An urban greening system based on big data and its optimization method

Abstract

The present invention discloses an urban greening system based on big data and an optimization method thereof, which relate to the field of artificial intelligence technology and solve the problems of difficulty in integrating heterogeneous data, insufficient real-time analysis capability of dynamic data, and low management efficiency in existing urban greening systems. The present invention collects multi-source heterogeneous data through an API interface and Internet of Things communication, and cleans, aligns, converts and fuses the data via a distributed real-time data stream processing platform. It utilizes image recognition and machine learning algorithms to monitor pests and diseases in real time, and combines an LSTM network to train a growth prediction model, thereby enhancing the system's ability to quickly respond to environmental changes. It adopts a deep Q network to dynamically adjust optimization target weights, and combines mixed integer linear programming, situational awareness and genetic algorithms to construct a multi-objective optimization model to achieve intelligent optimization of irrigation, fertilization, pest control and greening layout. The present invention significantly improves the management efficiency and resource utilization of the urban greening system.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and more particularly to an urban greening system based on big data and an optimization method thereof. Background Art

[0002] With the acceleration of urbanization, urban greening has become a crucial tool for enhancing urban livability and sustainable development. Modern urban greening involves not only plant cultivation and maintenance but also encompasses multiple aspects, such as ecological and environmental protection, water resource management, and public participation. The development of big data technology has brought new opportunities for urban greening management. By collecting and analyzing various data, such as meteorological data, soil moisture data, plant growth data, and remote sensing imagery, refined management can be achieved, resource utilization efficiency can be improved, and greening effects can be optimized. Furthermore, the application of the Internet of Things (IoT) technology makes data collection more convenient and real-time, providing powerful technical support for urban greening management. However, despite the enormous potential of big data technology in urban greening management, its practical application still faces numerous challenges, urgently requiring further research and improvement.

[0003] Existing big data-based urban greening management systems integrate diverse data sources, such as weather stations, soil moisture sensors, remote sensing satellites, and drones, to collect vast amounts of data to support refined management and decision-making. These systems typically employ data fusion techniques to integrate data from various sources for comprehensive analysis. Some systems, by integrating meteorological and soil moisture data, can more accurately predict plant growth and irrigation needs. Furthermore, existing systems utilize machine learning and artificial intelligence technologies for data mining and predictive analysis.

[0004] However, existing urban greening management systems still face several technical challenges in practical application. First, heterogeneous data integration is difficult. Urban greening management systems involve a vast array of data types, including but not limited to environmental sensor data (such as temperature, humidity, and light intensity), geographic information data (such as soil type and distribution), plant growth cycle data, and human activity data (such as pedestrian density and activity areas). These data come from diverse sources with varying formats, units, and sampling frequencies, resulting in complex and time-consuming cleaning, alignment, and conversion steps during data fusion. Second, real-time analysis capabilities for dynamic data are insufficient. Urban greening management systems require real-time monitoring and response to environmental changes, but existing systems are often based on static models and fixed rules, unable to effectively address dynamic changes. For example, systems may rely on models built from historical data and are unable to adjust irrigation plans in a timely manner to address extreme weather conditions such as droughts or heavy rains. Finally, management efficiency is low. Existing systems lack multi-objective optimization capabilities and struggle to simultaneously optimize multiple objectives, such as plant growth, water resource utilization, and environmental protection. This leads to resource conflicts and inefficient management.

[0005] Therefore, the present invention discloses an urban greening system based on big data and an optimization method thereof. Summary of the Invention

[0006] In view of the shortcomings of the existing technology, the present invention discloses an urban greening system based on big data and an optimization method thereof, aiming to solve the problems raised in the above background technology.

[0007] The present invention adopts the following technical solutions:

[0008] A method for optimizing urban greening based on big data, comprising the following steps:

[0009] S1. Obtain temperature and humidity, light intensity, soil moisture, and air quality data from environmental sensors through API interfaces and IoT wireless communication methods; obtain plant species and growth cycle data from the plant growth cycle database; obtain crowd density and activity area data from public data sources; and regularly obtain image data of green areas from the drone inspection system and store them in a temporary buffer;

[0010] S2. Input the data into a distributed real-time data stream processing platform to clean, align, transform, and fuse the collected data; the distributed real-time data stream processing platform integrates a unified format data set, including environmental parameters, vegetation growth status, and human activity patterns, in a unified format, through a deep heterogeneous fusion algorithm;

[0011] S3, based on the drone inspection system's green area image data, identifies pest and disease symptoms through image recognition-based machine learning algorithms. When abnormal environmental data is detected, an event-driven mechanism automatically triggers corresponding emergency response measures, including increasing monitoring frequency and adjusting irrigation strategies.

[0012] S4. Build a growth prediction model by training historical environmental data, plant growth data, and pest and disease symptoms through a long short-term memory network.

[0013] S5. Inputting the real-time collected environmental data into the growth prediction model to obtain growth prediction results for future environmental changes, and generating optimization strategies including irrigation plans, fertilization strategies, pest and disease control measures, and green area layout based on the prediction results;

[0014] S6. Define optimization objectives and initial weights, including plant growth status, pest and disease symptoms, water resource utilization efficiency, and environmental protection effects, and dynamically adjust the weight of each objective according to environmental changes through a deep Q-network algorithm;

[0015] S7. Constructing a context-based multi-objective optimization model using mixed integer linear programming, context-awareness, and genetic algorithms, inputting real-time environmental data, pest and disease data, growth prediction data, and optimization strategies into the context-based multi-objective optimization model to obtain an optimal comprehensive management strategy, including irrigation duration, fertilizer type, fertilizer amount, fertilizer frequency, and plant location and type, and storing the output results in a database;

[0016] S8. Sending the output results of the scenario-based multi-objective optimization model to a control system, and dynamically adjusting the operating parameters of the irrigation system and fertilization equipment and facilities through an adaptive control algorithm according to the optimization strategy.

[0017] As a further technical solution of the present invention, the working method of the deep heterogeneous fusion algorithm is as follows: missing values ​​are filled by mean interpolation, median interpolation and K-nearest neighbor interpolation, and outliers are identified and processed by the Z-Score method; timestamp alignment is performed by linear interpolation and spline interpolation; spatial coordinate alignment is performed by geocoding and coordinate transformation; based on the aligned data, the deep heterogeneous fusion algorithm fuses the features of data from different sources through multimodal data; the multimodal data fusion extracts the features of image and time series data through convolutional neural networks and recurrent neural networks, and adopts mutual information method, principal component analysis and feature engineering to generate a feature vector F through a dynamic weight adjustment mechanism. The fused feature vector F is used as the input of the deep neural network, and the nonlinear mapping relationship between input data and output data is learned through a multilayer perceptron and an autoencoder to generate a data set in a unified format; the multilayer perceptron learns the nonlinear mapping relationship between input data and output data through forward propagation and backpropagation algorithms, introduces nonlinear characteristics through the activation function ReLU, and improves the expressive ability of the model; the autoencoder learns the low-dimensional representation of the data through the encoder and decoder, and optimizes the model parameters through the mean square error MSE.

[0018] As a further technical solution of the present invention, the working principle of the machine learning algorithm based on image recognition is:

[0019] Step 301: Preprocess the image data collected by the drone through data cleaning and image enhancement. The data cleaning removes noise in the image through Gaussian filtering, and the image enhancement enhances the image details through histogram equalization and contrast adjustment.

[0020] Step 302: extract features from the image using a convolutional neural network. The convolutional neural network performs a convolution operation on the image through a convolution layer to extract local features, and performs a maximum pooling operation on the feature map through a pooling layer to reduce computational complexity. The output of the convolution layer is nonlinearly transformed using an activation function.

[0021] Step 303: Locate and identify pest symptoms in the extracted image features using a target detection algorithm; the target detection algorithm works as follows:

[0022] r1. Generate candidate regions of pest and disease symptoms through sliding window and region proposal network.

[0023] r2. Extract features from the candidate region through a convolutional neural network and generate a feature vector;

[0024] r3. Classify the feature vector and perform bounding box regression through a fully connected layer and a cross entropy loss function to determine the location and category of the pest and disease symptoms;

[0025] r4. Ensure the accuracy and reliability of the machine learning algorithm through data labeling and model training. The data labeling is performed by manually labeling the pest and disease symptoms in the image or semi-supervised learning to generate a training data set. The model training is performed by backpropagation algorithm and SGD optimization algorithm.

[0026] r5. The identified pest and disease symptoms are transmitted to the central server in real time through an event-driven mechanism.

[0027] As a further technical solution of the present invention, the working method of the event-driven mechanism is: reading pest and disease symptom data from the message queue through the event processor, generating response measure events, and pushing the response measure events to the message queue, triggering emergency response measures through the rule engine and decision tree algorithm, including increasing the monitoring frequency, adjusting the irrigation strategy, and initiating pest and disease control measures.

[0028] As a further technical solution of the present invention, the growth prediction model includes an input layer, an embedding layer, a long short-term memory layer, an attention mechanism layer, a fully connected layer, and an output layer; the operation process steps of the growth prediction model are as follows:

[0029] Step 501: receiving and processing input data through the input layer, including historical environmental data, plant growth data, and pest and disease symptoms;

[0030] Step 502: The input data is converted into a high-dimensional vector representation by the embedding layer to capture data features; the embedding layer converts plant species and pest types into continuous vector representations by using a word embedding method;

[0031] Step 503: The long short-term memory layer receives the high-dimensional vector output by the embedding layer, captures the long-term dependency in the time series data, and extracts the time series features; the long short-term memory layer dynamically adjusts the retention and discarding of information through a gating mechanism;

[0032] Step 504: Enhance the model's attention to important time steps through the attention mechanism layer; the attention mechanism uses the attention mechanism to calculate the weight of each time step, and the weighted sum is used as supplementary information output by the long short-term memory layer to enhance the model's sensitivity to key time steps;

[0033] Step 505: The fully connected layer performs a nonlinear transformation on the extracted features to generate a prediction result; the fully connected layer performs a nonlinear transformation on the outputs of the long short-term memory layer and the attention mechanism layer through the ReLU activation function, and maps the features extracted by the long short-term memory layer and the attention mechanism layer to the output space;

[0034] Step 506: Output the final growth prediction result through the output layer, including the plant growth status and pest and disease prediction under future environmental changes; the output layer generates a binary classification prediction result through the Sigmoid activation function and generates a multi-classification prediction result through the Softmax activation function.

[0035] As a further technical solution of the present invention, the method for generating an irrigation plan, fertilization strategy, pest and disease control measures and an optimization strategy for greening area layout based on the prediction results in S5 is as follows: based on vegetation growth requirements and soil moisture data, an irrigation plan is generated through an expert system, including irrigation time, frequency and water volume; based on vegetation growth status and soil nutrient data, a fertilization plan is generated through a nutritional model, including fertilizer type, amount and frequency; based on pest and disease warning information and vegetation growth status, prevention and control measures are formulated through an expert system, including spraying pesticides and pruning diseased branches; based on human activity patterns and vegetation growth status, the layout of greening areas and vegetation types are generated through spatial planning methods.

[0036] As a further technical solution of the present invention, the dynamic adjustment method of the deep Q network algorithm is:

[0037] Step 701: define optimization objectives, including plant growth status, pest and disease symptoms, water resource utilization efficiency, and environmental protection effects, and set initial weight values ​​for each optimization objective;

[0038] Step 702: Construct a deep neural network using convolutional neural networks and fully connected layers. This includes an online network and a target network. The online network is used to update and calculate the Q value of the current strategy in real time. The weights of the target network are periodically copied from the online network to calculate the target Q value to stabilize the training process. The convolutional neural network extracts features of the input data through convolutional layers and pooling layers to generate feature vectors. The fully connected layer maps the feature vectors to Q values, i.e., the expected reward value for each action, through a multilayer perceptron.

[0039] Step 703: Collect environmental data in real time through the sensor network and the drone inspection system, including temperature, humidity, light intensity, and soil moisture. Preprocess the collected environmental data through normalization and standardization methods to construct an initialized environmental state space.

[0040] Step 704: Combine the collected environmental data and the state information of the optimization target into a state vector as the input of the deep Q network; the state vector includes the current environmental data and the state information of the optimization target;

[0041] Step 705: Select an action using the adaptive decay mechanism of the greedy strategy. The greedy strategy randomly selects an action or selects the action with the highest Q value in the current state based on the exploration rate. The exploration rate is gradually reduced during training to balance exploration and exploitation.

[0042] Step 706: Execute the selected action and collect environmental data and optimization target status information after the action is executed through the sensor network and drone inspection system. If the plant growth condition improves, a positive reward is given; if the pest and disease symptoms worsen, a negative reward is given.

[0043] Step 707: Update the Q value using the Bellman equation. The Bellman equation calculates the new Q value using the discount factor γ and the immediate reward r. The mean square error loss function (MSE) is used to calculate the difference between the current Q value and the target Q value, and the network parameters are updated through the back-propagation mechanism.

[0044] Step 708: Periodically copy the weights of the online network to the target network through weight repetition to maintain the stability of the target value and reduce fluctuations during training;

[0045] Step 709: Continuously execute steps 704 to 708, interact with the environment, collect data, update network weights, and stop when the predetermined number of training rounds or convergence criteria are met.

[0046] As a further technical solution of the present invention, the context-based multi-objective optimization model includes a data reading layer, a feature extraction layer, a context-awareness layer, a goal setting layer, an optimization modeling layer, and an algorithm solving layer; the working method based on context-awareness and the multi-objective optimization model is:

[0047] Step 801: Read pre-processed data from the central database through the data layer; including real-time environmental data, pest and disease data, and growth prediction data;

[0048] Step 802: extract key features of the data through the feature extraction layer; the feature extraction layer extracts features through a principal component analysis method and stores the extracted feature data in a feature database;

[0049] Step 803: The context perception layer reads the processed feature data from the feature database and identifies the current context state, including drought, heavy rain, high population density, high incidence of pests and diseases, and vigorous plant growth period. The context feature perception layer classifies the data into different categories through k-cluster analysis to identify the current context state. The k-cluster analysis selects the best clustering result through multiple initializations, and the formula expression is:

[0050]

[0051] In formula (1), c i represents the i-th cluster center; X represents the set of all data points; ||Xc j || represents the data point X and the selected cluster center c j The Euclidean distance between j represents the variance of the jth cluster center, which is used to adjust the influence of distance;

[0052] Step 804: Optimization goals are set by the goal setting layer, including plant growth status, pest and disease symptoms, water resource utilization efficiency, and environmental protection effects. The goal setting layer uses an expert system combined with historical data to set initial weight values ​​for each optimization goal. The goal setting layer dynamically adjusts the weight values ​​based on actual environmental changes and management needs through a feedback mechanism based on reinforcement learning, continuously optimizing the weight settings.

[0053] Step 805: Construct a mixed integer linear programming model through the optimization modeling layer, which includes the objectives of maximizing plant growth efficiency, minimizing the incidence of pests and diseases, and optimizing resource use costs. The mixed integer linear programming model dynamically adjusts the weights between different objectives based on real-time context using a dynamic adjustment function. The formula of the dynamic adjustment function is:

[0054]

[0055] In formula (2), f(x) represents the comprehensive objective function; w i (t) represents the weight of the i-th target at time t; g i (x) represents the value of the i-th objective function; γ i represents the penalty coefficient of the i-th objective, which is used to adjust the nonlinear effect of the objective function; θ i represents the threshold of the i-th target, which is used to define the benchmark value of the objective function; β i represents the smoothing parameter of the i-th objective, which is used to control the smoothness of the objective function;

[0056] Step 806: Generate the optimal management strategy through the algorithm solution layer, including irrigation duration, fertilizer type, fertilizer quantity, fertilizer frequency, and plant location and type; the optimization solution layer solves the optimization model through a genetic algorithm, which gradually optimizes individuals in the population through selection, crossover and mutation operations, and handles conflicts between objectives through non-dominated sorting and crowding distance calculation; the algorithm solution layer also continuously monitors the optimization effect through a feedback mechanism based on online learning, and dynamically adjusts the optimization strategy according to actual conditions.

[0057] As a further technical solution of the present invention, the adaptive control algorithm realizes dynamic adjustment of the irrigation system and fertilization equipment through PID control algorithm and fuzzy control, and ensures that the system maintains optimal performance in a dynamic environment through an online learning algorithm; wherein, the PID control algorithm dynamically adjusts control parameters through proportional control, integral control and differential control; the fuzzy control converts complex nonlinear relationships into fuzzy rules through a fuzzy inference system and makes decisions based on the fuzzy rule base; the online learning algorithm updates the control parameters in real time through recursive least squares method.

[0058] As a further technical solution of the present invention, a big data-based urban greening system includes a multi-source data acquisition module for collecting multimodal data of urban greening areas, including temperature and humidity, light intensity, soil moisture, air quality, plant species, growth cycle, vegetation growth status, and pedestrian density data;

[0059] A data integration module is used to clean, align, convert and fuse the multimodal data collected by the multi-source data collection module through a deep heterogeneous fusion algorithm;

[0060] The real-time data analysis module is used to analyze the integrated data in real time. The analysis content includes vegetation growth trend prediction, pest and disease warning, and human activity pattern recognition. The module also outputs real-time analysis results, including irrigation demand prediction, fertilization recommendations, pest and disease warning information, and human activity hotspots.

[0061] An adaptive decision-making module, which uses mixed-integer linear programming, situational awareness, and genetic algorithms to build a context-based multi-objective optimization model based on real-time analysis results. This module dynamically adjusts irrigation plans, fertilization strategies, pest and disease control measures, and greening area layout, enabling data-driven intelligent management.

[0062] User interaction and feedback module, which is used to provide user interaction functions, including user query, feedback, and activity participation, while displaying real-time data and results of greening management;

[0063] The output end of the multi-source data acquisition module is connected to the input end of the data integration module; the output end of the data integration module is connected to the input end of the real-time data analysis module; the output end of the real-time data analysis module is connected to the input end of the adaptive decision module; and the user interaction feedback module is connected to the above modules.

[0064] The positive beneficial effects of the present invention are:

[0065] 1. Data is collected from a variety of sources through APIs and IoT wireless communication technologies. These include not only environmental sensors but also plant growth cycle databases, public data sources, and drone inspection systems. This data is centrally stored in a temporary buffer. All data is then imported into a distributed real-time data stream processing platform, which utilizes deep heterogeneous fusion algorithms to clean, align, transform, and fuse the data, ensuring data consistency and usability. This process resolves the issues of inconsistent data formats, units, and sampling frequencies caused by diverse data sources, laying a solid foundation for subsequent advanced analytics.

[0066] 2. Based on the image data of green areas collected by the drone inspection system, combined with machine learning algorithms for image recognition, the system can efficiently identify symptoms of pests and diseases. Furthermore, when the system detects abnormal environmental data, it can automatically trigger emergency response measures, such as increasing monitoring frequency or adjusting irrigation strategies immediately. This mechanism significantly enhances the system's response speed and accuracy to environmental changes. Furthermore, by training historical data using a long short-term memory (LSTM) network, an accurate growth prediction model is constructed. This enables the system to make appropriate adjustments in advance in the face of unforeseen circumstances such as extreme weather, further improving the system's dynamic adaptability.

[0067] 3. In order to overcome the limitations of existing systems in multi-objective optimization, this solution adopts a deep Q-network algorithm to dynamically adjust the weight values ​​of optimization objectives according to environmental changes, ensuring a balance between different objectives. Specifically, the system defines optimization objectives and initial weight values ​​based on multiple dimensions such as plant growth conditions, pest and disease symptoms, and water resource utilization efficiency. In addition, a context-based multi-objective optimization model is constructed through mixed integer linear programming, context perception, and genetic algorithms. This model can comprehensively consider factors such as real-time environmental data, pest and disease data, and growth forecast data to generate the most optimized comprehensive management strategy. Ultimately, these strategies are applied in actual operations, such as adjusting irrigation duration, selecting appropriate fertilizer types and quantities, planning plant locations and types, etc., thereby achieving effective resource allocation and improving overall management efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1This is a step diagram of an urban greening optimization method based on big data according to the present invention;

[0069] Figure 2 This is a structural framework step diagram of the growth prediction model of the present invention;

[0070] Figure 3 This is a schematic diagram of the urban greening system based on big data of the present invention;

[0071] Figure 4 This is a step diagram of the dynamic adjustment method of the deep Q network algorithm of the present invention. DETAILED DESCRIPTION

[0072] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0073] In the present invention, Figure 1-Figure 4 As shown: A method for optimizing urban greening based on big data includes the following steps:

[0074] S1. Obtain temperature and humidity, light intensity, soil moisture and air quality data from environmental sensors through API interfaces and IoT wireless communication methods, obtain plant species and growth cycle data from plant growth cycle databases, obtain human density and activity area data from public data sources, and regularly obtain image data of green areas from drone inspection systems and store them in temporary buffers; S2. Input the data into a distributed real-time data stream processing platform to clean, align, convert and fuse the collected data; the distributed real-time data stream processing platform integrates data sets in a unified format through a deep heterogeneous fusion algorithm, including environmental parameters, vegetation growth status, and human activity patterns in a unified format; S3. Based on the green area image data of the drone inspection system, identify pest and disease symptoms through an image recognition-based machine learning algorithm, and automatically trigger corresponding emergency response measures when abnormal environmental data is detected through an event-driven mechanism, including increasing monitoring frequency and adjusting irrigation strategies; S4. Train historical environmental data, plant growth data, and pest and disease symptoms through a long short-term memory network to construct a growth prediction Model; S5, input the real-time collected environmental data into the growth prediction model to obtain the growth prediction results of future environmental changes, and generate an optimization strategy including irrigation plan, fertilization strategy, pest and disease control measures and greening area layout based on the prediction results; S6, define the optimization objectives and initial weight values, including plant growth status, pest and disease symptoms, water resource utilization efficiency, and environmental protection effect, and dynamically adjust the weight value of each objective according to environmental changes through the deep Q network algorithm; S7, construct a situation-based multi-objective optimization model through mixed integer linear programming, situational perception and genetic algorithm, input the real-time environmental data, pest and disease data, growth prediction data and optimization strategy into the situation-based multi-objective optimization model, obtain the optimal comprehensive management strategy, including irrigation time, fertilizer type, fertilizer quantity, fertilizer frequency and plant location and type, and store the output results in a database; S8, send the output results of the situation-based multi-objective optimization model to the control system, and the control system dynamically adjusts the operating parameters of the irrigation system and fertilization equipment facilities through the adaptive control algorithm according to the optimization strategy.

[0075] In practice, S2 fills missing values ​​using mean interpolation, median interpolation, and K-nearest neighbor interpolation. Mean interpolation is applicable to numerical data and fills missing values ​​by calculating the mean of the same feature; median interpolation is applicable to numerical data and fills missing values ​​by calculating the median of the same feature; and K-nearest neighbor interpolation is applicable to both numerical and categorical data and fills missing values ​​by finding the values ​​of similar samples. Outliers are then identified and processed using the Z-Score method. The Z-Score method calculates a standardized score for each data point and identifies data points whose standardized scores exceed a certain threshold (e.g., 3σ) as outliers, which are then corrected or deleted. Timestamp alignment is then performed using linear interpolation and spline interpolation. Linear interpolation is applicable to data with small time intervals and fills missing values ​​within the time interval; spline interpolation is applicable to data with large time intervals and fills missing values ​​by fitting a time series curve. Spatial coordinate alignment is performed using geocoding and coordinate transformation. Geocoding converts address information into longitude and latitude coordinates, while coordinate transformation unifies data from different coordinate systems into the same coordinate system. Based on the aligned data, convolutional neural networks (CNNs) and recurrent neural networks (RNNs) are used to extract features from image and time series data. Mutual information, principal component analysis (PCA), and feature engineering are employed to generate feature vectors through a dynamic weight adjustment mechanism. The fused feature vectors serve as input to the deep neural network, where a multilayer perceptron (MLP) and autoencoder (AE) learn the nonlinear mapping between input and output data, generating a unified data set. The nonlinear mapping between input and output data is learned through forward and backpropagation algorithms, and the ReLU activation function introduces nonlinear characteristics to enhance the model's expressiveness. The encoder and decoder learn a low-dimensional representation of the data, and the model parameters are optimized using mean squared error (MSE).

[0076] In practice, the deep heterogeneous fusion algorithm generates a comprehensive dataset encompassing environmental parameters, vegetation growth status, and pedestrian activity patterns, providing comprehensive decision-making support for urban greening management. Based on this data, managers can more accurately assess greening effectiveness, optimize resource allocation, and enhance urban greening levels. S2's real-time processing and high reliability ensure data timeliness and accuracy, enabling urban greening management to rapidly respond to environmental changes and emergencies. This not only improves management efficiency but also reduces costs.

[0077] In practice, the hardware environment for the deep heterogeneous fusion algorithm includes: high-performance computing servers equipped with multi-core CPUs and high-performance GPUs; large-capacity, high-speed storage devices; a distributed real-time data stream processing platform for real-time data collection, cleaning, alignment, conversion, and fusion; stable network communication equipment; and a power and cooling system.

[0078] Based on the above hardware equipment, a comparative experiment was designed to verify the advantages of the deep heterogeneous fusion algorithm in processing big data and performing data fusion. The experiment included the deep heterogeneous fusion algorithm (Group A) and the rule-based data fusion algorithm (Group B). The rule-based data fusion algorithm relies on predefined rules and logic to integrate data from different sources. The same data set was selected and experiments were conducted on Group A and Group B respectively. Each group of experiments was conducted five times to ensure the stability and reliability of the results. The running time, data fusion accuracy, data fusion completeness, CPU usage, memory usage, and disk I / O parameters of each experiment were recorded. The experimental data record table is shown in Table 1:

[0079] Table 1 Experimental data record table

[0080]

[0081] Comparative experimental results show that the deep heterogeneous fusion algorithm (Group A) outperforms the traditional algorithm (Group B) in terms of runtime, data fusion accuracy, and data fusion completeness. Group A's average runtime was only approximately 120 seconds, while Group B required approximately 240 seconds, demonstrating that the deep heterogeneous fusion algorithm is more efficient when processing big data. In terms of data fusion accuracy, Group A achieved an average accuracy of 98.6%, significantly higher than Group B's 90.0%, demonstrating the deep heterogeneous fusion algorithm's superior data fusion accuracy. Group A also achieved better data fusion completeness than Group B, demonstrating that the deep heterogeneous fusion algorithm is able to more completely preserve the features and information of the original data. Furthermore, Group A demonstrated superior performance in terms of resource utilization, with lower CPU utilization, memory usage, and disk I / O than Group B. In summary, the deep heterogeneous fusion algorithm exhibits significant advantages in processing big data and performing data fusion, providing strong technical support for the intelligent management of urban greening systems.

[0082] Specifically, the working principle of the deep heterogeneous fusion algorithm is as follows: first, missing values ​​are filled, including mean interpolation: when the data distribution is relatively uniform, the mean is used to fill the missing values. Median interpolation: for skewed distributed data, the median can better represent the central trend of the data. K nearest neighbor interpolation: missing values ​​are filled according to the values ​​of the K nearest neighbors, and the proximity is calculated using the Euclidean distance; after standardizing the data, the Z-Score method is used to identify and process outliers that exceed the threshold. The expression is Where u is the original data, k is the mean, a kis the standard deviation. The z value indicates the degree to which the data point deviates from the mean. If the absolute value of the z value exceeds the set threshold, the data point is treated as an outlier and processed. Next, for data with uniform time intervals, linear interpolation is used to align timestamps. For data with uneven time intervals, spline interpolation is used to align timestamps. Spline interpolation is a more flexible and accurate interpolation method that approximates the original data by constructing a piecewise polynomial function. For data with spatial attributes, we need to convert the address into latitude and longitude coordinates through geocoding, and convert data in different coordinate systems into a unified coordinate system through coordinate conversion.

[0083] In the feature extraction stage, the convolution operation formula of the convolutional neural network is expressed as r = h + f (Wx + b), where x is the input image, W is the convolution kernel, b is the bias term, f is the convolution operation, and h is the activation function used to introduce nonlinear characteristics. The convolution kernel slides on the input image and calculates the weighted sum to extract features. Next, the loop structure is used to capture the time dependency in the time series data. The time series data contains dynamic information in time. The formula is expressed as h t =1+ε(W h h t-1 +W x x t +b), h c is the hidden state at time step t, W h and W x is the weight matrix, x t is the input data at time step t, b is the bias term, and ε is the activation function; the temporal dependency in time series data is captured by transferring the hidden state between time steps.

[0084] After the feature extraction stage, these feature vectors are fused into a feature vector of a unified format, wherein the weights are dynamically adjusted according to the importance of the feature vectors through a dynamic weight adjustment mechanism. The dynamic weight adjustment mechanism fuses the feature vectors from different sources into a feature vector of a unified format through weighted summation, i.e., by using ρ, τ and These three dynamic weights dynamically adjust feature vectors from CNNs, RNNs, and other sources. To learn the nonlinear mapping between input and output data, a deep learning model is trained using a multilayer perceptron (MLP) and an autoencoder. The MLP uses a multi-layer neural network structure to learn the nonlinear mapping between input and output data. First, a weighted sum is calculated, and nonlinear characteristics are introduced through an activation function to obtain the output. The autoencoder then performs the encoding and decoding process, minimizing the reconstruction error to learn a low-dimensional representation of the data.

[0085] In the specific implementation of S3, this step first involves using drones to regularly inspect green areas according to a predetermined flight route and schedule, capturing high-definition images. The drones are equipped with high-resolution cameras and GPS positioning systems to ensure image clarity and accurate geographic location information. The drone images are then subjected to Gaussian filtering to remove noise and reduce environmental interference. Gaussian filtering smoothes the image using a convolution kernel, preserving key features. Next, histogram equalization and contrast adjustment are used to enhance image detail. Histogram equalization redistributes pixel values ​​to achieve a more uniform brightness and contrast. Contrast adjustment adjusts the pixel value range to enhance image detail. The preprocessed images are then passed through a convolutional neural network for feature extraction. The convolution layer applies multiple filters to the image to generate a feature map. The pooling layer performs a max pooling operation on the feature map to reduce computational complexity. Activation functions (such as ReLU) apply nonlinear transformations to the convolution layer outputs, increasing the model's expressive power. Next, candidate regions representing pest and disease symptoms are generated. A sliding window method slides across the image at different locations and scales to generate candidate regions. The RPN uses a convolutional neural network to generate more accurate candidate regions, improving detection efficiency. Feature extraction is then performed on the candidate regions to generate feature vectors. This process typically uses deep feature maps from the convolutional neural network. Fully connected layers and the cross-entropy loss function are used to classify and regress the feature vectors to determine the location and category of the pest and disease symptoms. The fully connected layers map the feature vectors to class labels and bounding box coordinates. The cross-entropy loss function is used to optimize the classification task, and the smoothed L1 loss function is used to optimize the bounding box regression task. Pest and disease symptoms in images are annotated using manual annotation or semi-supervised learning to generate a training dataset. Manual annotation ensures data accuracy, while semi-supervised learning improves model generalization by using a small amount of annotated data and a large amount of unlabeled data. The model is trained using backpropagation and optimization algorithms (such as SGD) to optimize model parameters and improve accuracy and reliability. Finally, the identified pest and disease symptoms are transmitted to a central server in real time through an event-driven mechanism. Based on the detection results, the central server automatically triggers appropriate emergency response measures, such as increasing monitoring frequency and adjusting irrigation strategies.

[0086] The event-driven mechanism uses an event processor to read pest and disease symptom data from the message queue, generate response action events, and push the response action events to the message queue. Emergency response measures are triggered through the rule engine and decision tree algorithm. The operation process of the event processor is as follows:

[0087] Initialize the event handler: Set the connection parameters of the message queue and initialize the event handler.

[0088] Listening to the message queue: The event processor continuously listens to the message queue, waiting to receive pest and disease symptom data.

[0089] Read data: When there is new pest and disease symptom data in the message queue, the event processor reads the data.

[0090] Generate response measure events: Generate corresponding response measure events based on the read pest and disease symptom data.

[0091] Push events: Push the generated response action events to the message queue for subsequent processing.

[0092] In specific implementation, the drone inspection system and image recognition technology can quickly and accurately detect pest and disease symptoms in green areas, identifying potential problems promptly. This helps prevent the spread of pests and diseases and protect greening achievements. Furthermore, the highly automated inspection process reduces reliance on manual labor and reduces labor costs. The drone inspection system can operate automatically at predetermined times and routes, alleviating the burden on staff. Secondly, using convolutional neural networks and object detection algorithms, it can precisely locate the location and type of pest and disease symptoms, providing a scientific basis for urban greening management. Based on the detection results, managers can formulate more precise management strategies and improve greening management. Furthermore, timely detection and treatment of pest and disease problems helps maintain ecological balance and promote sustainable urban development. Adjusting irrigation strategies and increasing monitoring frequency can effectively reduce water waste and improve resource utilization efficiency. Finally, through an event-driven mechanism, the detection of abnormal environmental data can automatically trigger appropriate emergency response measures. This facilitates rapid response to emergencies, minimizes losses, and ensures the safety and stability of urban greening.

[0093] In specific implementation, the hardware working environment of the image recognition-based machine learning algorithm in actual work includes: UAV inspection system: Model: DJI Mavic 2 Enterprise; Camera: 4K HD camera, 12 million pixels; Battery life: 31 minutes; Flight control: GPS positioning, automatic flight path planning; Central server: Processor: Intel Xeon Gold 6248R (2.4GHz, 24Cores); Memory: 128GB DDR4 RAM; Storage: 2TB NVMe SSD; Operating system: Ubuntu 20.04LTS; GPU: NVIDIA Tesla V100 (32GB); Communication equipment: Wireless communication module: 4G / 5G wireless communication module; Data transmission protocol: MQTT protocol; Image processing software: Development environment: Python 3.8; Deep learning framework: TensorFlow 2.4; Image processing library: OpenCV 4.5;

[0094] Based on the above hardware, an image dataset containing symptoms of pests and diseases was prepared. The dataset contains 1,000 images, 500 of which are normal images and 500 are images containing symptoms of pests and diseases. The dataset is divided into a training set (70%) and a test set (30%). By comparing the performance of the machine learning algorithm based on image recognition (Group A) and the traditional algorithm (Group B) in pest and disease detection, the positive and beneficial effects of the machine learning algorithm based on image recognition are verified. Among them, Group A: Machine learning algorithm based on image recognition, using convolutional neural network (CNN) for feature extraction and target detection. Group B: Traditional algorithm, using rule-based methods (such as color segmentation, edge detection, etc.) for pest and disease detection. The experimental record table is shown in Table 2:

[0095] Table 2 Experimental record form

[0096] Number of experiments Group Detection time (seconds) Detection accuracy (%) False alarm rate (%) 1 Group A 12.3 95.6 2.1 1 Group B 18.5 82.4 12.3 2 Group A 12.1 95.8 2.0 2 Group B 18.7 82.2 12.5 3 Group A 12.4 95.5 2.2 3 Group B 18.6 82.3 12.4 4 Group A 12.2 95.7 2.1 4 Group B 18.8 82.1 12.6 5 Group A 12.3 95.6 2.1 5 Group B 18.5 82.4 12.3

[0097] As shown in Table 2, image recognition-based machine learning algorithms demonstrate higher detection accuracy, lower false alarm rates, and faster processing speeds in pest and disease detection, demonstrating significant positive benefits. This provides more efficient and accurate technical support for urban greening management.

[0098] In the specific implementation of S4 and S5, the construction of the growth prediction model first collects historical environmental data (such as temperature, humidity, light), plant growth data (such as growth rate, health status) and pest and disease symptom data through various sensors and devices. These data are cleaned, aligned and standardized to ensure the quality and consistency of the data. Data preprocessing is the basis of model training, which improves the availability of data by removing noise and filling missing values. The model includes an input layer, an embedding layer, a long short-term memory layer, an attention mechanism layer, a fully connected layer and an output layer: the input layer receives preprocessed data, which includes time series environmental parameters, plant growth status and pest and disease symptoms. The input layer converts the raw data into a format that the model can process. Data is usually input in the form of a two-dimensional array, where each row represents a time step and each column represents a feature. The input layer is also responsible for passing the data to the next layer for further processing.

[0099] The embedding layer converts the input data into a fixed-length vector representation for subsequent processing. The embedding layer captures the intrinsic characteristics of the data by learning a low-dimensional representation of the input data. The embedding layer captures the intrinsic characteristics of the data by learning a low-dimensional representation of the input data. The embedding layer is typically used to process discrete features (such as categorical variables) and convert them into continuous vector form. In this model, the embedding layer learns the optimal vector representation through the training process, so that similar input data are closer in the vector space;

[0100] Long Short-Term Memory (LSTM) is a special type of recurrent neural network (RNN) that can capture long-term dependencies in time series data and extract temporal features that are important for plant growth. By using multiple LSTM units, the model can effectively extract temporal features and understand the impact of environmental changes on plant growth. LSTM is a special type of recurrent neural network (RNN) that effectively handles long-term dependencies in time series data. LSTM uses gating mechanisms (input gate, forget gate, and output gate) to control the flow of information, thereby avoiding the vanishing and exploding gradient problems. At each time step, an LSTM unit receives the input data of the current time step and the hidden state of the previous time step. An LSTM unit includes: a forget gate, which determines which information is discarded. The forget gate uses a sigmoid activation function to calculate a value between 0 and 1, indicating the degree to which each cell state should be retained. An input gate, which determines which new information is stored in the cell state. The input gate uses a sigmoid activation function and a tanh activation function to calculate new candidate cell states. Cell state update, which combines the outputs of the forget and input gates to update the cell state. An output gate, which determines which information is output. The output gate calculates a value between 0 and 1 through a sigmoid activation function, indicating the output degree of each cell state. Finally, a tanh activation function is used to convert the cell state into an output value.

[0101] The attention mechanism layer improves the model's explanatory power and accuracy by dynamically adjusting the importance of different time steps. The attention mechanism layer assigns different weights to the features of different time steps based on the input of the current time step, thereby focusing on key information. The attention mechanism dynamically adjusts the importance of different time steps to improve the model's explanatory power and accuracy. The attention mechanism enables the model to focus on key information by calculating the weight of each time step. Specifically, it includes: Calculating the attention score: Calculating the attention score of each time step through a feedforward neural network (usually a fully connected layer). The attention score reflects the importance of the time step to the current prediction. Normalization: The attention score is normalized through the softmax function so that its sum is 1 to form the attention weight. Weighted summation: Multiply the feature vector of each time step by its corresponding attention weight, and then perform weighted summation to obtain the final feature representation.

[0102] The fully connected layer performs a nonlinear transformation on the extracted features to generate the final prediction result. Through multiple layers of neurons, the fully connected layer learns complex nonlinear relationships, improving the model's expressiveness. Through multiple layers of neurons, the fully connected layer transforms the features extracted by the LSTM layer and the attention mechanism layer into the final prediction result. A fully connected layer typically contains multiple hidden layers, each of which performs a nonlinear transformation using an activation function (such as ReLU) to increase the model's expressiveness. The final layer is typically a linear layer that outputs the prediction result.

[0103] The output layer outputs predicted plant growth under future environmental changes, including growth rate, health status, and pest and disease risk. The output layer converts the output of the fully connected layer into specific predictions. The output layer typically uses an appropriate activation function (such as sigmoid or softmax) for a final nonlinear transformation to generate the predictions. For example, a linear activation function might be used to predict growth rate, while a sigmoid activation function might be used to generate probability values ​​for pest and disease risk predictions.

[0104] S5 generates optimized management strategies in the following steps: Real-time data collection: Sensors collect real-time environmental data (such as temperature, humidity, and light) and plant growth data (such as growth rate and health status). This real-time data is transmitted to a central server via IoT technology to ensure real-time and accurate data.

[0105] Data input: The real-time collected data is input into the growth prediction model. The model generates growth prediction results for future environmental changes based on historical data and real-time data.

[0106] Irrigation Planning: An expert system generates an irrigation plan based on vegetation growth needs and soil moisture data. The expert system determines the timing, frequency, and volume of irrigation based on plant water requirements and soil moisture. Using an optimization algorithm, the irrigation plan ensures the optimal use of water resources and minimizes waste.

[0107] Fertilization Strategy: A nutritional model generates a fertilization plan based on vegetation growth status and soil nutrient data. The nutritional model determines the type, amount, and frequency of fertilizer application based on plant nutritional requirements and soil nutrient status. Fertilization strategies utilize optimization algorithms to ensure optimal fertilizer application, improve fertilizer utilization, and reduce environmental pollution. The nutritional model first integrates and cleans individual or group nutritional data through data collection and preprocessing techniques. This step utilizes a database management system and data cleaning algorithms to ensure the accuracy and consistency of input data, laying the foundation for subsequent analysis. Next, the model utilizes a personalized nutritional assessment algorithm to substantively analyze the nutritional needs of different individuals. Based on factors such as age, gender, weight, height, body fat percentage, health status, and dietary habits, the algorithm uses regression analysis and machine learning models (such as decision trees and random forests) to calculate nutritional recommendations and dietary plans tailored to individual needs. Furthermore, the model utilizes real-time monitoring and feedback mechanisms (such as smart devices monitoring dietary intake and physical condition) to dynamically adjust nutritional management, ensuring the timeliness and effectiveness of the plan. This mechanism relies on Internet of Things technology and mobile application development technology, enabling users to easily obtain personalized nutritional advice and continuously optimize nutritional plans based on feedback.

[0108] Pest and disease control measures: Based on pest and disease warning information and vegetation growth status, an expert system formulates control measures. Based on the severity of the pest and disease and the health of the plants, the expert system determines measures such as spraying pesticides and pruning diseased branches. Pest and disease control measures utilize optimization algorithms to ensure effectiveness and timeliness, minimizing the impact of pests and diseases on plants. The expert system first utilizes knowledge acquisition and representation technologies to extract and structure domain expert knowledge. This step leverages data mining and natural language processing techniques to extract valuable knowledge from extensive literature, case studies, and expert interviews, representing it in the form of ontologies and rule bases. Next, the system utilizes a reasoning engine to effectively apply this knowledge and solve problems. Based on forward reasoning and rule-matching algorithms, the reasoning engine retrieves relevant facts and rules from the knowledge base based on user-entered questions or requests, gradually deriving new conclusions or recommendations. Furthermore, the expert system significantly improves system usability and accuracy through user interaction and feedback mechanisms. This mechanism uses user interface design technology and machine learning algorithms to enable users to easily input information and view results. At the same time, the inference results are evaluated and adjusted based on user feedback to continuously optimize the performance of the system.

[0109] Green Area Layout: Based on pedestrian activity patterns and vegetation growth status, spatial planning methods are used to generate the layout of green areas and the types of vegetation. Spatial planning methods consider pedestrian density, activity areas, and plant growth patterns to optimize the layout of green areas, improve greening effectiveness, and enhance the aesthetics of the city. Spatial planning methods first utilize data collection and integration techniques to substantially complete the collection and processing of basic spatial planning data. This step utilizes geographic information systems (GIS) and remote sensing technology to acquire multi-source data on topography, natural resources, and socioeconomic factors. Data fusion algorithms and spatial database technology enable unified management and efficient access to this data. Next, planning methods utilize spatial analysis and evaluation techniques to substantively assess the potential for spatial development and the carrying capacity of resources and the environment. This step utilizes spatial overlay analysis, multi-criteria decision analysis (MCDA), and ecosystem service assessment models to comprehensively consider multiple factors, including resources and the environment, the economy and society, and policies and regulations, providing a scientific basis for spatial planning. Furthermore, planning methods utilize planning scheme design and optimization techniques to substantively develop planning schemes that meet regional development goals and constraints. This step utilizes intelligent optimization algorithms (such as genetic algorithms and particle swarm optimization) and planning simulation techniques to compare and optimize multiple planning options to ensure their feasibility and sustainability. Ultimately, through planning implementation and monitoring and evaluation techniques (such as planning implementation effectiveness evaluation and environmental impact assessment), the planning scheme is implemented and dynamically adjusted to ensure the achievement of planning objectives.

[0110] In specific implementations, the model, using LSTM and attention mechanisms, can capture long-term dependencies in time series data, improving the accuracy of growth predictions. This helps identify potential problems in advance, implement preventative measures, and mitigate losses. Compared to traditional time series prediction methods, growth prediction models based on deep learning offer greater accuracy and robustness, and are better able to cope with complex and changing environmental conditions. Furthermore, irrigation plans generated by an expert system can rationally schedule irrigation times and frequencies based on plant water requirements and soil moisture, reducing water waste and improving irrigation efficiency. Fertilization plans generated by a nutritional model can rationally schedule fertilizer types, amounts, and frequencies based on plant nutritional requirements and soil nutrient availability, improving fertilizer utilization and reducing environmental pollution. Pest control measures developed by the expert system can be tailored to the severity of pests and diseases and the health of plants, enabling timely implementation of measures such as spraying pesticides and pruning diseased branches, minimizing their impact on plants and protecting greening achievements. Greening area layouts and vegetation types generated by spatial planning methods can be optimized based on pedestrian activity patterns and plant growth habits, enhancing greening effectiveness and improving urban aesthetics. Secondly, the system can respond in real time to environmental changes and plant growth status, generating optimized management strategies to ensure timely and effective management. Leveraging expert systems and nutritional models, the system generates intelligent management strategies based on plant growth needs and environmental conditions, reducing human intervention and improving management efficiency. Furthermore, the system coordinates multiple management processes, including irrigation, fertilization, pest control, and greening area layout, achieving comprehensive optimization and improving management efficiency and effectiveness.

[0111] In S6, a deep Q-network (DQN) defines multiple optimization objectives, including plant growth status, pest and disease symptoms, water resource utilization efficiency, and environmental protection effectiveness. These objectives reflect the core requirements of urban greening management. Each optimization objective is assigned an initial weight, reflecting its importance in the overall optimization process. These initial weights can be determined through expert experience or historical data statistics. The DQN algorithm combines deep learning and reinforcement learning to solve complex decision-making problems. Its core concept is to use the Q-function to evaluate the expected reward of taking a certain action under a given state. In urban greening systems, the Q-function represents the expected effect of taking a management measure (such as irrigation, fertilization, or pest control) under the current environmental conditions. The state represents comprehensive information about the current environment, including plant growth status, pest and disease symptoms, soil moisture, and light intensity, and is typically represented as a vector. The action represents the possible management measures, such as increasing irrigation frequency, adjusting fertilization strategies, or initiating pest and disease control measures, and is also represented as a vector. The reward function evaluates the effect of an action and is designed based on the changing optimization objective, such as improved plant growth, reduced pests and diseases, and increased water resource efficiency. The experience replay mechanism improves learning efficiency by storing and reusing past experiences. At each time step, the current state, action, reward, and next state are stored in the experience replay buffer. The target network stabilizes the learning process, and its parameters are periodically copied from the current network to ensure the stability of the target Q value. The ε-greedy strategy balances exploration (trying new actions) and exploitation (selecting the current optimal action), ensuring that the algorithm gradually converges to the optimal policy.

[0112] In practice, step S6 works as follows: First, multiple optimization objectives are defined, such as plant growth status, pest and disease symptoms, water resource utilization efficiency, and environmental protection effectiveness. Initial weights are assigned to each objective. These weights can be determined through expert experience or historical data statistics to ensure that the importance of each objective in the overall optimization is appropriately reflected. Next, various sensors and devices collect real-time environmental data (such as temperature, humidity, and light), plant growth data (such as growth rate and health status), and pest and disease symptom data. This collected data is integrated into a state vector, representing comprehensive information about the current environment. Based on the current state, the DQN algorithm selects the optimal action. In the initial stage, an ε-greedy strategy is used to balance exploration and exploitation, ensuring that the algorithm gradually converges to the optimal strategy. Based on the selected action, corresponding management measures are implemented, such as increasing irrigation frequency, adjusting fertilization strategies, and initiating pest and disease control measures. A reward value is calculated based on the action taken and the change in the optimization objective. The reward value reflects the effect of the action, such as improved plant growth, reduced pests and diseases, and increased water resource utilization efficiency. The current state, action, reward, and next state are stored in an experience replay buffer, and training is performed through random sampling to improve learning efficiency. Backpropagation and optimization algorithms (such as Adam) are used to update the DQN network's parameters and optimize weights, enabling the model to better adapt to environmental changes. The parameters of the current network are regularly copied to the target network to ensure the stability of the target Q value. Through continuous iterative optimization, the model can dynamically adjust the weight of each optimization target, achieving multi-objective optimization.

[0113] Traditional urban greening management systems often rely on fixed rules and manual adjustments, lacking real-time and adaptability. For example, rule-based irrigation systems can only irrigate according to a preset schedule and are unable to adapt to real-time environmental changes. The DQN algorithm, however, collects environmental data in real time and dynamically adjusts management measures based on current environmental conditions, ensuring timely and effective management. Through continuous iterative optimization, the DQN algorithm adaptively adjusts the weight of each optimization objective, improving management flexibility and adaptability. Furthermore, traditional urban greening management systems typically focus on a single objective, such as plant growth or water resource efficiency, making it difficult to achieve multi-objective optimization. For example, a rule-based pest control system may only focus on reducing pests and diseases while ignoring the impact on plant growth and water resource efficiency. The DQN algorithm, on the other hand, can simultaneously consider multiple optimization objectives and achieve multi-objective optimization by dynamically adjusting the weight of each objective. This helps to comprehensively improve the effectiveness of urban greening management and ensure the coordinated development of all aspects. Furthermore, traditional urban greening management systems rely on expert experience and fixed rules, lacking learning and intelligent decision-making capabilities. For example, rule-based irrigation systems can only irrigate according to preset rules and are unable to intelligently adjust to actual conditions. However, the DQN algorithm, using deep learning technology, can learn optimal management strategies from extensive historical data, demonstrating strong learning and intelligent decision-making capabilities. Through continuous iterative optimization, the DQN algorithm can gradually converge to the optimal strategy, improving the level of intelligent management. Furthermore, traditional urban greening management systems suffer from significant resource waste, such as over-irrigation and over-fertilization. For example, rule-based irrigation systems may irrigate when not necessary, resulting in water waste. The DQN algorithm, however, can dynamically adjust management measures based on current environmental conditions and optimization objectives, improving resource utilization efficiency. Through precise irrigation and fertilization strategies, the DQN algorithm can reduce resource waste, improve resource utilization efficiency, and promote sustainable development. Finally, traditional urban greening management systems have shortcomings in environmental protection, such as the overuse of chemical pesticides. For example, rule-based pest control systems can overuse chemical pesticides, leading to environmental pollution. The DQN algorithm, however, comprehensively considers environmental protection outcomes and dynamically adjusts management measures to reduce chemical pesticide use and improve environmental protection. Through precise pest and disease control strategies, the DQN algorithm can reduce environmental pollution and protect the ecological environment.

[0114] The hardware environment for the Deep Q Network algorithm in a big data-based urban greening system primarily includes high-performance computing servers, data acquisition and processing equipment, network communication equipment, and intelligent execution devices. The high-performance computing servers are responsible for running the Deep Q Network algorithm, performing complex data processing and model training. Data acquisition and processing equipment, including various sensors and environmental monitoring stations, collects multi-dimensional data related to urban greening in real time. Network communication equipment ensures efficient data transmission between the server, data acquisition equipment, and intelligent execution devices. The intelligent execution devices then implement appropriate greening management measures, such as irrigation, fertilization, and pruning, based on the optimal strategy output by the algorithm.

[0115] Based on the above hardware, a comparative experiment was conducted using the Deep Q Network algorithm (Group A) and the traditional algorithm (Group B). Both groups of experiments were conducted under the same experimental conditions, including the same urban greening area, the same data collection frequency, and the same greening management implementation standards. The traditional algorithm used a rule-based algorithm. During the experiment, the plant growth status, pest control effect, water resource utilization efficiency, and environmental protection effect indicators were recorded in each group of experiments under five repeated runs. The data record table is shown in Table 3:

[0116] Table 3 Experimental data record table

[0117]

[0118]

[0119] Comparative experiments show that the Deep Q-Network algorithm (Group A) outperforms the traditional algorithm (Group B) in terms of plant growth, pest control, water resource utilization efficiency, and environmental protection. Group A achieved an average improvement of approximately 10% in plant growth, 10% in pest control, 12% in water resource utilization, and 8% in environmental protection. This demonstrates that the Deep Q-Network algorithm can more accurately adjust greening management strategies based on environmental changes, enabling more efficient and intelligent urban greening management.

[0120] In S7, mixed-integer linear programming (MILP) is a mathematical optimization technique used to solve linear optimization problems involving both continuous and discrete variables. In urban greening systems, MILP can be used to optimize decision variables such as irrigation duration, fertilizer type, quantity, and frequency, as well as plant location and species. By defining an objective function and constraints, MILP can find the optimal solution that satisfies all constraints. In multi-objective optimization, MILP can be used to maximize plant growth, minimize pest and disease risks, and improve water resource efficiency and environmental protection. By defining multiple objective functions, MILP can balance multiple objectives to find the optimal overall solution. Contextual awareness refers to the system's ability to automatically adjust its behavior and strategies based on the current environment and user needs. In urban greening systems, contextual awareness dynamically adjusts the input parameters of the optimization model by collecting real-time environmental data (such as temperature, humidity, and light), pest and disease data, and growth prediction data. Contextual awareness ensures that the input data of the optimization model is always up-to-date, improving the accuracy and timeliness of the optimization results. By monitoring environmental changes in real time, the system can promptly adjust management strategies to respond to emergencies. Genetic algorithms are global optimization algorithms based on natural selection and genetic mechanisms. By simulating the biological evolution process, genetic algorithms can gradually improve the quality of solutions through selection, crossover, and mutation operations. In multi-objective optimization, genetic algorithms can find multiple Pareto optimal solutions, providing decision makers with a variety of optimization options. In urban greening systems, genetic algorithms can be used to optimize irrigation strategies, fertilization strategies, and plant layouts. By defining a fitness function, genetic algorithms can evaluate the effectiveness of different management strategies and select the optimal one.

[0121] In actual applications, S7 first collects environmental data (such as temperature, humidity, and light), pest and disease data, and growth forecast data in real time through various sensors and devices. The collected data is preprocessed, including data cleaning, alignment, and standardization, to ensure data quality and consistency. Next, an optimization model is constructed to define multiple optimization objectives, such as plant growth status, pest and disease risk, water resource utilization efficiency, and environmental protection effect. Decision variables are defined, including irrigation duration, fertilizer type, fertilizer quantity, fertilizer frequency, and plant location and type. Constraints are defined based on actual conditions and restrictions, such as maximum irrigation water volume, maximum fertilizer amount, etc. A MILP solver (such as Gurobi or CPLEX) is used to solve the optimization model to find the optimal solution that meets all constraints. The input parameters of the optimization model are dynamically adjusted through the context awareness module to ensure the accuracy and timeliness of the optimization results. A genetic algorithm is used to globally optimize the optimization model, find multiple Pareto optimal solutions, and provide a variety of optimization schemes for decision makers to choose from. Based on the optimization results, an optimal comprehensive management strategy is generated, including irrigation duration, fertilizer type, amount, frequency, and plant location and type. The generated management strategy is stored in a database for subsequent management and decision-making. The generated management strategy is then applied to actual management, such as adjusting the irrigation system, fertilization system, and plant layout. By monitoring environmental changes and management effectiveness in real time and collecting feedback data, the management strategy can be further optimized.

[0122] In specific implementation, the model runs as follows: Initialize the parameters of the MILP model, including the objective function, decision variables, and constraints. Initialize the parameters of the genetic algorithm, including population size, crossover probability, mutation probability, and maximum number of iterations. Initialize the context-aware module and set parameters for data collection and processing. Next, sensors and devices collect real-time environmental data, pest and disease data, and growth forecast data. Preprocess the collected data, including data cleaning, alignment, and standardization, to ensure data quality and consistency. Input the preprocessed data into the MILP model as input parameters for the optimization model. The context-aware module dynamically adjusts the input parameters of the optimization model to ensure the accuracy and timeliness of the optimization results. The optimization model is solved using the MILP solver to find the optimal solution that satisfies all constraints. A genetic algorithm is used to globally optimize the optimization model, finding multiple Pareto optimal solutions and providing multiple optimization options for decision makers to choose from. Based on the optimization results, an optimal comprehensive management strategy is generated, including irrigation duration, fertilizer type, amount, frequency, and plant location and type. The generated management strategy is stored in a database for subsequent management and decision-making. Finally, the generated management strategies are applied to actual management, such as adjusting irrigation systems, fertilization systems, and plant layout. By monitoring environmental changes and management effectiveness in real time and collecting feedback data, management strategies can be further optimized.

[0123] In practice, traditional urban greening management systems typically focus on a single objective, such as plant growth or water resource efficiency, making it difficult to achieve multi-objective optimization. For example, rule-based pest control systems may focus solely on reducing pests and diseases while ignoring their impact on plant growth and water resource efficiency. Multi-objective optimization models, however, define multiple optimization objectives, such as plant growth, pest and disease risk, water resource efficiency, and environmental protection. These models can balance these objectives and find a comprehensive optimal solution, achieving multi-objective optimization. Furthermore, using context-aware technology, multi-objective optimization models can collect environmental data, pest and disease data, and growth forecast data in real time, dynamically adjusting the optimization model's input parameters to ensure the accuracy and timeliness of optimization results and enhance the real-time and adaptability of management. Furthermore, using genetic algorithms, multi-objective optimization models can perform global optimization, identify multiple Pareto optimal solutions, and provide decision makers with a variety of optimization options. This improves the quality of optimization results and ensures the scientific and rational nature of management strategies. Furthermore, by optimizing irrigation and fertilization strategies, multi-objective optimization models can reduce resource waste and improve resource efficiency. For example, by using precise irrigation and fertilization strategies, the model can reduce water and fertilizer waste, improve resource efficiency, and promote sustainable development. Furthermore, by comprehensively considering environmental protection effects, the multi-objective optimization model can reduce the use of chemical pesticides and improve environmental protection. For example, by using precise pest control strategies, the model can reduce environmental pollution and protect the ecological environment. Finally, using MILP and genetic algorithms, the multi-objective optimization model can learn optimal management strategies from large amounts of historical data, demonstrating strong learning and intelligent decision-making capabilities. Through continuous iterative optimization, the model can gradually converge to the optimal strategy, enhancing the level of intelligent management.

[0124] In practical applications, the hardware environment for scenario-based multi-objective optimization models in big data-based urban greening systems primarily includes: high-performance computing servers, big data storage and processing equipment, an IoT sensor network, an intelligent irrigation and fertilization system, and network communication equipment. High-performance computing servers are responsible for running complex optimization algorithms and data processing tasks; big data storage and processing equipment is used to store massive amounts of urban greening data and perform efficient data processing and analysis; the IoT sensor network is responsible for collecting real-time environmental data, plant growth data, and pest and disease data; the intelligent irrigation and fertilization system implements specific greening management measures based on the strategies output by the optimization model; and network communication equipment ensures data communication and command transmission between various devices.

[0125] Based on the above hardware environment, set up Group A: using a scenario-based multi-objective optimization model (MILP+situational awareness+genetic algorithm). Group B: using a traditional rule-based optimization algorithm. A comparative experiment was conducted using Group A and Group B respectively. Both groups of experiments were conducted under the same experimental conditions, including the same urban greening area, the same data collection frequency, and the same greening management measures implementation standards. During the experiment, the plant growth efficiency, resource consumption (such as irrigation water volume, fertilizer application amount), pest and disease control effect, and system response time of each group of experiments were recorded under five repeated runs. The experimental record data table is shown in Table 4:

[0126] Table 4 Comparative experiment record

[0127]

[0128] As shown in Table 4, the context-based multi-objective optimization model (Group A) outperformed the traditional algorithm (Group B) in terms of plant growth efficiency, resource consumption, pest and disease control effectiveness, and system response time. Group A's average plant growth efficiency increased by approximately 7%, resource consumption decreased by approximately 10%, pest and disease control effectiveness increased by approximately 7%, and system response time decreased by approximately 50%. This demonstrates that the context-based multi-objective optimization model can more accurately adjust greening management strategies based on real-time environmental conditions and plant growth conditions, achieving more efficient and intelligent urban greening management.

[0129] In S8, adaptive control algorithms dynamically adjust control parameters based on real-time system data. Their core concept is to enable the control system to adapt to environmental changes and achieve optimal control performance through continuous learning and adjustment during system operation. Adaptive control algorithms include model reference adaptive control (MRAC), self-tuning control (STC), and fuzzy adaptive control. MRAC defines an ideal reference model to ensure that the actual system output is as close as possible to the reference model output. It estimates system parameters online and continuously adjusts controller gains to gradually approximate the system output to the reference model. STC dynamically adjusts controller parameters by estimating system model parameters online, ensuring good control performance in a changing environment. Fuzzy adaptive control uses a fuzzy logic system to dynamically adjust control parameters based on the system's real-time status and historical data, enabling the system to adapt to complex and changing environments. The control system is responsible for dynamically adjusting the operating parameters of the irrigation system and fertilization equipment based on optimization strategies. A control system typically includes sensors, actuators, controllers, and communication modules. Sensors monitor the environment and plant status in real time, actuators execute control commands, controllers process data and generate control signals, and communication modules transmit data. In urban greening systems, the control system dynamically adjusts parameters such as the irrigation system's duration and the type, quantity, and frequency of fertilizers applied by fertilization equipment through adaptive control algorithms based on the output of a scenario-based multi-objective optimization model. This ensures optimal plant growth, pest and disease control, water resource utilization, and environmental protection.

[0130] In practice, step S8 works as follows: First, various sensors and devices collect real-time environmental data (such as temperature, humidity, and light), plant growth data (such as growth rate and health), and pest and disease data. The collected data undergoes preprocessing, including data cleaning, alignment, and standardization, to ensure data quality and consistency. Next, the preprocessed data is input into a scenario-based multi-objective optimization model to generate an optimal comprehensive management strategy, including irrigation duration, fertilizer type, amount, frequency, and plant location and type. The generated management strategy is stored in a database for use by the control system. Based on the optimization strategy, the control system dynamically generates control signals using an adaptive control algorithm. The adaptive control algorithm dynamically adjusts control parameters based on real-time data to ensure the accuracy and timeliness of the control signals. For example, if real-time data indicates low soil moisture, the adaptive control algorithm increases irrigation duration; if plant growth is poor, the adaptive control algorithm adjusts fertilizer type and amount. Based on the generated control signals, the actuators execute the corresponding control commands. For example, the automatic irrigation system adjusts irrigation duration based on the control signal, while the automatic fertilization system adjusts fertilizer type and amount based on the control signal. By real-time monitoring of environmental changes and management effects, feedback data can be collected to further optimize control strategies.

[0131] In specific implementations, control systems based on adaptive control algorithms can collect environmental data in real time and dynamically adjust the operating parameters of irrigation systems and fertilization equipment to ensure the accuracy and timeliness of control signals. This improves the system's real-time and adaptability, enabling it to better respond to environmental changes. Furthermore, control systems based on adaptive control algorithms can dynamically adjust control parameters based on real-time data, ensuring that the system maintains excellent control performance in a changing environment. This improves control effectiveness and ensures optimal achievement of goals such as plant growth, pest control, water resource utilization, and environmental protection. Furthermore, control systems based on adaptive control algorithms can dynamically adjust irrigation duration and fertilization amounts based on real-time data, reducing resource waste and improving resource utilization efficiency. For example, through precise irrigation and fertilization strategies, the system can reduce water and fertilizer waste, improve resource utilization efficiency, and promote sustainable development. Furthermore, control systems based on adaptive control algorithms can dynamically adjust pest control strategies based on real-time data, reducing the use of chemical pesticides and improving environmental protection. For example, through precise pest control strategies, the system can reduce environmental pollution and protect the ecological environment. Finally, control systems based on adaptive control algorithms can learn optimal control strategies from large amounts of historical data, demonstrating strong learning and intelligent decision-making capabilities. Through continuous iterative optimization, the system can gradually converge to the optimal strategy, improving the level of intelligent management.

[0132] Although specific embodiments of the present invention have been described above, those skilled in the art will appreciate that these specific embodiments are merely illustrative, and that those skilled in the art may omit, substitute, and modify the details of the methods and systems described above without departing from the principles and spirit of the present invention. For example, combining the above method steps to perform substantially the same functions in substantially the same manner to achieve substantially the same results falls within the scope of the present invention. Accordingly, the scope of the present invention is limited solely by the appended claims.

Claims

1. A method for optimizing urban greening based on big data, characterized by: The following steps are involved: S1. Obtain temperature and humidity, light intensity, soil moisture, and air quality data from environmental sensors through API interfaces and IoT wireless communication methods; obtain plant species and growth cycle data from the plant growth cycle database; obtain crowd density and activity area data from public data sources; and regularly obtain image data of green areas from the drone inspection system and store them in a temporary buffer; S2. Input the data into a distributed real-time data stream processing platform and clean, align, transform, and fuse the collected data. The distributed real-time data stream processing platform integrates a unified data set using a deep heterogeneous fusion algorithm, including unified environmental parameters, vegetation growth status, and pedestrian activity patterns. The deep heterogeneous fusion algorithm works by filling missing values ​​through mean interpolation, median interpolation, and K-nearest neighbor interpolation, and identifying and processing outliers using the Z-Score method; aligning timestamps through linear interpolation and spline interpolation; and aligning spatial coordinates through geocoding and coordinate transformation. Based on the aligned data, the deep heterogeneous fusion algorithm fuses the features of data from different sources through multimodal data; the multimodal data fusion extracts the features of image and time series data through convolutional neural networks and recurrent neural networks, and uses mutual information method, principal component analysis and feature engineering to generate feature vectors through dynamic weight adjustment mechanism. , the fused feature vector As the input of the deep neural network, the nonlinear mapping relationship between input data and output data is learned through a multilayer perceptron and an autoencoder to generate a data set in a unified format; the multilayer perceptron learns the nonlinear mapping relationship between input data and output data through forward propagation and backpropagation algorithms, and introduces nonlinear characteristics through the activation function ReLU to improve the expressiveness of the model; the autoencoder learns the low-dimensional representation of the data through the encoder and decoder, and optimizes the model parameters through the mean square error (MSE); S3, based on the drone inspection system's green area image data, identifies pest and disease symptoms through image recognition-based machine learning algorithms. When abnormal environmental data is detected, an event-driven mechanism automatically triggers corresponding emergency response measures, including increasing monitoring frequency and adjusting irrigation strategies. S4. Build a growth prediction model by training historical environmental data, plant growth data, and pest and disease symptoms through a long short-term memory network. S5. Inputting the real-time collected environmental data into the growth prediction model to obtain growth prediction results for future environmental changes, and generating optimization strategies including irrigation plans, fertilization strategies, pest and disease control measures, and green area layout based on the prediction results; S6. Define optimization objectives and initial weights, including plant growth status, pest and disease symptoms, water resource utilization efficiency, and environmental protection effects, and dynamically adjust the weight of each objective according to environmental changes through a deep Q-network algorithm; S7. Construct a context-based multi-objective optimization model using mixed integer linear programming, context-awareness, and genetic algorithms. Input real-time environmental data, pest and disease data, growth prediction data, and optimization strategies into the context-based multi-objective optimization model to obtain an optimal comprehensive management strategy, including irrigation duration, fertilizer type, fertilizer amount, fertilizer frequency, and plant location and type. The output results are stored in a database. The context-based multi-objective optimization model includes a data reading layer, a feature extraction layer, a context-awareness layer, a goal setting layer, an optimization modeling layer, and an algorithm solving layer. The working method of the context-based multi-objective optimization model is as follows: Step 801: Read pre-processed data from the central database through the data layer, including real-time environmental data, pest and disease data, and growth prediction data; Step 802: extract key features of the data through the feature extraction layer; the feature extraction layer extracts features through a principal component analysis method and stores the extracted feature data in a feature database; Step 803: The context perception layer reads the processed feature data from the feature database and identifies the current context state, including drought, heavy rain, high population density, high incidence of pests and diseases, and vigorous plant growth period. The context perception layer classifies the data into different categories through k-cluster analysis to identify the current context state. The k-cluster analysis selects the best clustering result through multiple initializations, and the formula expression is: （1） In formula (1), Indicates the cluster centers; represents the set of all data points; Represents a data point With the selected cluster center The Euclidean distance between Indicates the The variance of the cluster centers is used to adjust the influence of distance; Step 804: Optimization goals are set by the goal setting layer, including plant growth status, pest and disease symptoms, water resource utilization efficiency, and environmental protection effects. The goal setting layer uses an expert system combined with historical data to set initial weight values ​​for each optimization goal. The goal setting layer dynamically adjusts the weight values ​​based on actual environmental changes and management needs through a feedback mechanism based on reinforcement learning, continuously optimizing the weight settings. Step 805: Construct a mixed integer linear programming model through the optimization modeling layer, which includes the objectives of maximizing plant growth efficiency, minimizing the incidence of pests and diseases, and optimizing resource use costs. The mixed integer linear programming model dynamically adjusts the weights between different objectives based on real-time context using a dynamic adjustment function. The formula of the dynamic adjustment function is: （2） In formula (2), represents the comprehensive objective function; Indicates the Targets in time The weight of Indicates the The value of the objective function; Indicates the The penalty coefficient of each objective is used to adjust the nonlinear effect of the objective function; Indicates the The threshold of the target is used to define the benchmark value of the objective function; Indicates the The smoothing parameter of the objective is used to control the smoothness of the objective function; Step 806: Generate an optimal management strategy through the algorithm solution layer, including irrigation duration, fertilizer type, fertilizer amount, fertilizer frequency, and plant location and type. The algorithm solution layer solves the optimization model using a genetic algorithm. The genetic algorithm gradually optimizes individuals in the population through selection, crossover, and mutation operations, and resolves conflicts between objectives through non-dominated sorting and crowding distance calculation. The algorithm solution layer also continuously monitors the optimization effect through a feedback mechanism based on online learning and dynamically adjusts the optimization strategy according to actual conditions. S8. Sending the output results of the scenario-based multi-objective optimization model to a control system, and dynamically adjusting the operating parameters of the irrigation system and fertilization equipment and facilities through an adaptive control algorithm according to the optimization strategy.

2. The urban greening optimization method based on big data according to claim 1, characterized in that: The working principle of the image recognition-based machine learning algorithm is as follows: Step 301: Preprocess the image data collected by the drone through data cleaning and image enhancement. The data cleaning removes noise in the image through Gaussian filtering, and the image enhancement enhances the image details through histogram equalization and contrast adjustment. Step 302: extract features from the image using a convolutional neural network. The convolutional neural network performs a convolution operation on the image through a convolution layer to extract local features, and performs a maximum pooling operation on the feature map through a pooling layer to reduce computational complexity. The output of the convolution layer is nonlinearly transformed using an activation function. Step 303: Locate and identify pest symptoms in the extracted image features using a target detection algorithm; the target detection algorithm works as follows: r1. Generate candidate regions of pest and disease symptoms through sliding window and region proposal network. r2. Extract features from the candidate region through a convolutional neural network and generate a feature vector; r3. Classify the feature vector and perform bounding box regression through a fully connected layer and a cross entropy loss function to determine the location and category of the pest and disease symptoms; r4. Ensure the accuracy and reliability of the machine learning algorithm through data labeling and model training. The data labeling is performed by manually labeling the pest and disease symptoms in the image or semi-supervised learning to generate a training data set. The model training is performed by backpropagation algorithm and SGD optimization algorithm. r5. The identified pest and disease symptoms are transmitted to the central server in real time through an event-driven mechanism.

3. The urban greening optimization method based on big data according to claim 1, characterized in that: The working method of the event-driven mechanism is: read pest and disease symptom data from the message queue through the event processor, generate response measure events, and push the response measure events to the message queue, and trigger emergency response measures through the rule engine and decision tree algorithm, including increasing monitoring frequency, adjusting irrigation strategies, and initiating pest and disease control measures.

4. The urban greening optimization method based on big data according to claim 1, characterized in that: The growth prediction model includes an input layer, an embedding layer, a long short-term memory layer, an attention mechanism layer, a fully connected layer, and an output layer. The operation process of the growth prediction model is as follows: Step 501: receiving and processing input data through the input layer, including historical environmental data, plant growth data, and pest and disease symptoms; Step 502: Convert the input data into a high-dimensional vector representation through the embedding layer to capture data features; The embedding layer converts plant species and pest types into continuous vector representations through a word embedding method; Step 503: The long short-term memory layer receives the high-dimensional vector output by the embedding layer, captures the long-term dependency in the time series data, and extracts the time series features; the long short-term memory layer dynamically adjusts the retention and discarding of information through a gating mechanism; Step 504: Enhance the model's attention to important time steps through the attention mechanism layer; the attention mechanism uses the attention mechanism to calculate the weight of each time step, and the weighted sum is used as supplementary information output by the long short-term memory layer to enhance the model's sensitivity to key time steps; Step 505: The fully connected layer performs a nonlinear transformation on the extracted features to generate a prediction result; the fully connected layer performs a nonlinear transformation on the outputs of the long short-term memory layer and the attention mechanism layer through the ReLU activation function, and maps the features extracted by the long short-term memory layer and the attention mechanism layer to the output space; Step 506: Output the final growth prediction result through the output layer, including the plant growth status and pest and disease prediction under future environmental changes; the output layer generates a binary classification prediction result through the Sigmoid activation function and generates a multi-classification prediction result through the Softmax activation function.

5. The urban greening optimization method based on big data according to claim 1, characterized in that: The method for generating an optimization strategy for irrigation plans, fertilization strategies, pest and disease control measures, and green area layout based on the prediction results in S5 is as follows: based on vegetation growth requirements and soil moisture data, an irrigation plan is generated through an expert system, including irrigation time, frequency, and water volume; based on vegetation growth status and soil nutrient data, a fertilization plan is generated through a nutrition model, including fertilizer type, amount, and frequency; based on pest and disease warning information and vegetation growth status, control measures are formulated through an expert system, including spraying pesticides and pruning diseased branches; based on human activity patterns and vegetation growth status, a layout of green areas and vegetation types are generated through a spatial planning method.

6. The urban greening optimization method based on big data according to claim 1, characterized in that: The dynamic adjustment method of the deep Q network algorithm is: Step 701: define optimization objectives, including plant growth status, pest and disease symptoms, water resource utilization efficiency, and environmental protection effects, and set initial weight values ​​for each optimization objective; Step 702: Construct a deep neural network through a convolutional neural network and a fully connected layer; It includes an online network and a target network. The online network is used to update and calculate the Q value of the current strategy in real time. The weights of the target network are regularly copied from the online network to calculate the target Q value to stabilize the training process. The convolutional neural network extracts the features of the input data through convolutional layers and pooling layers to generate feature vectors. The fully connected layer maps the feature vectors to Q values, i.e., the expected reward value of each action, through a multi-layer perceptron. Step 703: Collect environmental data in real time through the sensor network and the drone inspection system, including temperature, humidity, light intensity, and soil moisture. Preprocess the collected environmental data through normalization and standardization methods to construct an initialized environmental state space. Step 704: Combine the collected environmental data and the state information of the optimization target into a state vector as the input of the deep Q network; the state vector includes the current environmental data and the state information of the optimization target; Step 705: Select an action using the adaptive decay mechanism of the greedy strategy. The greedy strategy randomly selects an action or selects the action with the highest Q value in the current state based on the exploration rate. The exploration rate is gradually reduced during training to balance exploration and exploitation. Step 706: Execute the selected action and collect environmental data and optimization target status information after the action is executed through the sensor network and drone inspection system. If the plant growth condition improves, a positive reward is given; if the pest and disease symptoms worsen, a negative reward is given. Step 707: Update the Q value using the Bellman equation. The Bellman equation calculates the new Q value using the discount factor γ and the immediate reward r. The mean square error loss function (MSE) is used to calculate the difference between the current Q value and the target Q value, and the network parameters are updated through the back-propagation mechanism. Step 708: Periodically copy the weights of the online network to the target network through weight repetition to maintain the stability of the target value and reduce fluctuations during training; Step 709: Continuously execute steps 704 to 708, interact with the environment, collect data, update network weights, and stop when the predetermined number of training rounds or convergence criteria are met.

7. The urban greening optimization method based on big data according to claim 1, characterized in that: The adaptive control algorithm uses a PID control algorithm and fuzzy control to achieve dynamic adjustment of the irrigation system and fertilization equipment, and uses an online learning algorithm to ensure that the system maintains optimal performance in a dynamic environment. The PID control algorithm dynamically adjusts control parameters through proportional control, integral control, and differential control. The fuzzy control converts complex nonlinear relationships into fuzzy rules through a fuzzy inference system and makes decisions based on the fuzzy rule base. The online learning algorithm updates control parameters in real time through a recursive least squares method.

8. An urban greening system based on big data, characterized by: An urban greening optimization method based on big data, as applied to any one of claims 1 to 7, comprising: Multi-source data acquisition module, used to collect multimodal data of urban green areas, including temperature and humidity, light intensity, soil moisture, air quality, plant species, growth cycle, vegetation growth status and pedestrian density data; A data integration module is used to clean, align, convert and fuse the multimodal data collected by the multi-source data collection module through a deep heterogeneous fusion algorithm; The real-time data analysis module is used to analyze the integrated data in real time. The analysis content includes vegetation growth trend prediction, pest and disease warning, and human activity pattern recognition. The module also outputs real-time analysis results, including irrigation demand prediction, fertilization recommendations, pest and disease warning information, and human activity hotspots. An adaptive decision-making module, which uses mixed-integer linear programming, situational awareness, and genetic algorithms to build a context-based multi-objective optimization model based on real-time analysis results. This module dynamically adjusts irrigation plans, fertilization strategies, pest and disease control measures, and greening area layout, enabling data-driven intelligent management. User interaction and feedback module, which is used to provide user interaction functions, including user query, feedback, and activity participation, while displaying real-time data and results of greening management; The output end of the multi-source data acquisition module is connected to the input end of the data integration module; the output end of the data integration module is connected to the input end of the real-time data analysis module; the output end of the real-time data analysis module is connected to the input end of the adaptive decision module; and the user interaction feedback module is connected to the above modules.

Images