Environment monitoring system based on big data

Abstract

The invention relates to the technical field of environment monitoring, discloses an environment monitoring system based on big data, and aims to solve the defects of an existing environment monitoring system in the aspects of data processing, analysis, risk early warning and suggestion coping. The system collects various environmental data in real time through the environmental monitoring data acquisition module, constructs a model by using the environmental quality evaluation module, and converts the collected data into visual indexes such as an air quality index, a water quality index and a soil quality index. And the pollution prediction module predicts the future environmental pollution level and trend according to the historical data and the external influence factors. And the risk early warning module carries out accurate early warning on environmental risks in combination with evaluation and prediction results. And the coping suggestion module generates targeted environmental protection and coping measure suggestions according to the early warning result, and triggers emergency early warning and provides a detailed scheme when the risk is high. According to the system, the accuracy and timeliness of environment monitoring are improved, and powerful support is provided for environment protection and emergency management.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental monitoring, and specifically to an environmental monitoring system based on big data. Background Art

[0002] With the acceleration of industrialization and urbanization, environmental problems have become increasingly prominent. Problems such as air quality deterioration, water pollution, soil degradation, and ecological damage seriously threaten the living environment and health of humans. The rapid development of the Internet of Things technology has provided new means for environmental monitoring. By deploying various sensors and monitoring devices, a large amount of environmental data can be collected in real time, including air quality data (such as PM2.5, PM10, SO2, NOx, etc.), water quality data (such as dissolved oxygen, pH value, heavy metal content, etc.), soil data (such as soil humidity, nutrient content, pollutant concentration, etc.), meteorological data (such as temperature, humidity, wind speed, wind direction, etc.), and ecological data (such as vegetation coverage, biodiversity index, etc.). However, the huge volume and complexity of these data pose higher requirements for data processing and analysis.

[0003] Existing environmental monitoring systems have many deficiencies in data processing and analysis. On the one hand, there is a lack of an efficient data integration and evaluation mechanism, and it is unable to convert various types of collected environmental data into intuitive environmental quality indices, such as air quality index, water quality index, and soil quality index, so as to facilitate managers and the public to quickly understand the environmental quality status. On the other hand, existing environmental monitoring systems often focus on the assessment of the current environmental quality, while ignoring the prediction and early warning of future environmental pollution levels, resulting in the inability to take effective countermeasures in a timely manner when environmental pollution incidents occur.

[0004] Existing environmental monitoring systems also have deficiencies in risk early warning and response suggestions. Due to the lack of an effective risk early warning model, it is unable to accurately warn of environmental risks based on environmental quality assessment results and pollution prediction results. In terms of response suggestions, existing systems often can only provide simple suggestions for countermeasures, lacking pertinence and operability, and unable to meet the actual needs of environmental protection and emergency management. Summary of the Invention

[0005] The purpose of the present invention is to provide an environmental monitoring system based on big data to solve the problems raised in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solution: An environmental monitoring system based on big data, the system includes: An environmental monitoring data collection module, constructed using the Internet of Things technology, for collecting air quality data, water quality data, soil data, meteorological data, and ecological data in real time; Environmental Quality Assessment Module, which constructs an environmental quality assessment model. The input data of the environmental quality assessment model is the data collected by the environmental monitoring data collection module, and the output data is the environmental quality index, including the air quality index, water quality index, and soil quality index. The model is trained using a dataset containing historical environmental quality monitoring data and corresponding quality indices, and the corresponding labels are various environmental quality indices. Pollution Prediction Module, which constructs a pollution prediction model. The input data of the pollution prediction model is the historical environmental quality index and external influencing factor data, and the output data is the predicted possible environmental pollution level and its trend in a future period. The model is trained using a time series dataset containing historical environmental quality indices and external influencing factors, and the corresponding label is the future environmental pollution level. Risk Warning Module, which constructs a risk warning model. The input data of the risk warning model is the evaluation result of the environmental quality assessment model and the prediction result of the pollution prediction model, and it combines the information of both to warn of environmental risks. The model is trained using a combined dataset containing environmental quality assessment results and pollution prediction results, and the corresponding label is the environmental risk level. Response Suggestion Module, which is connected to the risk warning module and generates corresponding environmental protection and response measure suggestions according to the warning result. If the warning result is a high risk, it triggers an emergency warning notice and provides a detailed response plan.

[0007] Preferably, the environmental quality assessment model is constructed using the Convolutional Neural Network (CNN) algorithm. It extracts the features of the input data through convolutional layers and pooling layers. Its network structure includes: Input Layer: Receives the preprocessed environmental monitoring data, and the data format is unified into a multi-dimensional matrix form. Convolutional Layer: Extracts data features through multiple convolutional kernels to generate feature maps. Pooling Layer: Performs downsampling on the feature maps to reduce the data dimension. Fully Connected Layer: Maps the features output by the pooling layer to various environmental quality indices through the fully connected layer. Output Layer: Outputs the air quality index, water quality index, and soil quality index.

[0008] Preferably, the pollution prediction model is constructed using the Long Short-Term Memory (LSTM) algorithm. It captures the long-term dependencies in the time series data through LSTM units. Its network structure includes: Input Layer: Receives the historical environmental quality index and external influencing factor data. LSTM Layer: Contains multiple LSTM units, and each unit processes one time step in the sequence data, passing the cell state and hidden state to the next unit. Fully Connected Layer: Maps the hidden state output by the LSTM layer to the predicted environmental pollution level through the fully connected layer. Output layer: Output the possible environmental pollution levels and their trends in the future for a period of time.

[0009] Preferably, the risk early warning model separates data of different risk levels by constructing a hyperplane, and its construction steps include: S1: Use the environmental quality assessment results and pollution prediction results as input features, and use the environmental risk level as the target variable; S2: Select the radial basis function as the kernel function; S3: Use the support vector machine algorithm to train the model and find the optimal hyperplane; S4: For new input data, judge its risk level according to the distance of its position in the feature space relative to the hyperplane.

[0010] Preferably, the optimization steps of the risk early warning model include: Step 1: Adjust the parameters of the support vector machine, including the penalty parameter C and the kernel function parameters; Step 2: Adopt grid search combined with cross-validation to evaluate the model performance and select the optimal parameter combination; Step 3: Use the validation set to verify the optimized model and evaluate the accuracy rate and F1 score of the model; Step 4: When the model performance reaches the preset standard, save the model parameters to obtain the trained risk early warning model.

[0011] Preferably, the implementation method of the response suggestion module includes: Pre-define a variety of environmental protection and response measure suggestions, each suggestion corresponding to different environmental risk levels and pollution types, and the suggestion content covers the level of early warning notice, the activation conditions of emergency response and specific environmental protection measures; Receive the early warning results of the risk early warning module, and match the corresponding response suggestions in the suggestion library according to the risk level and pollution type in the early warning results; Execute specific environmental protection and response measures according to the response suggestions matched by the decision engine, including adjusting industrial emission standards, starting water purification procedures, issuing public health warnings and implementing ecological restoration plans.

[0012] Preferably, the specific method for constructing the suggestion library is: Set up a hash table as the data structure for storing response suggestions. Each key contains a unique identifier, which is used to represent a specific combination of environmental risk levels and pollution types; the value part contains a text string that represents a specific response suggestion. The hash table uses open addressing or chaining to handle collisions, so as to efficiently store and retrieve suggestion information. The definition of the hash table structure is: H = {(k, v) | k ∈ {risk level_pollution type}, v ∈ {response suggestion}}; where k represents the combination of environmental risk levels and pollution types, and v represents the corresponding response suggestion.

[0013] Preferably, the response suggestion module further includes a key matcher, which has a built-in hash function. Using the hash function, the key matcher quickly calculates the hash value of the input combination of environmental risk levels and pollution types, and then searches for the response suggestion that matches this hash value in the hash table.

[0014] Preferably, the response suggestion module further includes a suggestion editor, which provides an editing function for users to add, modify or delete keys and values in the hash table according to the actual environmental conditions and policy changes, so as to update the response suggestion library.

[0015] Preferably, the environmental monitoring data collection module uses the MQTT protocol to achieve real-time data transmission and cloud storage.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This system uses Internet of Things technology to collect multi-dimensional environmental data such as air quality, water quality, soil, meteorology and ecology in real time, ensuring the comprehensiveness and timeliness of monitoring data. By constructing an environmental quality assessment model, the original data is converted into intuitive environmental quality indices (such as air quality index, water quality index, soil quality index), improving the accuracy and readability of environmental monitoring, and facilitating managers and the public to quickly understand the environmental quality status.

[0017] The system integrates a pollution prediction module. Using historical environmental quality indices and external influencing factor data, it can predict the possible environmental pollution levels and their trends in a future period, providing forward-looking information support for environmental management. The risk warning module combines the environmental quality assessment results and pollution prediction results to accurately warn of environmental risks, helping to take preventive measures before environmental pollution events occur and reducing the impact of environmental pollution on human health and the ecological environment. The response suggestion module generates targeted environmental protection and response measure suggestions based on the risk warning results, improving the practicability and operability of the response plan. In the case of high-risk warnings, the system can automatically trigger emergency warning notifications and provide detailed response plans, ensuring the rapidity and effectiveness of emergency response.

[0018] Through big data technology and machine learning algorithms, this system realizes the intelligent processing and analysis of environmental monitoring data, providing a scientific basis and decision-making support for environmental management. The high integration and automation features of the system reduce the costs of manual monitoring and analysis, and improve the efficiency and accuracy of environmental monitoring work. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 FIG. is the working principle diagram of the big data-based environmental monitoring system described in the present invention; Figure 2 FIG. is the flowchart of the construction and optimization process of the risk warning model; Figure 3 FIG. is the working flowchart of the response suggestion module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0021] Please refer to Figures 1-3 , the present invention provides a technical solution: a big data-based environmental monitoring system, the system includes: Environmental monitoring data acquisition module: This module is constructed using Internet of Things technology and includes the deployment of various sensors and monitoring devices. These sensors and devices are configured to collect real-time air quality data (such as PM2.5, PM10, SO2, NOx, etc.), water quality data (such as dissolved oxygen, pH value, heavy metal content, etc.), soil data (such as soil humidity, nutrient content, pollutant concentration, etc.), meteorological data (such as temperature, humidity, wind speed, wind direction, etc.) and ecological data (such as vegetation coverage, biodiversity index, etc.). The data is transmitted to the data center through a wireless network for storage and preprocessing for subsequent use by other modules.

[0022] Environmental quality assessment module: This module constructs an environmental quality assessment model, which takes the data collected by the environmental monitoring data acquisition module as input and outputs environmental quality indices, including air quality index, water quality index and soil quality index.

[0023] Pollution prediction module: This module constructs a pollution prediction model, which takes historical environmental quality indices and external influencing factor data (such as weather forecasts, human activity data, etc.) as input and outputs the predicted possible environmental pollution levels and their trends in a future period of time.

[0024] Risk warning module: This module constructs a risk warning model. This model takes the evaluation results of the environmental quality assessment model and the prediction results of the pollution prediction model as inputs, and combines the information of both to give early warnings about environmental risks.

[0025] Response suggestion module: This module is connected to the risk warning module and generates corresponding environmental protection and response measure suggestions according to the warning results. A series of response suggestions are preset according to different levels of warning results. When the warning result is a high risk, an emergency warning notice is triggered, and a detailed response plan is provided, including emergency response measures, resource allocation plans, etc. According to the actual response effects and feedback, the response suggestions are continuously optimized and updated to improve their pertinence and operability.

[0026] The present invention will be further described below in conjunction with Examples 1 to 4:

[0027] Example 1: The environmental quality assessment model is constructed using the Convolutional Neural Network (CNN) algorithm, aiming to extract the features of the input data through convolutional layers and pooling layers, and then map them to various environmental quality indices. Its network structure: Input layer: Receives preprocessed environmental monitoring data. After these data are preprocessed through steps such as cleaning and normalization, they are unified into the form of a multi-dimensional matrix. For example, air quality data can be represented as a three-dimensional matrix containing time, location, and pollutant concentrations.

[0028] Convolutional layer: Extracts data features through multiple convolutional kernels. The convolutional kernels slide on the input data, perform weighted summation of local regions, and generate feature maps. These feature maps characterize different patterns in the input data, such as the spatial distribution of pollutant concentrations, temporal change trends, etc.

[0029] Pooling layer: Performs downsampling on the feature maps to reduce the data dimension. The pooling operation can select the maximum value (max pooling) or the average value (average pooling) in the feature maps, thereby retaining important features and reducing the computational amount.

[0030] Fully connected layer: Maps the features output by the pooling layer to various environmental quality indices through the fully connected layer. Each neuron in the fully connected layer is connected to all neurons in the previous layer, and through the linear combination of weights and biases and the transformation of a non-linear activation function (such as ReLU), the predicted values of the environmental quality indices are generated.

[0031] Output layer: Outputs the air quality index, water quality index, and soil quality index. These indices are intuitive reflections of environmental quality, facilitating managers and the public to quickly understand the environmental quality status.

[0032] Taking the air quality index as an example, the input layer receives multi-dimensional matrix data containing the concentrations of pollutants such as PM2.5, PM10, SO2, and NOx. The convolutional layer extracts the spatial distribution characteristics and temporal variation trends of these pollutant concentrations through multiple convolutional kernels. The pooling layer downsamples the feature maps to reduce the data dimension. The fully connected layer maps the output of the pooling layer to the air quality index. The output layer outputs the real-time air quality index, which is used to reflect the quality of the current air. Traditional environmental quality assessment methods often rely on manual experience and simple statistical models, and it is difficult to accurately capture the complex patterns in environmental data. The CNN model improves the accuracy and efficiency of environmental quality assessment by automatically learning data features.

[0033] The pollution prediction model is constructed using the Long Short-Term Memory (LSTM) algorithm, aiming to capture the long-term dependencies in time series data through LSTM cells, and then predict the future environmental pollution level. Its network structure: Input layer: Receives historical environmental quality index and external influencing factor data. The historical environmental quality index reflects the environmental quality status over a past period of time, and the external influencing factor data includes weather forecasts, human activity data, etc. These factors may affect the future environmental pollution level.

[0034] LSTM layer: Contains multiple LSTM cells. Each LSTM cell processes one time step in the sequence data, passing the cell state and hidden state to the next cell. The cell state is used to store long-term information, and the hidden state is used to represent the output of the current time step. The LSTM cell controls the flow of information through the forget gate, input gate, and output gate, effectively capturing the long-term dependencies in time series data.

[0035] Fully connected layer: Maps the hidden state output by the LSTM layer to the predicted environmental pollution level through the fully connected layer. The role of the fully connected layer is similar to that in the environmental quality assessment model. Through the linear combination of weights and biases and the transformation of the non-linear activation function, the predicted value of the environmental pollution level is generated.

[0036] Output layer: Outputs the possible environmental pollution level and its trend in the future period. These prediction results can provide decision-making support for environmental managers, helping them take countermeasures in advance.

[0037] Taking air quality prediction as an example, the input layer receives historical air quality indices and weather forecast data. The LSTM layer captures the temporal variation trend of the air quality index and the impact of weather forecast data on future air quality through multiple LSTM cells. The fully connected layer maps the output of the LSTM layer to the future air quality level. The output layer outputs the possible air quality levels and their trends over a future period, such as the change curve of PM2.5 concentration. Traditional pollution prediction methods often struggle to capture long-term dependencies in time series data, resulting in low prediction accuracy. The LSTM model effectively addresses this issue through the transmission of cell states and hidden states, improving the accuracy and reliability of pollution prediction.

[0038] Embodiment 2: The risk warning model separates data of different risk levels by constructing a hyperplane to achieve accurate warning of environmental risks. The construction steps of the risk warning model include: S1: Use the environmental quality assessment results (such as air quality index, water quality index, soil quality index) and pollution prediction results (such as possible air pollution levels and their trends over a future period) as input features. Use the environmental risk level as the target variable, and the risk level can be set as low, medium, high or multiple specific levels according to actual needs.

[0039] S2: Select the radial basis function (RBF) as the kernel function. The RBF kernel function can map the original feature space to a high-dimensional space, making it easier to linearly separate data points in this high-dimensional space.

[0040] S3: Use the support vector machine (SVM) algorithm to train the model. The SVM algorithm separates data points of different risk levels by finding an optimal hyperplane. This hyperplane is the plane that maximizes the margin between two types of data points. During the training process, the algorithm adjusts the position and direction of the hyperplane to minimize classification errors and maximize the margin.

[0041] S4: For new input data, first map it to the feature space. Then, based on its position relative to the hyperplane in the feature space, determine its risk level. The closer a data point is to the hyperplane, the higher (or lower, depending on the definition of the hyperplane) its risk level may be.

[0042] Optimization steps of the model: Step 1: Adjust the parameters of the SVM, including the penalty parameter C and the kernel function parameters (such as the γ parameter of the RBF kernel). The penalty parameter C controls the degree of penalty for classification errors, and the kernel function parameters affect the mapping method of the feature space.

[0043] Step 2: Evaluate the model performance using the method of grid search combined with cross - validation. Grid search traverses a set of parameter combinations to find the parameter combination that optimizes the model performance. Cross - validation divides the dataset into multiple subsets and takes turns using one subset as the validation set and the remaining subsets as the training set to evaluate the generalization ability of the model.

[0044] Step 3: Validate the optimized model using an independent validation set. The validation set is data that has not been used during the training process and is used to evaluate the accuracy and F1 - score of the model. Accuracy represents the proportion of samples correctly classified by the model, while the F1 - score is the harmonic mean of precision and recall, used to comprehensively evaluate the model performance.

[0045] Step 4: When the model performance reaches the preset standard (such as the accuracy reaching a certain level or the F1 - score reaching a relatively high value), save the model parameters. Obtain a trained risk warning model, which can be used to predict the risk level of new environmental monitoring data.

[0046] Suppose there is an environmental monitoring system that needs to predict the air pollution risk level in a certain area in the future. First, use historical air quality indices and pollution prediction results (such as the change trend of PM2.5 concentration) as input features, and the actual air pollution risk level as the target variable. Then, select RBF as the kernel function and use the SVM algorithm to train the risk warning model. Optimize the model parameters through grid search and cross - validation, and finally obtain a risk warning model with relatively high accuracy and excellent F1 - score. When new environmental monitoring data is input, the model can judge the air pollution risk level of the area according to the distance of its position in the feature space relative to the hyperplane, providing decision - making support for environmental managers.

[0047] Example 3: The response suggestion module provides targeted environmental protection and response measure suggestions based on the risk warning results and guides the implementation of specific response actions. The specific implementation methods of this module include: Define multiple environmental protection and response measure suggestions, which are formulated based on the comprehensive consideration of environmental science, emergency management, and policies and regulations. Each suggestion corresponds to a specific environmental risk level and pollution type to ensure the pertinence and effectiveness of the suggestions. The suggestion content is detailed, including the level of warning notice (such as minor, moderate, severe alerts), the activation conditions of emergency responses (such as pollutant concentration exceeding a certain threshold), and specific environmental protection measures (such as reducing industrial emissions, increasing the frequency of water quality monitoring, etc.).

[0048] The response suggestion module is closely integrated with the risk warning module and receives the warning results output by the risk warning module in real - time. The warning results contain key information such as the risk level and pollution type, which are the basis for matching response suggestions.

[0049] According to the risk level and pollution type in the early warning result, search and match the corresponding response suggestions in the pre-defined response suggestion library. The matching process is based on rules or algorithms to ensure quickly and accurately finding the most appropriate response suggestions.

[0050] Execute specific environmental protection and response measures according to the response suggestions matched by the decision-making engine. The measures may include adjusting industrial emission standards to reduce pollutant emissions; starting a water purification program to improve water quality; issuing public health warnings to remind the public to take necessary protective measures; and implementing an ecological restoration plan to restore the damaged ecological environment.

[0051] Suppose the environmental monitoring system detects abnormal water quality in a certain river area. After the risk early warning module evaluates, it issues a medium-risk early warning and indicates that the pollution type is chemical pollutants. After receiving this early warning result, the response suggestion module immediately searches for the corresponding response suggestions in the suggestion library for medium risk and chemical pollutants. The matched response suggestions may include: starting a water purification program, increasing the frequency of water quality monitoring, and at the same time notifying industrial enterprises along the river to adjust their emission standards to reduce the emission of chemical pollutants. In addition, it may also be recommended to issue a public health warning to remind residents along the river to avoid direct contact with the river water and pay attention to the latest water quality information released by the official. Based on these suggestions, the environmental monitoring system or relevant management departments can quickly take actions to effectively control the spread of pollution and protect public health and the ecological environment.

[0052] Example 4: An efficient method for constructing and managing a response suggestion library is adopted in the response suggestion module. This method specifically includes: Set a hash table as the storage structure: Select a hash table as the data structure for storing response suggestions to achieve fast and efficient storage and retrieval. Each key in the hash table contains a unique identifier, which is used to represent a specific combination of environmental risk level and pollution type. For example, the identifier can be "high risk_chemical pollution" or "medium risk_biological pollution", etc. The value part contains a text string, which represents a specific response suggestion. For example, for "high risk_chemical pollution", the response suggestion may be "Immediately start an emergency response, notify enterprises along the river to stop emissions, and strengthen water quality monitoring".

[0053] Handle hash conflicts: The hash table uses open addressing or chaining to handle conflicts. Open addressing is to find the next free location to store the conflicting key, while chaining is to store the conflicting keys in a linked list. Both methods can effectively handle hash conflicts and ensure the performance and stability of the hash table.

[0054] Hash table structure definition: The structure of the hash table is defined as: H = {(k, v) | k ∈ {Risk level_Pollution type}, v ∈ {Response suggestions}}. Here, k represents the combination of environmental risk level and pollution type, which is the key of the hash table; v represents the corresponding response suggestions, which are the values of the hash table.

[0055] Implementation of the key matcher: The key matcher has a built-in hash function for calculating the hash value of the input combination of environmental risk level and pollution type. The design of the hash function should ensure the uniform distribution of hash values to reduce the occurrence of hash collisions. Using the hash function, the key matcher quickly calculates the hash value of the input combination of environmental risk level and pollution type. Then it looks up the corresponding response suggestions in the hash table and returns them to the response suggestion module for subsequent processing.

[0056] Implementation of the suggestion editor: The suggestion editor provides functions to add, modify, or delete keys and values in the hash table. Users can update the response suggestion library according to the actual environmental conditions and policy changes through the suggestion editor. When users need to add new response suggestions, they can input a new combination of environmental risk level and pollution type, as well as the corresponding response suggestions, through the suggestion editor. When users need to modify or delete existing response suggestions, they can select the corresponding keys and values through the suggestion editor for editing or deletion operations.

[0057] Suppose the environmental monitoring system needs to update the response suggestion library to deal with the newly emerged biological pollution risk. Users can add new response suggestions through the suggestion editor. First, users input a new combination of environmental risk level and pollution type, such as "High risk_Biological pollution". Then they input the corresponding response suggestions, such as "Immediately initiate an emergency response, notify relevant departments to conduct a source investigation of biological pollution, and strengthen water quality and ecological monitoring". The suggestion editor adds the new keys and values to the hash table to complete the update of the response suggestion library. When the environmental monitoring system detects high-risk biological pollution, the key matcher will calculate the corresponding hash value and look up the matching response suggestions in the hash table. The response suggestion module will execute specific environmental protection and response measures according to this suggestion.

[0058] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0059] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The environmental monitoring system based on big data is characterized by: The system comprises: Environmental monitoring data collection module, built using IoT technology, is used to collect air quality data, water quality data, soil data, meteorological data, and ecological data in real time; An environmental quality assessment module builds an environmental quality assessment model. The input data of the environmental quality assessment model is the data collected by the environmental monitoring data acquisition module, and the output data is the environmental quality index, including the air quality index, water quality index and soil quality index. The model training uses a data set containing historical environmental quality monitoring data and corresponding quality indexes, and the corresponding labels are various environmental quality indexes. The pollution prediction module builds a pollution prediction model, the input data of which are the historical environmental quality index and external influencing factor data, and the output data are the predicted possible environmental pollution level and its trend in the future. The model training uses a time series data set containing the historical environmental quality index and external influencing factors, and the corresponding label is the future environmental pollution level. The risk warning module builds a risk warning model. The input data of the risk warning model are the assessment results of the environmental quality assessment model and the prediction results of the pollution prediction model. The environmental risk is warned by combining the information of the two. The model training uses a combined data set containing environmental quality assessment results and pollution prediction results, and the corresponding label is the environmental risk level. The response suggestion module is connected to the risk warning module and generates corresponding environmental protection and response measures suggestions based on the warning results. If the warning result is a high risk, an emergency warning notification will be triggered and a detailed response plan will be provided.

2. The big data-based environmental monitoring system according to claim 1, characterized in that: The environmental quality assessment model is constructed using a convolutional neural network (CNN) algorithm, which extracts the features of input data through convolutional layers and pooling layers. Its network structure includes: Input layer: receives pre-processed environmental monitoring data, and the data format is unified into a multi-dimensional matrix form; Convolution layer: extract data features through multiple convolution kernels to generate feature maps; Pooling layer: downsample the feature map to reduce the data dimension; Fully connected layer: The features output by the pooling layer are mapped to various environmental quality indexes through the fully connected layer; Output layer: Output air quality index, water quality index and soil quality index.

3. The big data-based environmental monitoring system according to claim 1, characterized in that: The pollution prediction model is constructed using the long short-term memory network LSTM algorithm, which captures the long-term dependencies in time series data through LSTM units. Its network structure includes: Input layer: receiving historical environmental quality index and external influencing factor data; LSTM layer: contains multiple LSTM units, each of which processes a time step in the sequence data and passes the cell state and hidden state to the next unit; Fully connected layer: The hidden state output by the LSTM layer is mapped to the predicted environmental pollution level through the fully connected layer; Output layer: Output the possible environmental pollution level and its trend in the future.

4. The big data-based environmental monitoring system according to claim 1, characterized in that: The risk warning model separates data of different risk levels by constructing a hyperplane, and the construction steps include: S1: Environmental quality assessment results and pollution prediction results are used as input features, and environmental risk level is used as the target variable; S2: Select radial basis function as kernel function; S3: Use the support vector machine algorithm to train the model and find the optimal hyperplane; S4: For new input data, its risk level is determined based on the distance between its position in the feature space and the hyperplane.

5. The big data-based environmental monitoring system according to claim 4, characterized in that: The optimization steps of the risk early warning model include: Step 1: Adjust the parameters of the support vector machine, including the penalty parameter C and the kernel function parameters; Step 2: Use grid search combined with cross-validation to evaluate model performance and select the optimal parameter combination; Step 3: Use the validation set to validate the optimized model and evaluate the accuracy and F1 score of the model; Step 4: When the model performance reaches the preset standard, save the model parameters and obtain the trained risk warning model.

6. The big data-based environmental monitoring system according to claim 1, characterized in that: The implementation of the response suggestion module includes: Predefine a variety of environmental protection and response measures suggestions, each corresponding to different environmental risk levels and pollution types. The suggestions cover the level of early warning notification, the conditions for initiating emergency response, and specific environmental protection measures; Receive the warning results from the risk warning module, and match the corresponding response suggestions in the suggestion library according to the risk level and pollution type in the warning results; According to the response recommendations matched by the decision engine, specific environmental protection and response measures are implemented, including adjusting industrial emission standards, initiating water purification procedures, issuing public health warnings and implementing ecological restoration plans.

7. The big data-based environmental monitoring system according to claim 6, characterized in that: The specific method of building a suggestion library is: A hash table is set as the data structure for storing response suggestions, where each key contains a unique identifier, which is used to represent a specific combination of environmental risk level and pollution type; the value part contains a text string, which represents a specific response suggestion; the hash table uses open address method or chain address method to handle conflicts to efficiently store and retrieve suggestion information; the hash table structure is defined as: H={(k,v)|k∈{risk level_pollution type},v∈{response suggestion}}; where k represents the combination of environmental risk level and pollution type, and v represents the corresponding response suggestion.

8. The big data-based environmental monitoring system according to claim 7, characterized in that: The response suggestion module also includes a key matcher, which has a built-in hash function. Using the hash function, the key matcher quickly calculates the hash value of the input environmental risk level and pollution type combination, and then searches the hash table for the response suggestion that matches the hash value.

9. The big data-based environmental monitoring system according to claim 8, characterized in that: The response suggestion module also includes a suggestion editor, which provides an editing function for users to add, modify or delete keys and values ​​in the hash table according to actual environmental conditions and policy changes to update the response suggestion library.

10. The big data-based environmental monitoring system according to claim 1, characterized in that: The environmental monitoring data acquisition module adopts the MQTT protocol to achieve real-time data transmission and cloud storage.

Images