Smart Park Environmental Monitoring and Management System

Abstract

The present invention relates to the field of smart park environment monitoring technology, and in particular to a smart park environment monitoring and management system. It comprises a multi-source environmental data acquisition module, an edge computing module, a dynamic environment modeling module, an adaptive control module and a visual feedback module. The system collects data such as air quality, temperature and humidity, noise and light intensity through a heterogeneous sensor group, and uses edge computing and sparse coding algorithms to perform data compression and key feature extraction. Based on graph neural networks and attention mechanisms, the system constructs a dynamic environmental model to predict environmental changes and identify risks, and the adaptive control module optimizes the park environment control strategy in real time. Finally, the monitoring results are presented in a multi-dimensional visual form. The present invention improves the overall level of park environmental management, optimizes the park environment, and provides strong support for the sustainable development of the park by combining the comprehensiveness of environmental data collection, the efficiency of data processing, and the intelligence of environmental control.

Description

Technical Field

[0001] The present invention relates to the technical field of smart park environment monitoring, and in particular to a smart park environment monitoring and management system. Background Art

[0002] With the acceleration of urbanization and the increasing demand for modern park construction, the intelligent management of parks has become an important part of modern urban development. As a typical scenario for the application of emerging technologies such as the Internet of Things, big data, and artificial intelligence, smart parks can effectively improve the operational efficiency, energy conservation, environmental protection, and security capabilities of parks. Therefore, the research and development of smart park environmental monitoring and management systems has gradually become a hot topic in the industry. At present, there are still the following problems: the existing technology can usually only collect a single or limited type of environmental data, and it is difficult to achieve simultaneous monitoring of multiple environmental factors. Even if the collection of multi-source data can be achieved, there are still problems such as difficulty in data integration and slow response speed, which makes it impossible to understand the environmental status of the park in real time and comprehensively; the existing technology has limited ability to predict environmental change trends, often relying on simple statistical methods, lacking in-depth analysis of complex environmental data, resulting in insufficient prediction accuracy and difficulty in timely identification of potential environmental risks; the environmental control strategies in the existing system are mostly preset or based on simple rules, lacking dynamic adaptability, and difficult to flexibly adjust according to real-time environmental changes, resulting in environmental management measures often lagging behind actual needs, affecting the overall optimization effect of the park environment. Summary of the invention

[0003] To solve the above problems, the present invention provides a smart park environment monitoring and management system, which solves the problem of how to achieve real-time, comprehensive monitoring and integration of various environmental factors in the park, and through in-depth analysis and dynamically adaptive control strategies, thereby improving the accuracy of park environment change trend prediction and the flexibility of environmental management, optimizing the park environment, and providing strong support for the sustainable development of the park.

[0004] To achieve the above object, the technical solution adopted by the present invention is:

[0005] A smart park environment monitoring and management system, comprising a multi-source environment data acquisition module, an edge computing module, a dynamic environment modeling module, an adaptive control module and a visual feedback module connected in sequence;

[0006] The multi-source environmental data acquisition module is used to collect environmental data of the park through a heterogeneous sensor group; the environmental data includes air quality, temperature, humidity, noise and light intensity of the park;

[0007] The edge computing module is used to compress the collected environmental data from the edge nodes and extract key features from the environmental data through a sparse coding algorithm;

[0008] The dynamic environment modeling module is used to construct a dynamic environment model based on the key features using a graph neural network and an attention mechanism to predict environmental change trends and identify potential risks;

[0009] The adaptive control module is used to analyze the output of the dynamic environment model using a fuzzy logic control algorithm to adjust the park environment control strategy in real time;

[0010] The visualization feedback module is used to present the monitoring results and control strategies of multi-source data in a multi-dimensional visualization form by using multi-view geometry technology.

[0011] Furthermore, the operation process of the edge computing module includes the following steps:

[0012] Preprocessing the collected multi-source environmental data, wherein the preprocessing operations include data cleaning, normalization and noise reduction;

[0013] The pre-processed environmental data is compressed using sparse coding algorithm;

[0014] Based on the compressed data, the distributed computing resources of edge nodes are used to extract key features in the environmental data, including the main pollution factors of air quality, temperature change patterns, humidity fluctuation trends, noise source distribution, and time series characteristics of light intensity;

[0015] The key features are extracted in real time on the edge nodes through parallel computing and multi-threaded processing, and the extracted key features are passed to the dynamic environment modeling module.

[0016] Furthermore, the operation process of the dynamic environment modeling module includes the following steps:

[0017] Receiving key features of the edge computing module, constructing an environmental feature matrix, each matrix row represents a time series, and each matrix column represents a different type of environmental feature;

[0018] Based on the environmental feature matrix, a topology optimization algorithm is used to generate a graph structure model, environmental features are mapped into graph nodes, and edge weights between nodes are defined by an asymmetric distance function to generate a preliminary environmental graph;

[0019] Using graph neural networks, we extract features and model relationships on the generated environment graph, use the attention mechanism to adjust the edge weights between nodes, build a dynamic environment model, predict the evolution trend of environmental features, and identify potential environmental risks.

[0020] The structure and parameters of the dynamic environment model are updated in real time through a variant algorithm of recurrent neural network.

[0021] Furthermore, the construction process of the dynamic environment model includes the following steps:

[0022] Based on the time series of the environmental feature matrix, the initial structure of the environmental feature graph is generated using a multi-scale graph construction method. The node and edge structure of the initial graph is defined through clustering analysis and correlation measurement of nodes.

[0023] In the initial environment graph, an optimization algorithm based on spectral graph theory is applied to dynamically adjust the edge weights between nodes;

[0024] A graph convolutional neural network is used to train the optimized graph structure, and the attention mechanism is used to dynamically allocate weights in the node neighborhood to extract high-order features of each node.

[0025] In the model prediction process, a variant algorithm of the recurrent neural network is combined to continuously feedback and adjust the prediction results output by the graph neural network;

[0026] By conducting multi-dimensional analysis of the output results of the dynamic environmental model, potential environmental risks can be automatically identified, including sudden air pollution, abnormal temperature and humidity changes, and abnormal noise, and the corresponding alarm mechanism can be triggered to provide real-time risk warning information.

[0027] Furthermore, the formula of the dynamic environment model is as follows:

[0028] ;

[0029] in, Represents the predicted value of the park environment characteristics at time t+k; and Respectively represent the i-th and j-th environmental characteristics at time t; represents the edge weight between the i-th and j-th environmental features; Represents the weight coefficient of the lth layer of the graph neural network; Represents the entire feature matrix and the adjacency matrix The function that performs graph convolution operations to extract global features and use them to predict future environmental states; L represents the number of layers of the graph neural network, which is used to gradually extract deep information from environmental features; n represents the number of monitored environmental features; m represents the number of historical time steps for each environmental feature, which is used to analyze the change patterns of these features in the time dimension.

[0030] Furthermore, the operation process of the adaptive control module includes the following steps:

[0031] Based on the output results of the dynamic environment modeling module, a multi-objective control strategy is established using a fuzzy logic control algorithm, the control parameters corresponding to different environmental characteristics are fuzzified, and a preliminary control strategy is obtained based on the set fuzzy rules.

[0032] Adopting adaptive optimization algorithm to optimize and adjust the preliminary control strategy, and forming the final environmental control strategy through real-time learning and adjustment of optimized control parameters; the environmental control strategy includes the operating parameter adjustment plan of the environmental conditioning equipment of the air conditioning, ventilation system and lighting system in the park;

[0033] The actual status of the park environment is monitored in real time and compared with the predicted results, and the fuzzy rules and control parameters are dynamically updated through an adaptive adjustment mechanism.

[0034] Furthermore, the formula of the fuzzy logic control algorithm is as follows:

[0035] ;

[0036] in, represents the park environment control output at time t; The weight factor corresponding to the fuzzy rule is used to indicate the influence of different rules in the control decision; represents the fuzzy membership function, which is used to describe the degree of fuzziness corresponding to the environmental feature x(t); represents the consequent control action of the fuzzy rule, which is used to generate specific control strategy output; a represents the number of fuzzy rules; Represents the environmental feature vector at time t.

[0037] Furthermore, the visual feedback module specifically utilizes augmented reality technology to display the real-time status and control effects of the park environment, and interacts through gestures or voice commands.

[0038] Furthermore, the heterogeneous sensors include air quality sensors, temperature and humidity sensors, noise sensors and light sensors.

[0039] The beneficial effects of the present invention are:

[0040] The present invention uses a multi-source environmental data acquisition module, and the system can simultaneously collect multiple environmental data such as air quality, temperature, humidity, noise and light intensity. These data are collected through a heterogeneous sensor group, realizing all-round and multi-dimensional monitoring of the park environment, ensuring the comprehensiveness and diversity of the data. The edge computing module uses a sparse coding algorithm to compress and extract features of environmental data at the edge node, which greatly reduces the burden of data transmission, and can process and extract key environmental features in real time, improving the response speed and real-time performance of the system. The dynamic environment modeling module combines graph neural networks and attention mechanisms to conduct in-depth analysis of the extracted key features and construct an accurate dynamic environment model. This module can not only accurately predict environmental change trends, but also identify potential environmental risks, providing a scientific basis for subsequent control decisions. The adaptive control module analyzes the output results of the dynamic environment model through a fuzzy logic control algorithm, and can adjust the environmental control strategy of the park in real time, ensuring the accuracy and intelligence of environmental control, and improving the environmental management effect of the park. The visual feedback module uses multi-view geometry technology to present the monitoring results and control strategies of multi-source data in a multi-dimensional and visual form, allowing managers to intuitively understand the environmental status and control measures, facilitate quick decision-making, and enhance the user experience and operational convenience of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a module schematic diagram of a smart park environment monitoring and management system of the present invention.

[0042] Figure 2 It is a flowchart of the operation process of the dynamic environment modeling module provided by an embodiment of the present invention.

[0043] Figure 3 It is a flowchart of a dynamic environment model construction process provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0044] See also Figure 1-3 As shown, the present invention relates to a smart park environment monitoring and management system.

[0045] Example

[0046] A smart park environment monitoring and management system, comprising a multi-source environment data acquisition module, an edge computing module, a dynamic environment modeling module, an adaptive control module and a visual feedback module connected in sequence;

[0047] The multi-source environmental data acquisition module is used to collect environmental data of the park through a heterogeneous sensor group; the environmental data includes the air quality, temperature, humidity, noise and light intensity of the park; the heterogeneous sensors include air quality sensors, temperature and humidity sensors, noise sensors and light sensors.

[0048] It should be noted that the multi-source environmental data acquisition module includes multiple heterogeneous sensors, which are connected to the system through a wireless network. Specifically, it includes the following components:

[0049] Air quality sensor: This sensor can be a combination of an optical particle sensor and an electrochemical gas sensor. The optical particle sensor measures the concentration of PM2.5 and PM10 in the air by laser scattering, and the electrochemical gas sensor is used to detect harmful gases such as CO2, CO, O3, NO2, SO2, etc. Each sensor is equipped with a high-precision analog-to-digital converter (ADC) to ensure the accuracy of the data.

[0050] Temperature and humidity sensor: This sensor uses a digital temperature and humidity sensor module, such as DHT22 or SHT31, which has the characteristics of high accuracy and fast response. These sensors are usually built into a package and communicate with the data acquisition module through I2C or single-wire protocol.

[0051] Noise sensor: The noise sensor uses a high-sensitivity microphone module to detect the noise level in the environment and convert it into decibels (dB). The sensor passes through a preamplifier and filter to ensure the clarity and accuracy of the noise signal.

[0052] Light sensor: Light sensor uses photodiode or silicon photocell as a photosensitive element to detect the light intensity in the environment. The output of the sensor is an analog signal, which is converted into a digital signal by ADC and then transmitted to the main control unit.

[0053] Each sensor node contains the above sensors, a microcontroller (such as STM32 or ESP32) and a wireless communication module. The microcontroller is responsible for the collection, preliminary processing and transmission of sensor data. The wireless communication module uses low-power wireless technology such as Zigbee, LoRa or Wi-Fi to ensure that data can be stably transmitted between different nodes.

[0054] Each sensor node is connected to the edge computing module through a wireless communication protocol. For large-scale campuses, LoRa technology can provide long-distance, low-power communication; in smaller campuses, Zigbee or Wi-Fi technology can be used to obtain higher data transmission rates. The sensor nodes have self-organizing network functions and can automatically adjust the network topology according to network conditions. For nodes that cannot be directly connected to the edge computing module, the system supports multi-hop communication and forwards data through other relay nodes to ensure that all data can be reliably transmitted to the edge computing module.

[0055] The edge computing module is used to compress the collected environmental data from the edge nodes and extract key features from the environmental data through a sparse coding algorithm;

[0056] The operation process of the edge computing module includes the following steps:

[0057] Preprocessing the collected multi-source environmental data, wherein the preprocessing operations include data cleaning, normalization and noise reduction;

[0058] The pre-processed environmental data is compressed using sparse coding algorithm;

[0059] Based on the compressed data, the distributed computing resources of edge nodes are used to extract key features in the environmental data, including the main pollution factors of air quality, temperature change patterns, humidity fluctuation trends, noise source distribution, and time series characteristics of light intensity;

[0060] The key features are extracted in real time on the edge nodes through parallel computing and multi-threaded processing, and the extracted key features are passed to the dynamic environment modeling module.

[0061] The dynamic environment modeling module is used to construct a dynamic environment model based on the key features using a graph neural network and an attention mechanism to predict environmental change trends and identify potential risks;

[0062] The operation process of the dynamic environment modeling module includes the following steps:

[0063] Receiving key features of the edge computing module, constructing an environmental feature matrix, each matrix row represents a time series, and each matrix column represents a different type of environmental feature;

[0064] Specifically, key feature data after sparse coding and compression processing are received from the edge computing module. These data include the main pollution factors of air quality, temperature change patterns, humidity fluctuation trends, noise source distribution, and time series characteristics of light intensity. Construct an environmental feature matrix, in which each row of the matrix represents environmental data at a different time point, and each column represents a different type of environmental feature. For example, features such as air quality, temperature, humidity, noise, and light intensity correspond to different columns of the matrix. Encode the change information of the time dimension as the row index of the matrix to ensure that the model can capture the dynamic changes of environmental characteristics.

[0065] Based on the environmental feature matrix, a topology optimization algorithm is used to generate a graph structure model, environmental features are mapped into graph nodes, and edge weights between nodes are defined by an asymmetric distance function to generate a preliminary environmental graph;

[0066] Specifically, each feature in the environmental feature matrix is ​​mapped to a node in the graph structure, and each node represents the state of an environmental feature at a certain point in time. The edge weights between nodes are defined, and the asymmetric distance function is used to calculate the relationship between different features. For example, the impact between air quality and temperature may be directional, resulting in different edge weights. The asymmetric distance function can adjust the edge weights according to the degree of mutual influence between features. Based on the calculated edge weights, a preliminary environmental graph is generated. The graph structure reflects the relationship between different environmental features and their evolution over time.

[0067] Using graph neural networks, we extract features and model relationships on the generated environment graph, use the attention mechanism to adjust the edge weights between nodes, build a dynamic environment model, predict the evolution trend of environmental features, and identify potential environmental risks.

[0068] The structure and parameters of the dynamic environment model are updated in real time through a variant algorithm of recurrent neural network.

[0069] Furthermore, the construction process of the dynamic environment model includes the following steps:

[0070] Based on the time series of the environmental feature matrix, the initial structure of the environmental feature graph is generated using a multi-scale graph construction method. The node and edge structure of the initial graph is defined through clustering analysis and correlation measurement of nodes.

[0071] Specifically, based on the time series data of the environmental feature matrix, the initial structure of the environmental feature map is generated using a multi-scale graph construction method. The multi-scale graph method allows the interaction of environmental features to be captured at different scales, ensuring that the model can handle environmental changes at different time scales. Through clustering analysis, nodes with similar features are clustered together to form the node structure of the initial environmental graph. The clustering method can use algorithms such as K-means or hierarchical clustering. When generating the initial graph structure, it is also necessary to measure the correlation between nodes and determine the edge structure between nodes. The correlation can be calculated by methods such as the Pearson correlation coefficient or mutual information.

[0072] In the initial environment graph, an optimization algorithm based on spectral graph theory is applied to dynamically adjust the edge weights between nodes;

[0073] It should be noted that in the initial environment graph, an optimization algorithm based on spectral graph theory is applied to dynamically adjust the edge weights in the graph to ensure that the graph structure can accurately reflect the relationship between environmental features. On the optimized graph structure, a graph convolutional neural network (GCN) is used to train the graph to further extract high-order features of each node.

[0074] A graph convolutional neural network is used to train the optimized graph structure, and the attention mechanism is used to dynamically allocate weights in the node neighborhood to extract high-order features of each node.

[0075] In the model prediction process, a variant algorithm of the recurrent neural network is combined to continuously feedback and adjust the prediction results output by the graph neural network;

[0076] By conducting multi-dimensional analysis of the output results of the dynamic environmental model, potential environmental risks can be automatically identified, including sudden air pollution, abnormal temperature and humidity changes, and abnormal noise, and the corresponding alarm mechanism can be triggered to provide real-time risk warning information.

[0077] The formula of the dynamic environment model is as follows:

[0078] ;

[0079] in, Represents the predicted values ​​of the park environment characteristics at time t+k, including the future pollution concentration of air quality, temperature change trend, humidity fluctuation, noise level and light intensity; and They represent the i-th and j-th environmental characteristics at time t, such as PM2.5 concentration, temperature, humidity, noise, and light intensity of air quality; represents the edge weight between the i-th and j-th environmental features, reflecting the similarity or correlation between the two features. These edge weights are calculated through graph neural networks and attention mechanisms; Represents the weight coefficient of the lth layer of the graph neural network; Represents the entire feature matrix and the adjacency matrix The function that performs graph convolution operations extracts global features and uses them to predict future environmental states; L represents the number of layers of the graph neural network, which is used to gradually extract deep information from environmental features; n represents the number of monitored environmental features, such as different types of environmental data such as air quality, temperature, humidity, noise, and light intensity; m represents the number of historical time steps for each environmental feature, which is used to analyze the change patterns of these features in the time dimension.

[0080] The adaptive control module is used to analyze the output of the dynamic environment model using a fuzzy logic control algorithm to adjust the park environment control strategy in real time;

[0081] The operation process of the adaptive control module includes the following steps:

[0082] Based on the output results of the dynamic environment modeling module, a multi-objective control strategy is established using a fuzzy logic control algorithm, the control parameters corresponding to different environmental characteristics are fuzzified, and a preliminary control strategy is obtained based on the set fuzzy rules.

[0083] Specifically, the adaptive control module first receives the output results from the dynamic environment modeling module, which include the predicted values ​​of environmental characteristics, such as future air quality, temperature, humidity, noise level and light intensity. According to different environmental characteristics, the corresponding control targets are determined. For example, the air quality control target is to keep the PM2.5 and CO2 concentrations within a safe range, the temperature control target is to keep it in a comfortable range (such as 22°C-26°C), and the humidity control target is to maintain it between 40%-60%. The output results of the dynamic environment model (i.e., the predicted values ​​of environmental characteristics) are fuzzified. For example, the temperature data is divided into three fuzzy sets of "low temperature", "moderate temperature" and "high temperature", and the humidity data is divided into three fuzzy sets of "dry temperature", "moderate temperature" and "humid temperature". The control parameters of equipment such as air conditioners, ventilation systems, and lighting systems are also fuzzified. For example, the temperature setting of the air conditioner is divided into "lower temperature", "maintain temperature" and "increase temperature", the wind speed of the ventilation system is divided into "low speed", "medium speed" and "high speed", and the lighting intensity is divided into "dim temperature", "moderate temperature" and "bright temperature". Based on expert experience and historical data, a fuzzy rule base is established. Each fuzzy rule describes the best control strategy under different combinations of environmental characteristics. For example, a rule can be set as: "If the temperature is 'hot' and the humidity is 'dry', the air conditioning temperature should be 'lowered' and the humidity adjustment should be increased." These rules provide the basis for the reasoning of preliminary control strategies. According to the different control objectives of the park (such as air quality, temperature, humidity, etc.), the corresponding fuzzy rules are designed and the priority of each control objective is set. For example, under extreme high temperature conditions, temperature control may take precedence over humidity control.

[0084] According to the fuzzy environmental feature values ​​output by the dynamic environment model, the rules in the fuzzy rule base are matched. Through fuzzy reasoning, the credibility (i.e., membership) of each rule is obtained, and these memberships are used to calculate the fuzzy output of the preliminary control strategy. According to the matched fuzzy rules, the fuzzy output value of each control parameter is inferred. For example, the air conditioning temperature setting may be inferred to be between "lower" and "maintain", and the ventilation system may be inferred to be between "medium speed" and "high speed". The fuzzy control output is defuzzified by the weighted average method and converted into specific control parameter values. For example, the fuzzy output of the air conditioning temperature is converted from "lower" and "maintain" to a specific temperature setting value (such as 24°C), and the fuzzy output of the ventilation system is converted to a specific wind speed (such as 70% of the maximum wind speed). Based on the defuzzification results, a preliminary environmental control strategy is generated, which includes specific equipment operation parameter settings, such as air conditioning temperature setting, ventilation system wind speed adjustment, and lighting intensity adjustment.

[0085] Adopting adaptive optimization algorithm to optimize and adjust the preliminary control strategy, and forming the final environmental control strategy through real-time learning and adjustment of optimized control parameters; the environmental control strategy includes the operating parameter adjustment plan of the environmental conditioning equipment of the air conditioning, ventilation system and lighting system in the park;

[0086] Specifically, a suitable adaptive optimization algorithm is selected to optimize the preliminary control strategy, such as genetic algorithm (GA), particle swarm optimization algorithm (PSO) or adaptive neural fuzzy inference system (ANFIS). These algorithms can improve the effect of the control strategy by learning and adjusting the optimal control parameters in real time. Based on the preliminary control strategy, the adaptive optimization algorithm is used for multiple iterations to adjust the control parameters to minimize the objective function of the system (such as energy consumption, environmental fluctuations or user discomfort). Each iteration fine-tunes the control strategy according to the current environmental state and predicted trends to achieve the optimal control effect. After the optimized control strategy is verified, the final environmental control strategy is formed. The adaptive adjustment mechanism allows the system to dynamically adjust the control parameters according to the real-time changes in the environment during operation. For example, when the actual temperature deviates from the predicted value, the adaptive optimization algorithm will immediately adjust the air conditioner set temperature to ensure the effectiveness of the environmental control strategy. In the final control strategy, the adaptive optimization algorithm also needs to balance the control objectives of different environmental characteristics to ensure that the air conditioning, ventilation, lighting and other systems work together to avoid conflicts between subsystems. For example, the operation of the ventilation system is optimized while reducing the temperature to avoid discomfort caused by low humidity.

[0087] The actual status of the park environment is monitored in real time and compared with the predicted results, and the fuzzy rules and control parameters are dynamically updated through an adaptive adjustment mechanism.

[0088] Specifically, the adaptive control module monitors the actual state of the campus environment in real time, such as the current temperature, humidity, air quality, and lighting intensity, and compares these data with the prediction results of the dynamic environmental model. Through comparative analysis, the effectiveness of the environmental control strategy can be identified, and whether further adjustments are needed. By monitoring the actual environmental data, the system can detect abnormal situations that are inconsistent with the prediction, such as a sudden increase in actual temperature or a sharp deterioration in air quality. In response to these abnormal situations, the system will trigger an immediate adjustment mechanism to ensure a timely response to the environmental control strategy. During operation, the system continuously collects feedback information on actual environmental data and control effects, and dynamically updates the fuzzy rule base based on this information. The adaptive optimization algorithm runs continuously and dynamically optimizes the control parameters based on real-time environmental data and feedback information.

[0089] Furthermore, the formula of the fuzzy logic control algorithm is as follows:

[0090] ;

[0091] in, represents the park environment control output at time t; The weight factor corresponding to the fuzzy rule is used to indicate the influence of different rules in the control decision; represents the fuzzy membership function, which is used to describe the degree of fuzziness corresponding to the environmental feature x(t); represents the consequent control action of the fuzzy rule, which is used to generate specific control strategy output; a represents the number of fuzzy rules; Represents the environmental feature vector at time t, including various environmental data such as air quality, temperature, humidity, noise and light intensity.

[0092] The visual feedback module is used to present the monitoring results and control strategies of multi-source data in a multi-dimensional visual form using multi-view geometry technology. The visual feedback module also uses augmented reality technology to display the real-time status and control effects of the park environment, and interacts through gestures or voice commands.

[0093] Specifically, the system receives processed data transmitted from the multi-source environmental data acquisition module and the edge computing module in real time. These data include the real-time status of air quality, temperature, humidity, noise and light intensity, as well as the predicted environmental change trend. Based on the park's geographic information system (GIS) data and the actual building layout, a three-dimensional model of the park is generated. This model includes the deployment locations of all sensors, the structure of the building, the green belts, and the location of key facilities. The collected environmental data is mapped into the three-dimensional model. For example, air quality data can be directly displayed in the corresponding area of ​​the park model through color coding. Areas with poor air quality may be marked in red, while areas with better air quality are displayed in green. Temperature and humidity data can be overlaid on different areas of the park in the form of heat maps to show the distribution of temperature or humidity.

[0094] The module supports mainstream augmented reality (AR) devices, such as smartphones or tablets. Users can view the environmental status of the park in real time through these devices. When users scan the park through their mobile devices, the current environmental data will be superimposed on the device screen. In the AR view of the park, important sensor nodes, air purification devices, temperature control equipment, etc. can be marked. Users can click or gaze at these marks to view detailed data, such as historical records for a specific time period or the current operating status of the equipment.

[0095] With the integrated gesture recognition system, users can interact with the system using gestures. The module also supports voice recognition. Users can use voice commands to ask the system to display specific data or adjust environmental control policies. For example, the user can say "show the air quality at 2 pm yesterday", and the system will respond immediately and update the display in the AR view. After the user gives the command, the system will execute it immediately and feed the results back to the AR view. For example, when the user adjusts the temperature setting of a certain area, the system will update the temperature layer of the area in real time and show the effect of the new setting through color changes.

[0096] The module can provide a multi-perspective display of environmental monitoring results. For example, users can switch to a bird's-eye view to view the overall environmental status of the entire park, or choose a first-person perspective to view real-time data at a specific location. Users can conduct in-depth analysis of data in a specific area, such as zooming in on a building to view the temperature and humidity data of different rooms inside it, or analyzing changes in noise levels over a certain period of time. The system will clearly display this data through visualization tools such as line graphs, bar graphs, or pie charts.

[0097] The system can automatically generate environmental monitoring reports based on user selections, including detailed visualization of data charts, trend analysis, risk assessment, and implemented control strategies. Users can choose different report formats, such as PDF or PPT, and can also embed dynamic AR views or 3D models in the report. Users can share the current view or analysis results with team members or managers through mobile terminals. The system supports sharing data and visualization results through social media, email, or dedicated applications to ensure that information can be quickly transmitted and decisions can be made.

[0098] In summary, the present invention adopts a multi-source environmental data acquisition module, combined with a heterogeneous sensor group, to achieve comprehensive monitoring of key environmental data such as air quality, temperature and humidity, noise and light intensity. Through the sparse coding algorithm of the edge computing module and the graph neural network of the dynamic environment modeling module, the system can efficiently compress data and extract key features, accurately predict environmental change trends and identify potential risks, thereby improving the accuracy and real-time performance of environmental monitoring. The adaptive control module uses a fuzzy logic control algorithm combined with the output of the dynamic environment model to adjust the environmental control strategy of the park in real time. The system can adaptively optimize the operating parameters of equipment such as air conditioning, ventilation and lighting according to real-time monitoring data and prediction results, realize dynamic adjustment of the park environment, and effectively improve control accuracy and energy efficiency.

[0099] The visual feedback module in the present invention adopts multi-view geometry technology and augmented reality (AR) technology to provide an intuitive multi-dimensional visual display, and users can interact in real time through gestures or voice commands. This immersive interactive experience not only improves the user's operational convenience, but also makes environmental monitoring and management more intuitive and efficient. The sensor nodes in the multi-source environmental data acquisition module support self-organizing network functions and multi-hop communication technology, and are suitable for campus environments of different sizes. The application of wireless communication technologies such as LoRa, Zigbee and Wi-Fi ensures the stable transmission of sensor data and the reliability of the system, especially in a large-scale campus, where efficient data transmission can be maintained.

[0100] The present invention can continuously learn and adjust the control strategy, dynamically update the fuzzy rule base and optimize the control parameters through the adaptive optimization algorithm and the variant algorithm of the recursive neural network. This intelligent self-learning and optimization mechanism enables the system to continuously improve its performance as the environmental conditions change, ensuring the long-term stability and effectiveness of the park environment management.

[0101] The above implementation modes are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering and technical personnel in the field shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A smart park environment monitoring and management system, characterized in that: It includes a multi-source environmental data acquisition module, an edge computing module, a dynamic environmental modeling module, an adaptive control module and a visual feedback module connected in sequence; The multi-source environmental data acquisition module is used to collect environmental data of the park through a heterogeneous sensor group; the environmental data includes air quality, temperature, humidity, noise and light intensity of the park; The edge computing module is used to compress the collected environmental data from the edge nodes and extract key features from the environmental data through a sparse coding algorithm; The dynamic environment modeling module is used to construct a dynamic environment model based on the key features using a graph neural network and an attention mechanism to predict environmental change trends and identify potential risks; The adaptive control module is used to analyze the output of the dynamic environment model using a fuzzy logic control algorithm to adjust the park environment control strategy in real time; The visualization feedback module is used to present the monitoring results and control strategies of multi-source data in a multi-dimensional visualization form by using multi-view geometry technology; The operation process of the dynamic environment modeling module includes the following steps: Receiving key features of the edge computing module, constructing an environmental feature matrix, each matrix row represents a time series, and each matrix column represents a different type of environmental feature; Based on the environmental feature matrix, a topology optimization algorithm is used to generate a graph structure model, environmental features are mapped into graph nodes, and edge weights between nodes are defined by an asymmetric distance function to generate a preliminary environmental graph; Using graph neural networks, we extract features and model relationships on the generated environment graph, use the attention mechanism to adjust the edge weights between nodes, build a dynamic environment model, predict the evolution trend of environmental features, and identify potential environmental risks. The structure and parameters of the dynamic environment model are updated in real time through a variant algorithm of the recurrent neural network; The construction process of the dynamic environment model includes the following steps: Based on the time series of the environmental feature matrix, the initial structure of the environmental feature graph is generated using a multi-scale graph construction method. The node and edge structure of the initial graph is defined through clustering analysis and correlation measurement of nodes. In the initial environment graph, an optimization algorithm based on spectral graph theory is applied to dynamically adjust the edge weights between nodes; A graph convolutional neural network is used to train the optimized graph structure, and the attention mechanism is used to dynamically allocate weights in the node neighborhood to extract high-order features of each node. In the model prediction process, a variant algorithm of the recurrent neural network is combined to continuously feedback and adjust the prediction results output by the graph neural network; Through multi-dimensional analysis of the output results of the dynamic environmental model, potential environmental risks can be automatically identified, including sudden air pollution, abnormal temperature and humidity changes, and abnormal noise, and the corresponding alarm mechanism can be triggered to provide real-time risk warning information; The formula of the dynamic environment model is as follows: ； in, Represents the predicted value of the park environment characteristics at time t+k; and Respectively represent the i-th and j-th environmental characteristics at time t; represents the edge weight between the i-th and j-th environmental features; Represents the weight coefficient of the lth layer of the graph neural network; Represents the entire feature matrix and the adjacency matrix The function that performs graph convolution operations to extract global features and use them to predict future environmental states; L represents the number of layers of the graph neural network, which is used to gradually extract deep information from environmental features; n represents the number of monitored environmental features; m represents the number of historical time steps for each environmental feature, which is used to analyze the change patterns of these features in the time dimension; The operation process of the adaptive control module includes the following steps: Based on the output results of the dynamic environment modeling module, a multi-objective control strategy is established using a fuzzy logic control algorithm, the control parameters corresponding to different environmental characteristics are fuzzified, and a preliminary control strategy is obtained based on the set fuzzy rules. Adopting adaptive optimization algorithm to optimize and adjust the preliminary control strategy, and forming the final environmental control strategy through real-time learning and adjustment of optimized control parameters; the environmental control strategy includes the operating parameter adjustment plan of the environmental conditioning equipment of the air conditioning, ventilation system and lighting system in the park; Monitor the actual status of the park environment in real time and compare it with the predicted results, and dynamically update the fuzzy rules and control parameters through an adaptive adjustment mechanism; The formula of the fuzzy logic control algorithm is as follows: ； in, represents the park environment control output at time t; The weight factor corresponding to the fuzzy rule is used to indicate the influence of different rules in the control decision; represents the fuzzy membership function, which is used to describe the degree of fuzziness corresponding to the environmental feature x(t); represents the consequent control action of the fuzzy rule, which is used to generate specific control strategy output; a represents the number of fuzzy rules; Represents the environmental feature vector at time t.

2. The smart park environment monitoring and management system according to claim 1 is characterized in that: The operation process of the edge computing module includes the following steps: Preprocessing the collected multi-source environmental data, wherein the preprocessing operations include data cleaning, normalization and noise reduction; The pre-processed environmental data is compressed using sparse coding algorithm; Based on the compressed data, the distributed computing resources of edge nodes are used to extract key features in the environmental data, including the main pollution factors of air quality, temperature change patterns, humidity fluctuation trends, noise source distribution, and time series characteristics of light intensity; The key features are extracted in real time on the edge nodes through parallel computing and multi-threaded processing, and the extracted key features are passed to the dynamic environment modeling module.

3. The smart park environment monitoring and management system according to claim 1 is characterized in that: The visual feedback module specifically utilizes augmented reality technology to display the real-time status and control effects of the park environment, and interacts through gestures or voice commands.

4. The smart park environment monitoring and management system according to claim 1, characterized in that: The heterogeneous sensors include air quality sensors, temperature and humidity sensors, noise sensors and light sensors.

Images