A deep learning-based distributed farm bioaerosol simulation and health risk early warning method
By using distributed sensor networks and deep learning models, the concentration of bioaerosols can be monitored and warned in real time, which solves the problem of low efficiency of traditional methods, realizes dynamic analysis of bioaerosols and health risk assessment, and improves the environmental safety of farms.
Patent Information
- Application Number
- CN202411881482.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-19
- Publication Date
- 2025-12-05
- Estimated Expiration
- 2044-12-19
AI Technical Summary
Traditional bioaerosol monitoring methods are time-consuming and inefficient, unable to reflect dynamic changes in real time, and unable to effectively assess their health impact on farms and surrounding communities.
A distributed sensor network is constructed to collect environmental data from multiple regions in real time. A distributed model of bioaerosol concentration and a health risk early warning model are built using deep learning methods. Real-time monitoring and early warning are then performed using 3D visualization units.
It enables real-time acquisition and dynamic analysis of bioaerosol concentration, accurately identifies potential threats, optimizes emergency early warning mechanisms, reduces the risk of pathogen spread, and improves the environmental safety of farms.
Smart Images
Figure CN119808564B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of farm environment monitoring and health risk assessment, and particularly relates to a distributed farm bioaerosol simulation and health risk early warning method based on deep learning. BACKGROUND
[0002] With the continuous expansion of the scale and intensification of the farming industry, the bioaerosol concentration in the farm and its surrounding environment has increased significantly. Bioaerosols are rich in various microorganisms, including bacteria, fungi, viruses, and allergens, which not only pose a potential threat to the air quality of the farm and the public health of the surrounding community, but also affect the health of animals, leading to respiratory diseases, allergic reactions, and other health problems in humans. Therefore, it is particularly important to build an efficient monitoring and early warning system to assess the changes in bioaerosol concentration and their impact on health in a timely manner.
[0003] Traditional monitoring methods rely heavily on periodic sample collection and laboratory analysis, which is not only time-consuming and inefficient, but also cannot reflect the dynamic changes of bioaerosols in real time. To address this challenge, it is urgent to systematically analyze the sources, concentration changes, and microbial components of bioaerosols, and dynamically update them according to actual conditions. This system will provide strong support for early warning and management of the impact of bioaerosols on air quality and public health, lay a scientific foundation for the sustainable development of the farming industry, and help reduce resource waste, reduce economic losses for enterprises, and achieve more precise local prevention and control measures. SUMMARY
[0004] The purpose of the present application is to provide a distributed farm bioaerosol simulation and health risk early warning method based on deep learning, which solves the problems in the background art.
[0005] To achieve the above-mentioned purpose, the technical solution adopted by the present application is as follows:
[0006] A distributed farm bioaerosol concentration simulation and health risk early warning method based on deep learning, comprising the following steps:
[0007] A distributed sensor network is constructed and deployed to collect real-time environmental data of multi-region, multi-point farms, sample bioaerosol data, and obtain bioaerosol concentration data;
[0008] The collected data is standardized, the standardized data is formed into a structured data set and stored in a distributed database, the data set is preprocessed, and the data is divided into a training set and a validation set;
[0009] The biological aerosol concentration distributed model is constructed based on a deep learning method, a training set is input into the biological aerosol concentration distributed model for training, and the biological aerosol concentration distributed model is verified through a verification set. The model after verification can simulate the concentration distribution of biological aerosols in different regional farms and dynamically predict the diffusion behavior of biological aerosols in the time and space dimensions.
[0010] A health risk early warning model is constructed based on a deep learning method, biological aerosol concentrations in a data set are comprehensively analyzed, health risk levels are classified according to analysis results, risk level thresholds are set, results of simulation and prediction of the biological aerosol concentration distributed model are input into the health risk early warning model for health risk analysis, and if the results exceed the thresholds, a health risk early warning is automatically triggered, and if the prediction results are within a safe threshold, no alarm is triggered.
[0011] The simulation and prediction results of the biological aerosol concentration distributed model and the generated health risk levels and early warning information of the health risk early warning model are generated into visual 3D graphs through a central processing control and a 3D visualization unit.
[0012] Further, the standardization processing of the collected data includes removing outliers, normalization processing, and data type unification. The standardized data forms a structured data set and is stored in a distributed database, ensuring the reliability of data storage and cross-regional sharing capability.
[0013] Further, the pre-processing of the data set includes using a random forest algorithm combined with statistical methods for anomaly detection and processing, accurately identifying and removing noise data by analyzing data characteristics and distributed patterns, and using interpolation, mean filling, or deep learning-based prediction completion methods to process missing values.
[0014] Further, the environmental data includes but is not limited to temperature, humidity, air pressure, atmospheric pollutant concentration, and location information.
[0015] Further, the risk thresholds in the health risk early warning model are set based on regulatory standards, industry specifications, and historical risk data.
[0016] Further, the distributed database can be periodically updated to optimize and adjust the biological aerosol concentration distributed model and the health risk early warning model.
[0017] Further, for comprehensive analysis of the biological aerosol concentration in the data set, DNA / RNA extraction is performed on the samples in the data set by using an automated extraction device, sequence information is obtained by a high-throughput sequencing method, and sample images are collected by a microscopic image processing module, multi-dimensional feature extraction is performed on the images, and the extracted DNA / RNA sequence and image feature data are input into a health risk warning model, pathogenic bacteria data are compared with a regional epidemic disease database, pathogenicity is analyzed, and a health risk level is evaluated.
[0018] Further, the health risk warning is achieved by generating risk warning information and sending it to the device of the administrator to remind him to take countermeasures.
[0019] Further, the visual 3D graph includes generating dynamic heat maps, trend graphs, and risk level graphs.
[0020] Further, the central processing control and 3D visualization unit can control the external information acquisition unit, store historical data and real-time environmental data, and record biological aerosol data.
[0021] Beneficial effects:
[0022] The present application provides a distributed farm biological aerosol simulation and health risk warning method based on deep learning. The method can realize real-time acquisition of biological aerosol data in multiple regions and complex environments through a distributed data acquisition network, and comprehensively cover key information of different farm environments. Through a distributed architecture, multi-point data is aggregated and input into a deep learning model for dynamic analysis and simulation, thereby more accurately evaluating the concentration distribution and propagation trend of biological aerosol. The present application can dynamically adjust the health risk assessment model by combining deep learning technology, accurately identify potential threats, and optimize the emergency warning mechanism. When the risk exceeds the threshold, through real-time analysis and distributed control modules, the on-site equipment is linked to scientifically regulate the ventilation system to effectively suppress the spread of pathogens. By providing early warning and scientific decision support to farm managers, the present application significantly reduces the risk of pathogen transmission and significantly improves the safety level of the farm environment, providing an intelligent and systematic solution to protect the health of workers and animals. BRIEF DESCRIPTION OF DRAWINGS
[0023] Figure 1 is a flowchart of the present application. DETAILED DESCRIPTION
[0024] The present application will be further described in detail below in combination with the drawings and specific embodiments.
[0025] EMBODIMENT
[0026] A deep learning-based distributed farm bioaerosol concentration simulation and health risk early warning method, as shown in Figure 1 includes the following steps:
[0027] S1, a distributed sensor network is constructed and deployed, and environmental data of multi-region, multi-point farms are collected in real time, bioaerosol data are sampled, and bioaerosol concentration data are obtained.
[0028] The environmental data includes but is not limited to temperature, humidity, air pressure, atmospheric pollutant concentration (such as PM2.5, PM10, etc.) and location information. By constructing a distributed sensor network, the data of each region can be collected and aggregated in real time at multiple points, and the complexity of multi-region, multi-environmental characteristics can be fully covered.
[0029] S2, the collected data is standardized, the standardized data is formed into a structured data set and stored in a distributed database, the data set is preprocessed, and the data is divided into a training set and a validation set.
[0030] The collected data is standardized, which includes removing outliers, normalization processing and data type unification. The standardized data is formed into a structured data set and stored in a distributed database, ensuring the reliability and cross-regional sharing ability of data storage. The pre-processing of the data set includes using random forest algorithm combined with statistical method for anomaly detection and processing, identifying and removing noise data by analyzing data characteristics and distributed mode, and using interpolation method, mean filling or deep learning-based prediction completion method to process missing values. Ensure the integrity and consistency of the data, so as to provide high-quality data input for subsequent model training and dynamic health risk assessment. Feature extraction is also needed on the preprocessed data, and key feature variables such as temperature and humidity are used as input features of the model to provide high correlation data for subsequent bioaerosol concentration distribution simulation.
[0031] S3, a deep learning method is used to construct a bioaerosol concentration distributed model, the training set is input into the bioaerosol concentration distributed model for training, and the validation set is used to verify the bioaerosol concentration distributed model. The verified model can simulate the concentration distribution of bioaerosol in different regional farms and dynamically predict the diffusion behavior of bioaerosol in time and space dimensions.
[0032] The biological aerosol concentration distribution simulation is realized by the biological aerosol concentration distribution model, multi-source and multi-dimensional data are processed, spatial information is fully utilized, and the accuracy and expansibility of the concentration simulation are improved. Specifically, the model analyzes the collected data in depth, calculates the spatio-temporal distribution of biological aerosol concentration and its dynamic change rule. All data and analysis results are stored in a dynamic database, and combined with the regular collection and feedback mechanism of the distributed architecture, the real-time updating and efficient synchronization of the data are ensured. Distributed processing not only improves the adaptability to complex environments in multiple regions, but also significantly enhances the efficiency of data calculation and feedback, laying a solid foundation for health risk management of the farm, and providing strong technical support for precise early warning and automatic control.
[0033] In the verification of the biological aerosol concentration distribution model, cross-regional and multi-scenario verification is carried out through distributed data sets, and the model accuracy and prediction effect are evaluated based on multiple evaluation indexes such as mean square error and mean absolute error. The model that passes the verification is used for real-time or periodic prediction of the spatial distribution of biological aerosol concentration in the distributed farm.
[0034] S4, a health risk early warning model is constructed based on a deep learning method, the biological aerosol concentration in the data set is comprehensively analyzed, the health risk level is divided according to the analysis result, the risk level threshold is set, the simulation and prediction results of the biological aerosol concentration distribution model are input into the health risk early warning model for health risk analysis, and if the result exceeds the threshold, the health risk early warning is automatically triggered, and if the prediction result is within the safety threshold, the alarm is not triggered. The health risk early warning model dynamically adjusts the parameters of the model through a deep learning algorithm to adapt to the complexity of the environment in different regional farms, improve the accuracy and applicability of the prediction.
[0035] The health risk level is generated by a deep learning algorithm. Based on multi-point monitoring data, diffusion model results and real-time environmental parameters, the health risk is dynamically quantified. The dynamic early warning threshold is set by health risk assessment. When the health risk level exceeds the early warning threshold, the system will automatically trigger the health risk early warning. The early warning system generates risk warning information in real time and sends it to the devices of the management personnel, reminding them to take measures such as increasing ventilation and improving protective equipment to effectively reduce the potential health risk. If the system simulates the biological aerosol concentration within the safety threshold, the interface displays the safe state, and the system continues to dynamically monitor the changes of the biological aerosol concentration in the air and the environmental parameters, but does not trigger the alarm.
[0036] S5, the simulation and prediction results of the biological aerosol concentration distribution model and the generated health risk level and early warning information of the health risk early warning model are generated into a visual 3D graph by the central processing control and 3D visualization unit.
[0037] The central processing control and 3D visualization analysis unit can control the external information acquisition unit, combine the bioaerosol concentration simulation results with the environmental layout of different regional farms, and generate dynamic heat maps, trend charts, and risk level charts based on 3D visualization. Users can view the bioaerosol diffusion path, risk hot zone in real time through the interactive interface, and adjust the prevention and control measures combined with the early warning information. In addition, historical data and real-time environmental data can be stored, and bioaerosol data can be recorded. Through the 3D visualization unit, key data such as bioaerosol concentration and health risk warning are displayed intuitively, helping users to fully grasp the dynamic information of air quality and health risk. The system uses various 3D visualization forms such as heat maps, trend charts, and risk classification charts to show the spatial distribution of bioaerosols in different regional farms, concentration change trends, and health risk levels. The heat map directly presents the concentration distribution of bioaerosols in different regions, helping to identify high-risk areas; the trend chart shows the time trend of bioaerosol concentration, supporting risk trend prediction and early prevention. The risk classification chart compares the real-time health risk index with the safety threshold, allowing users to clearly understand the current health risk level. When the detected or simulated concentration level exceeds the warning threshold, the system displays the warning information in the interface and proposes countermeasures such as increasing ventilation, helping users to respond quickly.
[0038] Preferably, the risk threshold in the health risk warning model is set based on regulatory standards, industry specifications, and historical risk data. According to environmental regulations and health standards, a threshold system for air quality compliance is established to provide a basis for subsequent air quality judgment.
[0039] Preferably, the distributed database can be periodically updated to optimize and adjust the bioaerosol concentration distributed model and the health risk warning model.
[0040] Preferably, for comprehensive analysis of bioaerosol concentration in the data set, DNA / RNA extraction is performed on the samples in the data set using automated extraction equipment, and sequence information is obtained through high-throughput sequencing methods. At the same time, a microscopic image processing module is used to collect sample images, and multi-dimensional feature extraction is performed on the images, including morphological features and fluorescence features. Then the extracted DNA / RNA sequences and image feature data are input into the health risk warning model, and the pathogenic bacteria data are compared with the regional epidemic disease database to analyze pathogenicity and evaluate the health risk level.
[0041] Based on the health risk warning model, bioaerosol data samples and environmental data are input into the model, the model is pre-trained, the health risk level is calculated, a health risk report is generated, and an emergency response is triggered. When the risk value exceeds the set threshold, real-time warning is triggered and a health risk prompt is sent to the user, supporting the expansion needs of large-scale multi-point monitoring.
[0042] The deep learning-based distributed farm bioaerosol concentration simulation and health risk early warning method in this embodiment is based on environmental information data of different regional farms, provides multi-dimensional input for the model through distributed collection and processing, and constructs two types of core models, i.e., a bioaerosol concentration distributed model and a health risk early warning model, by using a deep learning method. The bioaerosol concentration distributed model simulates the concentration distribution of bioaerosols in different regional farms and dynamically predicts the diffusion behavior of bioaerosols in the time and space dimensions by fusing meteorological data, regional characteristics and monitoring data obtained by a distributed sensor network. The health risk early warning model comprehensively analyzes bioaerosol concentration and biological property data to assess the potential health risks of bioaerosols to farm workers and animals. The distributed application enables real-time aggregation and processing of regional data, thereby more comprehensively depicting the diffusion law and health risks of aerosols in complex environments. When the risk level exceeds the set threshold, the ventilation equipment distributed in the farm can be linked to dynamically adjust the wind speed and direction and accurately control the diffusion path of pollutants. Through periodic database updates, the latest monitoring data collected by the distributed collection are continuously introduced to optimize the model parameters and prediction accuracy, forming a closed-loop management mechanism from data collection, model optimization to risk feedback. The distributed architecture not only enhances the response speed and prediction reliability, but also flexibly adapts to the needs of farms of multiple scenarios and scales, providing efficient and intelligent risk early warning and control solutions for modern farms and comprehensively improving the health management level. In addition, a 3D visualization analysis unit is used to generate real-time updated bioaerosol concentration distribution maps and risk propagation path maps to intuitively show high-risk areas and their diffusion trends to users and provide scientific decision-making basis. The 3D visualization unit is embedded with a social and economic impact analysis tool to comprehensively evaluate factors such as environmental governance costs, health maintenance costs and production losses, calculate the social and economic benefits of current management measures, and generate an evaluation report. The report will be stored for a long time to provide a reference for future management optimization, helping managers make more scientific resource allocation and scheme optimization, and achieving an effective balance between economic benefits and health safety.
[0043] Although the embodiments of the present application are described in the specification, these embodiments are only as a hint and should not limit the protection scope of the present application. Various omissions, substitutions and changes made within the scope of the present application should be included in the protection scope of the present application.
Claims
1. A deep learning-based distributed farm bioaerosol concentration simulation and health risk early warning method, characterized in that, The method comprises the following steps: A distributed sensor network is constructed and deployed to collect environmental data of multi-area and multi-point breeding farms in real time, sample bioaerosol data, and obtain bioaerosol concentration data; The collected data is standardized, the standardized data is formed into a structured data set and stored in a distributed database, the data set is preprocessed, and the data is divided into a training set and a validation set; A bioaerosol concentration distributed model is constructed based on a deep learning method, the training set is input into the bioaerosol concentration distributed model for training, and the bioaerosol concentration distributed model is verified through the validation set; the verified model can simulate the concentration distribution of bioaerosol in different regional breeding farms and dynamically predict the diffusion behavior of bioaerosol in the time and space dimensions; A health risk early warning model is constructed based on a deep learning method, the bioaerosol concentration in the data set is comprehensively analyzed, the health risk level is divided according to the analysis result, the risk level threshold is set, the simulation and prediction results of the bioaerosol concentration distributed model are input into the health risk early warning model for health risk analysis, if the result exceeds the threshold, the health risk early warning is automatically triggered, if the prediction result is within the safety threshold, the alarm is not triggered; for the comprehensive analysis of the bioaerosol concentration in the data set, DNA / RNA extraction is performed on the samples in the data set by using an automatic extraction device, sequence information is obtained by using a high-throughput sequencing method, and sample images are collected by using a microscopic image processing module, multi-dimensional feature extraction is performed on the images, and the extracted DNA / RNA sequence and image feature data are input into the health risk early warning model, the pathogenic bacteria data are compared with a regional epidemic disease database, the pathogenicity is analyzed, and the health risk level is evaluated; The simulation and prediction results of the bioaerosol concentration distributed model and the generated health risk level and early warning information of the health risk early warning model are generated into a visual 3D graph by a central processing control and 3D visualization unit.
2. The deep learning-based distributed farm bioaerosol concentration simulation and health risk early warning method according to claim 1, characterized in that, The standardization of the collected data includes removing outliers, normalization processing, and data type unification, the standardized data forms a structured data set and is stored in a distributed database, and the reliability of data storage and cross-regional sharing capability are ensured.
3. The deep learning-based distributed farm bioaerosol concentration simulation and health risk early warning method according to claim 1, characterized in that, The preprocessing of the data set includes using a random forest algorithm combined with a statistical method for anomaly detection and processing, accurately identifying and removing noise data by analyzing data features and distributed patterns, and processing missing values by using an interpolation method, mean filling, or a deep learning-based prediction completion method.
4. The deep learning-based distributed farm bioaerosol concentration simulation and health risk early warning method according to claim 1, characterized in that, The environmental data includes temperature, humidity, air pressure, atmospheric pollutant concentration, and location information.
5. The deep learning-based distributed farm bioaerosol concentration simulation and health risk early warning method according to claim 1, characterized in that, The risk threshold in the health risk early warning model is set based on regulatory standards, industry specifications, and historical risk data.
6. The deep learning-based distributed farm bioaerosol concentration simulation and health risk early warning method according to claim 1, characterized in that, The distributed database can be periodically updated to optimize and adjust the bioaerosol concentration distributed model and the health risk early warning model.
7. The deep learning-based distributed farm bioaerosol concentration simulation and health risk early warning method according to claim 1, characterized in that, The health risk early warning generates risk warning information and sends it to the device of the management personnel to remind them to take countermeasures.
8. The deep learning-based distributed farm bioaerosol concentration simulation and health risk early warning method according to claim 1, characterized in that, The visualized 3D graph includes a dynamic heat map, a trend graph and a risk level graph.
9. The distributed deep learning-based bioaerosol concentration simulation and health risk early warning method for livestock farms according to any one of claims 1 to 8, characterized in that, The central processing control and 3D visualization unit can control the external information acquisition unit, store historical data and real-time environmental data, and record bioaerosol data.