Directional prediction and early warning method based on agricultural non-point source pollution treatment
Through the data collection, preprocessing, correlation analysis and early warning model establishment of agricultural non-point source pollution, a comprehensive prediction and early warning of agricultural non-point source pollution is achieved, and the problems of single and high cost of early warning in the existing methods are solved, and the early warning is improved Accuracy and agricultural production efficiency.
Patent Information
- Application Number
- CN202411236900.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-04
- Publication Date
- 2024-11-26
AI Technical Summary
The existing early warning methods for agricultural non-point source pollution control are single, and cannot effectively prevent and reduce the occurrence of pollution incidents from a global perspective, and there is a lack of comprehensive identification and targeted analysis of potential risks.
By collecting data related to agricultural production activities, using sensors and monitoring stations for real-time monitoring, conducting data preprocessing, correlation analysis and directed analysis of pollution causes, establishing early warning models and emergency response mechanisms to achieve comprehensive prediction and early warning of agricultural non-point source pollution.
It improves the accuracy of early warning, reduces monitoring costs, effectively prevents and reduces the occurrence of pollution incidents, protects the ecological environment and agricultural product quality and safety, and promotes agricultural production efficiency and sustainable development.
Smart Images

Figure CN119026749A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of data analysis and relates to a directional prediction and early warning method based on agricultural non-point source pollution control. Background Art
[0002] With the rapid development of China's economy and society, agriculture has achieved remarkable achievements. However, the overall level of China's agriculture is not high, the scientific and technological content and farmers' environmental awareness are low, especially the increasingly serious agricultural non-point source pollution has become an important factor restricting the sustainable development of China's agriculture and rural areas.
[0003] The severity of agricultural non-point source pollution is mainly manifested in water, atmosphere, soil, etc. Generally speaking, agricultural non-point source pollution refers to the pollution caused by nutrients such as nitrogen and phosphorus, pesticides and other organic or inorganic pollutants in agricultural production activities through surface runoff, farmland leakage or volatilization, mainly including pesticide pollution, fertilizer pollution, agricultural film pollution, intensive farm pollution and sewage irrigation pollution.
[0004] Due to the extensiveness and universality of agricultural activities, the randomness of the occurrence areas and the differences in the distribution of pollution loads, agricultural non-point source pollution has the characteristics of long-term, hidden and difficult to control. These characteristics make the prevention and control of agricultural non-point source pollution difficult. Therefore, early warning and prediction can identify potential non-point source pollution risks, and early warning of possible problems will help avoid the occurrence of pollution accidents. At present, the existing early warning methods for agricultural non-point source pollution control are relatively simple, often replacing the early warning of agricultural non-point source pollution prevention and control with a single water quality warning or atmospheric warning, and cannot solve the effective prevention of agricultural non-point source pollution and reduce the occurrence of pollution incidents from a global perspective. Summary of the invention
[0005] The invention relates to a directional prediction and early warning method based on agricultural non-point source pollution control. The method collects data related to agricultural production activities, including soil quality, pesticide application, farmland drainage, etc., and obtains farmland environmental index data through real-time monitoring of sensors, monitoring stations and other equipment. The obtained real-time data are subjected to data preprocessing and data screening, data correlation analysis and pollution cause directional analysis, model establishment, early warning information release, emergency response and disposal, etc., and finally forms an effective mechanism for predicting, warning, analyzing and processing the agricultural non-point source pollution that is expected to occur. The method effectively overcomes the defects of the existing method, can more comprehensively identify potential non-point source pollution risks, can effectively prevent and reduce the occurrence of pollution events, protect the ecological environment and the quality and safety of agricultural products, improve agricultural production efficiency, promote rural sustainable development, and provide an important basis for fundamentally solving the prevention and control of agricultural non-point source pollution.
[0006] The method of the present invention comprises the following key technical steps: Firstly, the curve fitting analysis method was used to preprocess and screen the real-time monitoring data of the surface source environment to obtain the correlation coefficient matrix between the parameters.
[0007] Then, the correlation data mining method is used to reduce the dimension of the correlation coefficient matrix and obtain a directional analysis of the pollution causes.
[0008] Finally, through machine learning, a mature model system and early warning information release mechanism are established, and emergency response measures are taken to achieve comprehensive prediction and early warning of agricultural non-point source monitoring data.
[0009] Compared with the existing early warning and prediction methods, the beneficial effect of the present invention is mainly to achieve directional analysis of the causes of non-point source pollution, thereby improving the accuracy of early warning and prediction, reducing the expenditure of monitoring costs, and thus achieving the purpose of reducing costs and increasing efficiency. The technology of the present invention can be applied to the prediction and early warning of non-point source pollution prevention and control or the emergency response and disposal, and has a relatively broad application prospect. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] Figure 1 It is a schematic diagram of the directional prediction and early warning process of the present invention. DETAILED DESCRIPTION
[0011] The technical solution and specific implementation scheme adopted by the present invention are as follows: Data collection The main aspects of agricultural non-point source pollution monitoring are soil quality, pesticide application, farmland drainage, and meteorological acid rain. The data that need to be collected usually include pH value, organic matter content, fecal coliform group, roundworm eggs, total phosphorus, ammonia nitrogen, heavy metal content and other parameters. The data collection formula is as follows: Wherein, Dm represents the collected original data set, mi represents the i-th sample data, n represents the number of samples contained in the data set, and mi represents the i-th sample parameter.
[0012] The main collection method is to add IoT devices, water quality monitoring stations, soil monitoring sensors and other hardware monitoring equipment in fields, farms, concentrated planting areas, irrigation water sources and other places to obtain original parameters such as pH value, organic matter content, fecal coliform group, ascaris eggs, total phosphorus, ammonia nitrogen, and heavy metal content through multiple channels.
[0013] All collected sample parameter sets (Dm), (m1, m2, m3...are sample data such as pH value, organic matter content, fecal coliform group, ascaris eggs, total phosphorus, ammonia nitrogen, heavy metal content, etc.) will be used in the data preprocessing and screening process.
[0014] Data preprocessing and screening Preprocess the data, including data cleaning, identification and processing of missing values, removal of outliers, data integration, data reduction, etc., to ensure the quality and credibility of the data.
[0015] Data preprocessing requirements: Data cleaning process: Process the redundant data in the Dm sample parameter set, such as the oxygen content of the collected samples that are not polluted by the surface source, and keep the important sample data such as pH value, organic matter content, fecal coliform group, ascaris eggs, total phosphorus, ammonia nitrogen, heavy metal content, etc.
[0016] Identification of missing values, processing of missing values and removal of outliers: Check whether there are missing values in the cleaned sample data, and supplement or delete them. If outliers are found, check whether they are important sample data. If there are, keep them; if not, delete them.
[0017] Data reduction process: Data preprocessing (least squares curve fitting) formula: Use to represent the data reduction process, where Li(x)(i=1, 2, 3, 4, ...m) is the original residual data, Li(x)(i=1, 2, 3, 4, ...m) is the linear function of x, It represents the reduction factor set of each sample data, etc. minf(x) is to minimize energy or maximize entropy, which is finally used to represent the data after data reduction.
[0018] Data brushing method: Among them, i represents the sample data ordinal number, t represents the monitoring time, Xit represents the reduced data set, Represents a relative parameter set, represents the safety factor, and Yit represents the final screened data set.
[0019] The collected data is cleaned and reduced using the least squares curve fitting method to form a relatively clean sampling data set, which is then further screened using the mixed data screening method to obtain an accurate agricultural non-point source sample data matrix. For example, a sample matrix with high fecal coliform bacteria, high total phosphorus content, low ammonia nitrogen content, and high heavy metal content will be obtained, and then the data correlation analysis stage will be entered.
[0020] Data correlation analysis Through data correlation analysis, key factors in surface source monitoring data are extracted to reduce the clutter and redundancy of data dimensions.
[0021] Correlation formula (Pearson correlation coefficient method): Among them, x¯ and y¯ are the means of the sample data. CosR is obtained by using the Pearson correlation coefficient method to represent the associated data set. This method can be used to calculate the sample set obtained in the previous step.
[0022] The preprocessed and screened data are subjected to data correlation analysis to ultimately obtain a set of agricultural non-point source status data sets with strong correlation. For example, this formula can be used to calculate the mean levels of sample values such as fecal coliform bacteria, ammonia nitrogen content, and heavy metal content, whether the impact on agriculture is reached, or whether it is above the threshold or close to the threshold.
[0023] Cause-oriented analysis After data correlation analysis, we will obtain multiple sample matrix sets. These sets, such as low ammonia nitrogen content and high heavy metal content, have more or less impact on agricultural non-point source pollution, so it is necessary to further determine and analyze the key factors that are decisive factors for agricultural non-point source pollution. Through the causal directed analysis method, we will eventually obtain the parameter set that is closest to the cause of agricultural non-point source pollution.
[0024] Cause-directed analysis formula (mixed effects data estimation method): in is the probability of the event occurring, is the independent variable (a set of parameters such as ammonia nitrogen or heavy metals), is the directional coefficient (the set warning threshold). Through the directional analysis of the causes, we can basically obtain a complete chain of evidence that is the most direct cause of agricultural non-point source pollution. This data provides an effective data basis for the establishment of the subsequent warning model. Early warning model establishment The data sets obtained through targeted analysis are subjected to machine learning methods or big data analysis methods, such as decision trees, random forests, support vector machines, neural networks, etc., to construct a complete prediction, early warning, analysis and processing model involving multiple dimensions such as scenarios, businesses, users, alarms, and decisions.
[0025] The data set obtained from the cause-oriented analysis is used in the early warning model to form a complete agricultural non-point source pollution prediction and early warning mechanism, set early warning rules and thresholds, and when the model prediction result exceeds a certain threshold, the system will issue a warning signal and transmit the signal to the person in charge of the agricultural non-point source through text messages or alarms.
[0026] Emergency response When the early warning model sends a signal or pollution has already occurred, the relevant person in charge can use the pre-planned governance methods and steps to respond quickly to the emergency situation and deal with it effectively; Summary of key elements of emergency response: Clearly allocate responsibilities: ensure that each role and team has various responsibilities and duties in emergency situations; formulate emergency plans: prepare detailed emergency plans, including main tasks, information reporting procedures, personnel evacuation and transfer plans, and the use of emergency resources; implement emergency response measures: according to the severity and nature of the incident, activate the corresponding level of emergency response mechanism, and the intensity will decrease in turn. Communication, coordination and recovery work.
[0027] When the early warning model sends a signal, the relevant person in charge who receives the signal can quickly find the corresponding team or role, and then proceed to check the specific situation of the corresponding location in an orderly manner according to the emergency plan, which will completely curb the situation that causes agricultural non-point source pollution, thereby fundamentally providing an important basis for solving the prevention and control of agricultural non-point source pollution. The operation of completing the precise early warning, compared with the previous early warning operation, this invention solves the problem of agricultural non-point source pollution prevention and control from the root, directly locates the pollution factor to a specific point, and can fundamentally solve the prevention and control of agricultural non-point source pollution.
[0028] Beneficial Effects 1. Accurate data mining and screening The curve fitting method is used to perform accurate preprocessing operations on the collected data. Compared with traditional data preprocessing methods, this method can further provide the required more accurate data, which is of great help to the accuracy of prediction and early warning.
[0029] Detailed implementation method: By establishing multiple automatic water quality monitoring stations and soil monitoring sensors, various parameter data in water and soil are obtained in real time using water quality monitoring stations and soil monitoring sensors, and these obtained data are processed and cleaned. At this time, this type of data does not have a linear relationship, so we use the data method with discrete fitting - curve fitting method to make a data precision processing module, correct and analyze this type of data, so as to obtain a relatively continuous discrete point curve that tends to fit composed of monitoring data.
[0030] 2. Data Correlation Analysis The correlation coefficient method is used to deeply mine the preprocessed data, extract the key factors in the surface source monitoring data, and reduce the clutter and redundancy of the data dimensions. This method can further make accurate judgments on pollution factors, thereby further improving the accuracy of prediction and early warning.
[0031] Detailed implementation method: After precise data mining and screening, a relatively continuous curve data is formed. At this time, the key factors of pollution cannot be fully given. Therefore, we use mathematical logic to establish a data correlation analysis module, and use the characteristics of the correlation coefficient method to analyze and screen out strongly correlated data and eliminate irrelevant data, so that the data on the fitting curve can be further accurately improved.
[0032] 3. Cause-oriented analysis The method of causal directional analysis can basically obtain the most direct and complete chain of evidence that ultimately leads to agricultural non-point source pollution. This data provides an effective data basis for the establishment of the early warning model in the later stage. Compared with the traditional fuzzy early warning, it can effectively prevent and reduce the occurrence of pollution incidents, protect the ecological environment and the quality and safety of agricultural products, and provide an important basis for fundamentally solving the prevention and control of agricultural non-point source pollution.
[0033] Detailed implementation method: When preventing and controlling surface source pollution, we usually set a threshold for each pollution factor. We establish a cause-oriented analysis module, set a warning threshold for each pollution factor, and then bring the fitting curve into it. When the curve tends to the position of the warning threshold, we can judge that pollution is about to occur, so as to accurately screen out the pollution factors. The chain formed by these pollution factors that tend to the warning threshold provides an early warning basis for the establishment of the subsequent early warning model.
[0034] Glossary Agricultural non-point source pollution refers to the irrational use of chemical inputs such as fertilizers, pesticides, and ground films in the agricultural production process, as well as the untimely or improper treatment of livestock and aquaculture waste, crop straw, etc., which, driven by rainfall and topography, use surface and underground runoff and soil erosion as carriers, accumulate excessively in the soil or enter the receiving water bodies, posing a threat to the ecological environment and food safety.
Claims
1. A directional prediction and early warning method based on agricultural non-point source pollution control, characterized in that The steps include: Firstly, the curve fitting analysis method is used to pre-process and screen the real-time monitoring data of the surface source environment, and the correlation coefficient matrix between the parameters is obtained; Then, the correlation data mining method is used to reduce the dimension of the correlation coefficient matrix and obtain the directional analysis of the pollution causes; Finally, a mature model system and early warning information release are established through machine learning.
2. A directional prediction and early warning method based on agricultural non-point source pollution control according to claim 1, characterized in that: Data collection The main aspects of agricultural non-point source pollution monitoring are soil quality, pesticide application, farmland drainage, and meteorological acid rain. The collected data usually include pH value, organic matter content, fecal coliform group, roundworm eggs, total phosphorus, ammonia nitrogen, heavy metal content and other parameters expressed as: Collection data formula Dm={m1,m2,m3,m4,m5,m6,mi,...mn} Wherein, Dm represents the original data set collected, mi represents the i-th sample data, n represents the number of samples contained in the data set, and mi represents the i-th sample parameter; Obtain original parameters such as pH value, organic matter content, fecal coliform group, ascaris eggs, total phosphorus, ammonia nitrogen, heavy metal content, etc.; all sample parameter sets (Dm), m1, m2, m3... after collection are sample data such as pH value, organic matter content, fecal coliform group, ascaris eggs, total phosphorus, ammonia nitrogen, heavy metal content, etc., which are used in data preprocessing and screening; Data preprocessing and screening Preprocess the data, including data cleaning, identification and processing of missing values, removal of outliers, data integration, data reduction, etc.; Process the redundant data in the Dm sample parameter set, such as the oxygen content of the data without pollution from the surface source in the collected samples, and keep the sample data such as pH value, organic matter content, fecal coliform group, roundworm eggs, total phosphorus, ammonia nitrogen, heavy metal content, etc.; Identification of missing values and processing of missing values and removal of outliers: Check whether there are missing values in the cleaned sample data, and supplement or delete them. If outliers are found, check whether they are important sample data. If there are, keep them, otherwise delete them; Data reduction process: data preprocessing, least squares curve fitting formula: Use to represent the data reduction process, where Li(x)(i=1,2,3,4,...m) is the original residual data, Li(x)(i=1,2,3,4,...m) is the linear function of x, It represents the reduction factor set of each sample data, etc., minf(x) is to minimize energy or maximize entropy, and is finally used to represent the data after data reduction; data screening method: Among them, i represents the sample data ordinal number, t represents the monitoring time, Xit represents the reduced data set, Represents a relative parameter set, represents the safety factor; Yit represents the final data set screened; Data correlation analysis Extract key factors from surface source monitoring data through data correlation analysis to reduce the clutter and redundancy of data dimensions; Correlation formula Pearson correlation coefficient method: Among them, x¯ and y¯ are the means of sample data. CosR is obtained by using the Pearson correlation coefficient method to represent the associated data set. This method is used to calculate the sample set obtained in the previous step. The preprocessed and screened data are subjected to data correlation analysis to finally obtain a set of agricultural non-point source status data sets with strong correlation. Cause-oriented analysis After data correlation analysis, multiple sample matrix sets will be obtained to determine the key factors that are decisive for agricultural non-point source pollution. Through the cause-oriented analysis method, the parameter set closest to the cause of agricultural non-point source pollution will eventually be obtained. Cause-directed analysis formula mixed effects data estimation method: in is the probability of the event occurring, is an independent variable, ammonia nitrogen or heavy metal parameter set, is the directional coefficient, setting the warning threshold; Early warning model establishment The data sets obtained through targeted analysis are analyzed using machine learning methods or big data analysis methods, such as decision trees, random forests, support vector machines, neural networks, etc., to build a complete prediction and early warning analysis and processing model involving multiple dimensions such as scenarios, businesses, users, alarms, and decisions. Early warning rules and thresholds are set for the data sets obtained through causal targeted analysis. When the results predicted by the model exceed a certain threshold, an early warning signal will be issued.
Citation Information
Cited By
Underground reactor operation monitoring system and method based on sensor network
CN119418509A
Agricultural non-point source pollution early warning method and system and electronic equipment
CN119918755A
Agricultural non-point source pollution early warning method and system, and electronic device
CN119918755B
Rainfall runoff pollution early warning system and method
CN120761606A
Regulation and control method and system for As-polluted paddy field soil
CN121936859A
