient, the sound source localization method in the spherical harmonic domain is studied to achieve accurate measurement of noise of typical industrial equipment; S122: With the help of noise occupational disease data, the hierarchical analysis method is adopted, combined with the relevant on-site engineers and workers’ survey data, the corresponding factors of the noise hazard index system are established, and the weight analysis of the main indicators and indicators at each level is carried out; then the regression analysis method is applied to collect noise occupational disease diagnosis data, hazard factor data, health examination data, the relationship model between noise and human health hazards, noise intensity, frequency and exposure time The contribution rate of illness, establish a noise hazard system and early warning model.

5. The comprehensive risk assessment and early warning method for occupational hazards according to claim 1 is characterized in that: In step S3, the specific steps of constructing a big data center for occupational hazard prevention and control decision support of multi-source heterogeneous data are as follows: S301: Collect or collect occupational environment monitoring, occupational diseases, occupational health supervision and occupational disease prevention data, process and integrate imaging data and online monitoring data according to the mode of distributed collection and cleaning, centralized integration and storage, and global application, divide resource data into four types of occupational hazard databases: dust, radioactivity, industrial noise and other physical factors, design data resource classification storage plan, and establish a theme-based occupational hazard warning information classification database; S302: After the data is collected, integrated and stored, the three computing frameworks of MapReduce, Storm and Spark are combined to process the data according to the different characteristics of the data and tasks.

6. The comprehensive risk assessment and early warning method for occupational hazards according to claim 5 is characterized in that: In step S302, the MapReduce computing framework is used to process massive historical data and offline data. MapReduce can be used to analyze low-latency and low-timeliness data; the Storm computing framework is used to process and convert the continuous collection data stream of the data access layer, and perform other real-time statistics and analysis; Real-time analysis and iterative computing use the Spark memory computing framework. Spark's distributed memory computing model can analyze and model massive data online.

7. The comprehensive risk assessment and early warning method for occupational hazards according to claim 5 or 6, characterized in that: In step S302, when performing data mining analysis and fault diagnosis based on the four types of occupational hazard databases, Mahout based on MapReduce and Spark MLlib based on Spark are used. Through the built-in computing models and various machine learning algorithms of Mahout and MLlib, various fault diagnosis modeling based on classification, clustering and time series association is performed.

8. The comprehensive risk assessment and early warning method for occupational hazards according to claim 1 is characterized in that: In step S4, an occupational hazard early warning support platform based on big data and cloud computing is established. The specific steps are as follows: S401: Based on the occupational hazard prevention and control decision-making big data center, using data distributed collection and distributed storage technology, parallel database technology and Spark technology, adopting hierarchical design and structured development methods, design a multidimensional indexing technology for occupational disease hazard data; S402: Collect real-time data of occupational hazard sites through online monitoring equipment, combine it with static data, and establish an occupational hazard early warning support platform based on big data and cloud computing.

Images