Water environment situation big data research, judgment and treatment integrated system

Abstract

The invention discloses a water environment situation big data research, judgment and treatment integrated system. The system comprises a plurality of big data of different sources, space unit division, a data warehouse module, an index system module, a deep learning model and a water environment situation early warning and prediction module. On the basis of a big data platform, through a source-sink relationship of a drainage basin water environment, big data feature values and label values of the drainage basin water environment and influence factors of the drainage basin water environment are identified and extracted, and on-line real-time prediction of the drainage basin water environment based on deep learning is carried out under a big data condition, so that a water environment situation research and judgment intelligent decision support system is constructed. According to the method, intelligent decision analysis and situation study and judgment of the water environment in the whole watershed scale are realized, the method is practically applied to watershed water environment management, and a new method and a new model are provided for intelligent decision of the watershed water environment. In addition, a watershed water environment intelligent decision informatization big data platform is constructed, and powerful support is provided for intelligent research and judgment and intelligent management of watershed water environment postures.

Description

Technical Field

[0001] The invention relates to the technical field of water environment situation analysis and judgment, especially to an integrated system for water environment situation big data analysis and judgment and management. Background Art

[0002] The ecological environment of a river basin plays a vital role in promoting the healthy, sustainable and coordinated development of a country's economy and society. Among them, the water environment of a river basin is the core component of the ecological environment of a river basin. At present, seeking effective methods for the governance and management of river basin water environment has become an important consideration for promoting the sustainable and healthy development of social economy.

[0003] The protection and management of river basin water environment is a complex system project. Local governments have difficulty in solving the problem of transboundary water pollution in river basins, which makes it impossible to optimize the allocation of water resources and water environment, and the comprehensive management of river basin water environment is still difficult.

[0004] Intelligent water environment assessment refers to building a suitable water environment big data analysis model based on the big data of online water environment and water quality monitoring and the big data of related factors affecting water quality, so as to explore the changing rules of various indicators in water quality, so as to accurately predict and judge the water environment quality, timely understand the current status of water environment quality and predict the possibility of water environment pollution and future development trends in advance. Accurately predicting the future trend of water environment quality helps to timely discover water environment problems in the basin, is the basis for the prevention and control of water environment pollution, and can provide an important scientific basis for early warning, supervision and maintenance of water environment quality in the basin.

[0005] At present, domestic and foreign scholars have done a lot of research in the field of intelligent assessment of water environment situation, which can be classified into mechanistic research and non-mechanistic research. Among them, mechanistic research is to establish a water environment quality prediction model for a specific water area based on the conservation principles of energy, mass, momentum, etc., which comprehensively describes the laws of migration and transformation of pollutants in the water body. It comprehensively considers various factors affecting water quality, such as flow, flow rate, topography, water depth, diffusion coefficient, degradation, precipitation, etc., and the research prediction results are good. However, this type of research is often only carried out on small-scale basins, and the model obtained by the research is only applicable to this water area. If you want to apply this model to other water areas, you need to collect a lot of data again to redefine the parameters. Therefore, the results of this type of research have a narrow scope of application and weak generalization ability. In addition, this type of research lacks the integration of large-scale basin information from a holistic perspective, and the model construction, organization, analysis and prediction, management and application methods of data can no longer adapt to the situation of multiple departments, wide data, diverse data types and huge amount of information in water environment management in the era of big data. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention aims to provide an integrated system for analyzing and managing big data of water environment situation.

[0007] In order to achieve the above object, the present invention adopts the following technical solution:

[0008] An integrated system for water environment situation big data analysis and management, the system includes big data from several different sources, spatial unit division, data warehouse module, indicator system module, deep learning model, and water environment situation early warning and prediction module; among them, based on the big data platform, through the source-sink relationship of the water environment in the river basin, the big data characteristic values ​​and label values ​​of the water environment in the river basin and its influencing factors are identified and extracted, and the water environment in the river basin is predicted online in real time based on deep learning under big data conditions, thereby building an intelligent decision-making support system for water environment situation analysis.

[0009] Preferably, the big data includes at least online monitoring data, "two pollution censuses" data, geographic data, meteorological data, remote sensing data, hydrological data, online public opinion data, socio-economic data, POI data, administrative review data, industrial and commercial electricity data and other auxiliary data.

[0010] Preferably, the inter-unit division includes using DEM to generate a watershed area, which can effectively divide different watershed areas in the basin, ensuring that the spatiotemporal context information of the water environment data in the basin is accurately used; wherein the watershed extraction step is as follows: based on DEM, the ArcGIS hydrological analysis tool is used to extract the flow direction, cumulative flow, flow length and river network of surface water, and on this basis, the basin of the study area is segmented, and finally the watershed is extracted according to the online monitoring points to determine the spatial range of the watershed of a single sample;

[0011] Geographic data, remote sensing data, and other spatial data such as meteorology and hydrology are superimposed in layer order, and data pooling is performed using a spatial pyramid. Corresponding cross-section water quality feature labels are generated to form sample feature image files. Convolutional neural networks are used for feature extraction, and deep learning is performed after adding attribute data of governance inputs to the fully expanded layer.

[0012] The monitored time series data is processed using a deep neural network, and the monitoring data of the monitoring sections connected by the water system are combined into sample feature data to form a fully expanded layer neural network for deep learning. The water quality is predicted based on the corresponding water quality section label data.

[0013] Preferably, the construction of the data warehouse includes cleaning of online monitoring data such as water quality and rainfall, enterprise pollutant emissions, "two pollution censuses" data, social economy and land use data, including verification of the validity of the accessed data and data deduplication to ensure the temporal, spatial and logical consistency of multi-source data; for the cleaned data, the data format of the external system is converted into the national standard data format according to environmental protection business rules, and data correction and desensitization are performed to ensure data security during data use.

[0014] Preferably, the establishment of the deep model includes, after preprocessing and spatializing the big data, firstly, superimposing spatial data such as land use, rainfall, social economy, pollutant emissions, etc. in layer order to form a sample feature image file, using CNN for feature extraction, adding attribute data such as governance input in the full expansion layer, and then conducting deep learning on the corresponding water quality label data; secondly, combining the monitoring data of the monitoring sections connected by the watershed system and the rainfall data as sample feature data to form a full expansion layer neural network, and using DNN to conduct deep learning on the corresponding water quality label data.

[0015] Preferably, both methods include two scale models: single catchment area and whole-basin catchment area. Through continuous iteration of the CNN-DNN hybrid layer, the spatiotemporal context information of the data is coupled into the deep learning model to obtain a big data model of the spatiotemporal evolution of water quality in the Pearl River Basin that can automatically extract spatiotemporal characteristics, and the model is continuously optimized through continuously improved online monitoring data.

[0016] The beneficial effect of the present invention is that, through the deep learning watershed water environment behavior prediction model coupled with spatiotemporal context information in the present invention, the spatiotemporal characteristics of watershed water environment behavior can be automatically extracted and abstracted, breaking through the current limitations that hinder the widespread application of mechanism models and machine learning, and realizing the intelligent decision-making analysis and situation assessment of the water environment at the whole watershed scale, and actually applying it to the watershed water environment management, providing new methods and models for intelligent decision-making of watershed water environment. In addition, based on the aforementioned proposed big data method and model for intelligent assessment of watershed water environment situation, a big data platform for intelligent decision-making of watershed water environment is constructed, providing strong support for the intelligent assessment and intelligent management of watershed water environment situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0018] Figure 2 It is a schematic diagram of the watershed extraction process from the perspective of big data of the present invention;

[0019] Figure 3 A schematic diagram of the structure of the water environment deep learning model coupled with spatiotemporal context information of the present invention;

[0020] Figure 4 It is a schematic diagram of the overall framework of the water environment situation analysis and intelligent decision support platform of the present invention. DETAILED DESCRIPTION

[0021] The present invention will be further described below in conjunction with the accompanying drawings. It should be noted that this embodiment is based on the technical solution and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to this embodiment.

[0022] like Figure 1 , Figure 2 and Figure 3 As shown, the present invention is an integrated system for water environment situation big data analysis and management, the system includes big data from several different sources, spatial unit division, data warehouse module, indicator system module, deep learning model, water environment situation early warning and prediction module; wherein, based on the big data platform, through the source-sink relationship of the water environment in the basin, the big data feature values ​​and label values ​​of the water environment in the basin and its influencing factors are identified and extracted, and the water environment in the basin is predicted online in real time based on deep learning under big data conditions, thereby constructing an intelligent decision support system for water environment situation analysis.

[0023] Preferably, the big data includes at least online monitoring data, "two pollution censuses" data, geographic data, meteorological data, remote sensing data, hydrological data, online public opinion data, socio-economic data, POI data, administrative review data, industrial and commercial electricity data and other auxiliary data.

[0024] Preferably, the inter-unit division includes using DEM to generate a watershed area, which can effectively divide different watershed areas in the basin, ensuring that the spatiotemporal context information of the water environment data in the basin is accurately used; wherein the watershed extraction step is as follows: based on DEM, the ArcGIS hydrological analysis tool is used to extract the flow direction, cumulative flow, flow length and river network of surface water, and on this basis, the basin of the study area is segmented, and finally the watershed is extracted according to the online monitoring points to determine the spatial range of the watershed of a single sample;

[0025] Geographic data, remote sensing data, and other spatial data such as meteorology and hydrology are superimposed in layer order, and data pooling is performed using a spatial pyramid. Corresponding cross-section water quality feature labels are generated to form sample feature image files. Convolutional neural networks are used for feature extraction, and deep learning is performed after adding attribute data of governance inputs to the fully expanded layer.

[0026] The monitored time series data is processed using a deep neural network, and the monitoring data of the monitoring sections connected by the water system are combined into sample feature data to form a fully expanded layer neural network for deep learning. The water quality is predicted based on the corresponding water quality section label data.

[0027] Preferably, the construction of the data warehouse includes cleaning of online monitoring data such as water quality and rainfall, enterprise pollutant emissions, "two pollution censuses" data, social economy and land use data, including verification of the validity of the accessed data and data deduplication to ensure the temporal, spatial and logical consistency of multi-source data; for the cleaned data, the data format of the external system is converted into the national standard data format according to environmental protection business rules, and data correction and desensitization are performed to ensure data security during data use.

[0028] Preferably, the establishment of the deep model includes, after preprocessing and spatializing the big data, firstly, superimposing spatial data such as land use, rainfall, social economy, pollutant emissions, etc. in layer order to form a sample feature image file, using CNN for feature extraction, adding attribute data such as governance input in the full expansion layer, and then conducting deep learning on the corresponding water quality label data; secondly, combining the monitoring data of the monitoring sections connected by the watershed system and the rainfall data as sample feature data to form a full expansion layer neural network, and using DNN to conduct deep learning on the corresponding water quality label data.

[0029] Preferably, both methods include two scale models: single catchment area and whole-basin catchment area. Through continuous iteration of the CNN-DNN hybrid layer, the spatiotemporal context information of the data is coupled into the deep learning model to obtain a big data model of the spatiotemporal evolution of water quality in the Pearl River Basin that can automatically extract spatiotemporal characteristics, and the model is continuously optimized through continuously improved online monitoring data.

[0030] Example

[0031] This embodiment takes the Pearl River as an example to further illustrate the present invention.

[0032] like Figure 4 As shown in the figure, through the Internet, big data and cloud computing technologies, coupled with the built big data platform of the ecological environment of the Pearl River Basin, a big data center for intelligent assessment of water environment situation is constructed. Focusing on the "water" and "ecology" in the ecological and environmental elements of the Pearl River Basin, the interaction between the basin's social economy, terrestrial ecological space and river systems is systematically considered to achieve intelligent assessment and scientific forecasting and early warning of water environment situation, and provide reliable support for the source tracing analysis of water environment pollution sources in the basin and emergency response and decision-making of sudden ecological and environmental events. Integrate the results of the above research contents such as intelligent assessment and governance technology of water environment situation, and build an intelligent decision-making support platform for water environment situation assessment in the Pearl River Basin with support for key breakthroughs, comprehensive analysis, situation assessment, one map and comprehensive decision-making, so as to provide technical support and management tools for comprehensive supervision of the water environment in the basin.

[0033] For those skilled in the art, various corresponding changes and modifications can be made according to the above technical solutions and concepts, and all of these changes and modifications should be included in the protection scope of the claims of the present invention.

Claims

1. The integrated system of water environment situation big data analysis and management is characterized by: The system includes big data from several different sources, spatial unit division, data warehouse module, indicator system module, deep learning model, and water environment situation warning and prediction module; among them, based on the big data platform, through the source-sink relationship of the water environment in the river basin, the big data characteristic values ​​and label values ​​of the water environment in the river basin and its influencing factors are identified and extracted, and the water environment in the river basin is predicted online in real time based on deep learning under big data conditions, thereby building an intelligent decision-making support system for water environment situation analysis and judgment.

2. The integrated system for water environment situation big data analysis and management according to claim 1 is characterized in that: The big data includes at least online monitoring data, "two pollution censuses" data, geographic data, meteorological data, remote sensing data, hydrological data, online public opinion data, socio-economic data, POI data, administrative review data, industrial and commercial electricity data and other auxiliary data.

3. The water environment situation big data analysis and management integrated system according to claim 1 is characterized by: The inter-unit division includes using DEM to generate watersheds, which can effectively divide different watershed areas in the basin, ensuring that the spatiotemporal context information of the water environment data in the basin is accurately used; the watershed extraction steps are as follows: based on DEM, ArcGIS hydrological analysis tools are used to extract the flow direction, cumulative runoff, flow length and river network of surface water, and on this basis, the basin of the study area is segmented, and finally the watershed is extracted according to the online monitoring points to determine the spatial range of the watershed of a single sample; Geographic data, remote sensing data, and other spatial data such as meteorology and hydrology are superimposed in layer order, and data pooling is performed using a spatial pyramid. Corresponding cross-section water quality feature labels are generated to form sample feature image files. Convolutional neural networks are used for feature extraction, and deep learning is performed after adding attribute data of governance inputs to the fully expanded layer. The monitored time series data is processed using a deep neural network, and the monitoring data of the monitoring sections connected by the water system are combined into sample feature data to form a fully expanded layer neural network for deep learning. The water quality is predicted based on the corresponding water quality section label data.

4. The water environment situation big data analysis and management integrated system according to claim 1 is characterized by: The construction of the data warehouse includes cleaning of online monitoring data such as water quality and rainfall, enterprise pollutant emissions, "two pollution censuses" data, social economy and land use data, including validation of the accessed data and data deduplication to ensure the temporal, spatial and logical consistency of multi-source data; After the data is cleaned, the data format of the external system is converted into the national standard data format according to environmental protection business rules, and the data is corrected and desensitized to ensure data security during use.

5. The integrated system for water environment situation big data analysis and management according to claim 1 is characterized in that: The establishment of the deep model includes, after pre-processing and spatial processing of big data, firstly, superimposing spatial data such as land use, rainfall, social economy, and pollutant emissions in layer order to form a sample feature image file, using CNN for feature extraction, adding attribute data such as governance input in the full expansion layer, and then carrying out deep learning corresponding to the water quality label data; The second is to combine the monitoring data of the monitoring sections connected by the river system and the rainfall data as sample feature data to form a fully expanded layer neural network, and use DNN to carry out deep learning on the corresponding water quality label data.

6. The integrated system for water environment situation big data analysis and management according to claim 5 is characterized in that: Both methods include two scale models: single catchment area and whole-basin catchment area. Through continuous iteration of the CNN-DNN hybrid layer, the spatiotemporal context information of the data is coupled into the deep learning model to obtain a big data model of the spatiotemporal evolution of water quality in the Pearl River Basin that can automatically extract spatiotemporal characteristics. The model is continuously optimized through continuously improved online monitoring data.

Images