A method and device for monitoring vegetation growth environment

Abstract

This invention discloses a method and apparatus for monitoring vegetation growth environment, relating to the technical field of environmental monitoring. It acquires real-time environmental monitoring data of a target area through multiple sensor nodes, obtains a full-area image of the target area, determines vegetation characteristics based on the full-area image, and predicts environmental parameters based on vegetation characteristics and historical environmental parameters. Based on real-time and predicted data, it identifies abnormal nodes; acquires high-resolution local images of the target area for pest assessment, obtaining a first type of alarm information; if no abnormal nodes are found, it determines whether the vegetation growth environment is suitable based on real-time data, obtaining a second type of alarm information. By deploying sensor nodes to monitor environmental data in real time, utilizing image processing and other technologies to quantify vegetation characteristics, and combining predictive models and anomaly detection, it can provide data-driven scientific management solutions, thereby reducing reliance on management personnel experience and improving management efficiency and accuracy.

Description

Technical Field

[0001] This invention relates to the technical field of data analysis, and specifically to a method and device for monitoring vegetation growth environment. Background Technology

[0002] With the acceleration of urbanization, people have increasingly higher requirements for the quality of their living environment, and urban greening has become an important means to enhance the city's image and improve residents' quality of life. As an important component of urban greening, the management and maintenance of parks are receiving increasing attention.

[0003] In the park management industry, vegetation growth monitoring relies excessively on the experience and judgment of management personnel, lacking quantifiable data for scientific management. During vegetation management, different plants have varying requirements for commonly used parameters such as water, nutrients, and pH levels depending on the season, region, and soil type. Conventional management often relies on the historical experience of maintenance personnel, which places high demands on staff training for park management companies. Furthermore, frequent staff turnover makes it difficult to ensure the quality of management. This experience-based approach can lead to improper vegetation management, such as unnecessary irrigation, fertilization, or pruning, wasting resources and even damaging vegetation health. Furthermore, manual assessment of vegetation growth status and environmental needs is inefficient and inaccurate. Summary of the Invention

[0004] The purpose of this invention is to solve the problem of low accuracy mentioned in the background art, and to propose a method and device for monitoring vegetation growth environment.

[0005] A first aspect of this invention provides a method for monitoring vegetation growth environment, the method comprising:

[0006] The park is divided into multiple sub-regions based on soil type and vegetation type; multiple sensor nodes are deployed in each sub-region.

[0007] Environmental monitoring data of the target sub-region is acquired, and the environmental monitoring data within a first preset time period is used as the baseline data, while the real-time monitoring values ​​of multiple environmental parameters at the first target time are used as the real-time data; the target sub-region is any one of multiple sub-regions;

[0008] Obtain a global image of the target sub-region, determine vegetation features based on the global image, and obtain predicted data based on the vegetation features and the baseline data.

[0009] Based on the real-time data and the predicted data, abnormal nodes are identified;

[0010] If abnormal nodes exist, multiple local images of the target sub-region are acquired, and pest judgment is made based on the multiple local images to obtain a first judgment result. Based on the first judgment result, a first type of alarm information is generated and issued.

[0011] If there are no abnormal nodes, the vegetation growth environment of the target sub-region is judged to be qualified based on the real-time data, and a second judgment result is obtained. A second type of alarm information is generated and issued based on the second judgment result, and the sprinkler irrigation equipment is controlled based on the second type of alarm information.

[0012] Optionally, the global image includes multispectral images and infrared thermal imaging; the vegetation features include normalized difference vegetation index, leaf area index, and leaf surface temperature; the baseline data includes soil moisture, soil temperature, soil nutrients, photosynthetically active radiation, air temperature, air humidity, wind speed, and precipitation; the prediction data includes predicted values ​​for soil moisture, soil temperature, and soil nutrients.

[0013] The step of determining vegetation features based on the global image and obtaining predicted data based on the vegetation features and the baseline data includes:

[0014] Based on the baseline data, the monitored values ​​of multiple environmental parameters at the second target time are combined into a target column vector; the first preset time period includes multiple monitoring times; the second target time is any one of the multiple monitoring times;

[0015] Based on the multispectral image, the normalized vegetation index is calculated, and the leaf area index is calculated based on the normalized vegetation index.

[0016] The surface temperature of the blade is obtained based on the infrared thermal imaging at the second target time.

[0017] The leaf surface temperature, normalized vegetation index, leaf area index and the target column vector are concatenated to obtain the feature column vector;

[0018] The feature column vectors of multiple monitoring times within the first preset time period are concatenated to obtain a feature matrix;

[0019] The feature matrix is ​​used as input to a pre-trained prediction model to obtain prediction data for the first target time.

[0020] Optionally, determining the abnormal nodes based on the real-time data and the predicted data includes:

[0021] The offset value of the target parameter is calculated based on the predicted value and real-time monitoring value of the target parameter at the first target time; the target parameter is any one of soil moisture, soil temperature, and soil nutrients.

[0022] The desired offset vector is obtained based on the offset values ​​at different times within the second preset time period;

[0023] Calculate the standard deviation of the offset based on the desired offset vector;

[0024] If the offset standard deviation is greater than the first preset threshold, it is determined that the sensor node corresponding to the target parameter is abnormal.

[0025] Optionally, multiple local images include high-resolution images of vegetation at each node; the first type of alarm information includes pest alarm information and fault alarm information;

[0026] The step of determining pest infestation based on multiple local images, obtaining a first determination result, and generating and issuing a first type of alarm information based on the first determination result includes:

[0027] Each high-resolution image is used as input to a pre-trained pest detection model to obtain the first judgment result;

[0028] If the first judgment result indicates that there is pest infestation, then a pest infestation alarm message is generated;

[0029] If the first judgment result is no pests, then a fault alarm message is generated.

[0030] Optionally, the sprinkler irrigation equipment includes a spraying mode and a drip irrigation mode; the step of determining whether the vegetation growth environment of the target sub-area is qualified based on the real-time data, obtaining a second determination result, generating and issuing a second type of alarm information based on the second determination result, and controlling the sprinkler irrigation equipment based on the second type of alarm information includes:

[0031] Based on the vegetation type of the target sub-region, determine the threshold range of each environmental parameter;

[0032] If the real-time monitoring value of the target environmental parameter exceeds the corresponding threshold range, the target environmental parameter is determined to be unqualified, and the determination result is included in the second type of alarm information; the target environmental parameter is any one of multiple environmental parameters;

[0033] If the second type of alarm information contains information about unqualified soil moisture, then the spraying mode of the sprinkler irrigation equipment is turned on;

[0034] If the second type of alarm information contains information about unqualified soil temperature, then the drip irrigation mode of the sprinkler irrigation equipment shall be turned on.

[0035] A second aspect of the present invention provides a vegetation growth environment monitoring device, the device comprising:

[0036] The zoning module is used to divide the park into multiple sub-regions based on the soil quality and vegetation type; each sub-region is equipped with multiple sensor nodes.

[0037] The data acquisition module is used to acquire environmental monitoring data of the target sub-region, using environmental monitoring data within a first preset time period as baseline data, and real-time monitoring values ​​of multiple environmental parameters at a first target time as real-time data; the target sub-region is any one of multiple sub-regions;

[0038] The prediction module is used to acquire a global image of the target sub-region, determine vegetation features based on the global image, and obtain prediction data based on the vegetation features and the baseline data.

[0039] An anomaly detection module is used to determine abnormal nodes based on the real-time data and the predicted data;

[0040] The first alarm module is used to acquire multiple local images of the target sub-region if there are abnormal nodes, and to make pest judgment based on the multiple local images to obtain a first judgment result, and to generate and issue a first type of alarm information based on the first judgment result.

[0041] The second alarm module is used to determine whether the vegetation growth environment of the target sub-region is qualified based on the real-time data if there are no abnormal nodes, obtain a second judgment result, generate and issue a second type of alarm information based on the second judgment result, and control the sprinkler irrigation equipment based on the second type of alarm information.

[0042] Optionally, the global image includes multispectral images and infrared thermal imaging; the vegetation features include normalized difference vegetation index, leaf area index, and leaf surface temperature; the baseline data includes soil moisture, soil temperature, soil nutrients, photosynthetically active radiation, air temperature, air humidity, wind speed, and precipitation; the prediction data includes predicted values ​​for soil moisture, soil temperature, and soil nutrients.

[0043] The prediction module includes:

[0044] The first time step component module is used to form a target column vector from the monitoring values ​​of multiple environmental parameters at the second target time based on the reference data; the first preset time period includes multiple monitoring times; the second target time is any one of the multiple monitoring times;

[0045] The first vegetation feature calculation module is used to calculate the normalized vegetation index based on the multispectral image, and to calculate the leaf area index based on the normalized vegetation index.

[0046] The second vegetation feature calculation module obtains the leaf surface temperature based on the infrared thermal imaging of the second target time.

[0047] The second time step component module is used to concatenate the leaf surface temperature, normalized vegetation index, leaf area index and the target column vector to obtain the feature column vector.

[0048] The feature determination module is used to concatenate the feature column vectors of multiple monitoring times within the first preset time period to obtain a feature matrix;

[0049] The model prediction module is used to take the feature matrix as input to a pre-trained prediction model to obtain prediction data for the first target time.

[0050] Optionally, the anomaly detection module includes:

[0051] The offset calculation module is used to calculate the offset value of the target parameter based on the predicted value and real-time monitoring value of the target parameter at the first target time; the target parameter is any one of soil moisture, soil temperature, and soil nutrients;

[0052] The multi-time-step offset value statistics module is used to obtain the desired offset vector based on the offset values ​​at different times within the second preset time period.

[0053] The standard deviation calculation module is used to calculate the offset standard deviation based on the expected offset vector;

[0054] An abnormal node determination module is used to determine that the sensor node corresponding to the target parameter is abnormal if the offset standard deviation is greater than a first preset threshold.

[0055] Optionally, multiple local images include high-resolution images of vegetation at each node; the first type of alarm information includes pest alarm information and fault alarm information;

[0056] The first alarm module includes:

[0057] The pest detection module is used to take each high-resolution image as input to the pre-trained pest detection model and obtain the first judgment result.

[0058] The pest alarm module is used to generate pest alarm information if the first judgment result is that there are pests.

[0059] The fault alarm module is used to generate fault alarm information if the first judgment result is no pests.

[0060] Optionally, the sprinkler irrigation device includes a spray mode and a drip irrigation mode; the second alarm module includes:

[0061] The threshold determination module is used to determine the threshold range of each environmental parameter based on the vegetation type of the target sub-region.

[0062] The parameter alarm module is used to determine that the target environmental parameter is unqualified if the real-time monitoring value of the target environmental parameter exceeds the corresponding threshold range, and to include the determination result in the second type of alarm information; the target environmental parameter is any one of multiple environmental parameters;

[0063] A humidity control module is used to activate the spraying mode of the sprinkler irrigation equipment if the second type of alarm information contains information about unqualified soil moisture.

[0064] The temperature control module is used to activate the drip irrigation mode of the sprinkler irrigation equipment if the second type of alarm information contains information about unqualified soil temperature.

[0065] The beneficial effects of this invention are:

[0066] This invention proposes a method for monitoring vegetation growth environment. The method includes: dividing the park into multiple sub-regions based on soil quality and vegetation type; deploying multiple sensor nodes in each sub-region; acquiring environmental monitoring data of the target sub-region, using environmental monitoring data within a first preset time period as baseline data, and using real-time monitoring values ​​of multiple environmental parameters at a first target time as real-time data, wherein the target sub-region is any one of the multiple sub-regions; acquiring a full-area image of the target sub-region, determining vegetation characteristics based on the full-area image, and obtaining predicted data based on the vegetation characteristics and baseline data; identifying abnormal nodes based on real-time data and predicted data; if abnormal nodes exist, acquiring multiple local images of the target sub-region, and judging pests based on the multiple local images to obtain a first judgment result, generating and issuing a first type of alarm information based on the first judgment result; if there are no abnormal nodes, judging whether the vegetation growth environment of the target sub-region is qualified based on real-time data to obtain a second judgment result, generating and issuing a second type of alarm information based on the second judgment result.

[0067] By deploying sensor nodes to monitor environmental data in real time, using technologies such as image processing to quantify vegetation characteristics, and combining predictive models and anomaly detection, data-driven scientific management solutions can be provided, thereby reducing reliance on the experience of management personnel and improving management efficiency and accuracy. Attached Figure Description

[0068] The invention will now be further described with reference to the accompanying drawings.

[0069] Figure 1 A flowchart of a vegetation growth environment monitoring method is provided as an embodiment of the present invention;

[0070] Figure 2 The present invention provides a structural diagram of a vegetation growth environment monitoring device. Detailed Implementation

[0071] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0072] This invention provides a method for monitoring vegetation growth environment. See also... Figure 1 , Figure 1 A flowchart illustrating a method for monitoring vegetation growth environment according to an embodiment of the present invention. The method includes the following steps:

[0073] S101 is divided into multiple sub-regions based on the soil type and vegetation type of the park.

[0074] S102, acquire environmental monitoring data of the target sub-region, use the environmental monitoring data within the first preset time period as the baseline data, and use the real-time monitoring values ​​of multiple environmental parameters at the first target time as the real-time data.

[0075] S103: Obtain the global image of the target sub-region, determine the vegetation features based on the global image, and obtain the prediction data based on the vegetation features and the baseline data.

[0076] S104. Identify abnormal nodes based on real-time and predicted data.

[0077] S105, if there are abnormal nodes, acquire multiple local images of the target sub-region, and make pest judgment based on the multiple local images to obtain a first judgment result. Generate and issue a first type of alarm information based on the first judgment result.

[0078] S106 If there are no abnormal nodes, the vegetation growth environment of the target sub-region is judged to be qualified based on real-time data, a second judgment result is obtained, a second type of alarm information is generated and issued based on the second judgment result, and the sprinkler equipment is controlled based on the second type of alarm information.

[0079] Each sub-region is equipped with multiple sensor nodes; the target sub-region is any one of the multiple sub-regions.

[0080] Based on the vegetation growth environment monitoring method provided in this embodiment of the invention, by deploying sensor nodes to monitor environmental data in real time, using technologies such as image processing to quantify vegetation characteristics, and combining predictive models and anomaly detection, a data-driven scientific management plan can be provided, thereby reducing reliance on the experience of management personnel, improving management efficiency and accuracy, and ensuring high-quality vegetation management.

[0081] In one implementation, by comparing real-time data and predicted data, abnormal nodes can be quickly identified—areas where sensors have malfunctioned or where environmental conditions have changed abnormally. This anomaly detection mechanism helps to take timely remedial measures, prevent problems from escalating, and ensure the healthy growth of vegetation.

[0082] In one implementation, the first preset time period can be 6 hours before the first target time.

[0083] In one implementation, this method can reduce the park management company's reliance on personnel experience, reduce personnel training costs, and enable new employees to carry out effective maintenance based on system data, thus ensuring the stability of maintenance quality.

[0084] In one embodiment, the global image includes multispectral images and infrared thermal imaging; vegetation features include normalized vegetation index, leaf area index, and leaf surface temperature; baseline data include soil moisture, soil temperature, soil nutrients, photosynthetically active radiation, air temperature, air humidity, wind speed, and precipitation; and prediction data include predicted values ​​for soil moisture, soil temperature, and soil nutrients.

[0085] Step S103 includes:

[0086] Step 1: Based on the baseline data, construct a target column vector from the monitoring values ​​of multiple environmental parameters at the second target time.

[0087] Step 2: Calculate the normalized vegetation index (NDI) based on the multispectral images, and then calculate the leaf area index based on the NDI.

[0088] Step 3: Obtain the blade surface temperature based on the infrared thermal imaging at the second target time.

[0089] Step four: Concatenate the leaf surface temperature, normalized vegetation index, leaf area index and target column vector to obtain the feature column vector.

[0090] Step 5: Concatenate the feature column vectors of multiple monitoring times within the preset time period to obtain the feature matrix.

[0091] Step six: Use the feature matrix as input to the pre-trained prediction model to obtain the prediction data for the first target time.

[0092] The first preset time period includes multiple monitoring times; the second target time is any one of the multiple monitoring times.

[0093] In one implementation, combining multispectral imaging and infrared thermal imaging can yield more comprehensive vegetation information. The multispectral images can be acquired using drones, while the infrared thermal images can be obtained using an infrared thermal imager fixed at a high location.

[0094] In one implementation, soil parameters are predicted using data that are highly correlated with soil temperature, soil moisture, and soil nutrients, which can improve the accuracy of the prediction.

[0095] Parameter 1. Photosynthetically Active Radiation (PAR): PAR affects plant photosynthesis and transpiration. Higher PAR increases transpiration, leading to reduced soil moisture. PAR directly affects soil surface temperature. Stronger light radiation increases soil surface temperature. PAR indirectly influences plant nutrient requirements and absorption by affecting photosynthesis and growth.

[0096] Parameter 2. Air Temperature: Air temperature affects the rate of soil moisture evaporation. Higher air temperatures accelerate soil moisture evaporation, thus reducing soil moisture content. Air temperature directly affects soil surface temperature. Higher air temperatures generally lead to increased soil temperature. Air temperature influences soil microbial activity, which in turn affects the decomposition and release of soil nutrients. Higher temperatures generally promote microbial activity and facilitate nutrient release.

[0097] Parameter 3. Air humidity: Air humidity affects the rate of water evaporation. Higher air humidity can slow down the evaporation of soil moisture, thereby increasing soil moisture.

[0098] Parameter 4. Wind speed: Wind speed affects the rate of water evaporation. Higher wind speeds directly accelerate soil moisture evaporation, thus reducing soil humidity. Wind speed also affects soil temperature by promoting the evaporation of surface moisture. Higher wind speeds can carry away more heat, affecting soil temperature. Wind speed also affects the transpiration rate of plants, indirectly accelerating the absorption of water by plant roots, thereby reducing soil humidity.

[0099] Parameter 5. Precipitation: Precipitation directly increases soil moisture, thereby increasing soil humidity. Precipitation has a relatively small direct impact on soil temperature, but it can indirectly affect temperature by influencing soil thermal conductivity through wetting the soil surface. Precipitation affects soil nutrient cycling. Large amounts of precipitation can wash away and dilute nutrients in the soil, but they can also increase soil nutrient availability by promoting plant growth and nutrient dissolution.

[0100] Parameter 6. Normalized Difference Vegetation Index (NDVI): NDVI reflects vegetation density and health. Plant transpiration and root water uptake directly affect soil moisture maintenance. Areas with high NDVI usually have more vegetation cover, which helps reduce solar radiation to the soil surface and lower soil temperature. High NDVI generally indicates good plant growth and a strong ability for plants to absorb soil nutrients.

[0101] Parameter 7. Leaf Area Index (LAI): LAI reflects the leaf area of ​​a plant. Vegetation with a large leaf area can reduce soil moisture evaporation, thus maintaining higher soil moisture. A higher LAI value is usually accompanied by denser vegetation cover, providing more shade and lowering soil temperature.

[0102] Parameter 8. Leaf Surface Temperature (LST): Leaf surface temperature reflects the intensity of transpiration. High leaf surface temperature is usually associated with higher transpiration, thus reducing soil moisture. Leaf surface temperature indirectly affects plant growth and nutrient absorption; excessively high temperatures may affect plant health, thereby impacting the absorption of soil nutrients.

[0103] In one implementation, prediction can be made using a Long Short-Term Memory (LSTM) network model. LSTM, relying on its unique gating mechanism, can effectively capture long-term dependencies in time-series data, providing more accurate predictions than traditional statistical models. This model includes an input layer, a first LSTM layer, a second LSTM layer, and three output layers; the predicted values ​​for soil moisture, soil temperature, and soil nutrients are obtained through independent output layers.

[0104] The training process of this model includes:

[0105] 1. Data Preparation

[0106] We collected environmental and vegetation monitoring data from the past two years and extracted the monitoring data when the vegetation was growing healthily as a sample dataset.

[0107] The dataset was divided into training, validation, and test sets in a 14:3:3 ratio.

[0108] Based on the preset time step (30 minutes), time series data is extracted from the sample data;

[0109] Divide the time series data into subsequences of fixed length (number of time steps), for example, each subsequence can contain 12 time steps;

[0110] Create a corresponding label (the value at the next time step) for each subsequence;

[0111] Normalize the eigenvalues, that is, scale the eigenvalues ​​to the range [0,1].

[0112] 2. Training process

[0113] The training set subsequences are input into the model to obtain the prediction results;

[0114] The loss value is calculated based on the prediction results and the true labels (using the root mean square error as the loss function);

[0115] The gradient is calculated using the backpropagation algorithm, and the model parameters are updated using the optimization algorithm (Adam optimizer).

[0116] Iterative training is conducted to complete multiple training cycles.

[0117] 3. Model Evaluation

[0118] After each training cycle, the model performance is evaluated using a validation set, and the model's hyperparameters, such as the number of hidden units and the learning rate, are adjusted based on the evaluation results.

[0119] After training, the final performance of the model is evaluated using the test set. The loss and evaluation metrics on the test set are calculated, such as accuracy and F1 score.

[0120] 4. Model optimization

[0121] The model's hyperparameters can be optimized using the Sparrow Search algorithm.

[0122] In one embodiment, step S104 includes:

[0123] Step 1: Calculate the offset value of the target parameter based on the predicted value and real-time monitoring value of the target parameter at the first target time.

[0124] Step 2: Obtain the desired offset vector based on the offset values ​​at different times within the second preset time period.

[0125] Step 3: Calculate the standard deviation of the offset based on the expected offset vector.

[0126] Step 4: If the standard deviation of the offset is greater than the first preset threshold, it is determined that the sensor node corresponding to the target parameter is abnormal.

[0127] The target parameter is any one of soil moisture, soil temperature, or soil nutrients.

[0128] In one implementation, the system can detect anomalies in real time by calculating the offset between real-time monitored values ​​and predicted values. For example, if the actual soil moisture value deviates from the predicted value by more than expected, it may indicate a sensor malfunction or abnormal environmental conditions. Accurately identifying anomalies in sensor nodes helps in the timely maintenance and calibration of sensors, preventing data quality degradation from affecting subsequent analysis and decision-making. Timely detection of environmental anomalies helps in the timely maintenance of vegetation, preventing the spread of disasters, such as insect infestations.

[0129] In one implementation, the second preset time period can be 5.5 hours before the first target time plus the first target time.

[0130] In one implementation, by accurately detecting and locating abnormal sensor nodes, resources can be effectively managed and allocated, reducing the need for manual inspection and improving the overall efficiency of the system.

[0131] In one embodiment, the multiple local images include high-resolution images of vegetation at each node; the first type of alarm information includes pest alarm information and fault alarm information;

[0132] Step S105 includes:

[0133] Step 1: Use each high-resolution image as input to the pre-trained pest detection model to obtain the first judgment result.

[0134] Step 2: If the first judgment result indicates the presence of pests, then generate a pest alarm message.

[0135] Step 3: If the first judgment result is no pests, then generate a fault alarm message.

[0136] In one implementation, high-resolution images can capture vegetation details more precisely, thereby improving the accuracy of pest detection. High-resolution images can more clearly show signs of pests, such as insect holes and eggs, enabling detection models to identify even minute anomalies. These high-resolution images can be acquired using drones.

[0137] In one implementation, managers can take different actions based on the type of alarm (pest or malfunction). For example, a pest alarm might require the application of pesticides or physical removal, while a malfunction alarm might require checking sensors or image acquisition equipment. By clearly defining the alarm type, managers can allocate resources more effectively, optimize processes, and improve overall response efficiency and problem-solving speed.

[0138] In one implementation, the pest detection model can be a YOLOv5 model, trained using collected images of vegetation when pests occur.

[0139] In one embodiment, step S106 includes:

[0140] Step 1: Determine the threshold range of each environmental parameter based on the vegetation type of the target sub-region.

[0141] Step 2: If the real-time monitoring value of the target environmental parameter exceeds the corresponding threshold range, the target environmental parameter is judged to be unqualified, and the judgment result is included in the second type of alarm information; the target environmental parameter is any one of multiple environmental parameters.

[0142] In one embodiment, the sprinkler irrigation device includes a spraying mode and a drip irrigation mode; step S106 includes:

[0143] Step 1: Determine the threshold range of each environmental parameter based on the vegetation type of the target sub-region.

[0144] Step 2: If the real-time monitoring value of the target environmental parameter exceeds the corresponding threshold range, the target environmental parameter is deemed unqualified, and this judgment result is included in the second type of alarm information.

[0145] Step 3: If the second type of alarm information contains information about unqualified soil moisture, then turn on the spraying mode of the sprinkler irrigation equipment.

[0146] Step 4: If the second type of alarm information contains information about unqualified soil temperature, then turn on the drip irrigation mode of the sprinkler irrigation equipment.

[0147] The target environmental parameter is any one of multiple environmental parameters.

[0148] In one implementation, different vegetation types have different requirements for environmental parameters, so setting threshold ranges based on specific vegetation types makes monitoring more accurate. For example, some plants are more sensitive to soil moisture, while others are more sensitive to temperature changes. This flexibility improves the adaptability and accuracy of monitoring.

[0149] In one implementation, the environmental parameters measured by the sensor are compared with the threshold parameters. If the parameters are not within the threshold range, the system issues an early warning and notifies the relevant maintenance personnel to carry out maintenance, thereby enhancing the timeliness and scientific nature of vegetation maintenance.

[0150] In one implementation, automated environmental control can be achieved through coordinated sprinkler irrigation equipment, ensuring the quality of vegetation maintenance. Spraying and drip irrigation modes can be activated simultaneously to achieve efficient cooling and humidification.

[0151] In one embodiment, the acquisition of sensor data includes the following steps:

[0152] 1. Sensor Connection: First, connect various types of sensors to the corresponding interfaces of the data acquisition instrument. These sensors are responsible for sensing and measuring the physical parameters in the environment or equipment.

[0153] 2. Data Acquisition: The sensor converts the detected physical quantities into electrical signals, and the data acquisition instrument converts these analog electrical signals into digital signals through its built-in analog-to-digital converter (ADC) circuit.

[0154] 3. Data Processing: The acquired digital data will undergo preliminary processing within the acquisition unit. This may include data filtering to remove noise, data compression to reduce storage space requirements, and other operations.

[0155] 4. Data storage: The processed data will be temporarily stored in the internal memory of the acquisition unit, such as flash memory or random access memory (RAM).

[0156] 5. Data transmission: According to the preset time interval or triggering conditions, the data acquisition device transmits the stored data to the host computer (such as a computer or server) or cloud platform through wired (such as Ethernet, USB, etc.) or wireless (such as Wi-Fi, Bluetooth, cellular network, etc.) communication methods.

[0157] 6. Remote control and configuration: In some applications, administrators can perform operations such as parameter configuration, start / stop acquisition, and change acquisition frequency of the data acquisition instrument through remote communication.

[0158] This invention provides a vegetation growth environment monitoring device. See also... Figure 2 , Figure 2 A structural diagram of a vegetation growth environment monitoring device provided in an embodiment of the present invention. The device includes:

[0159] The zoning module is used to divide the park into multiple sub-regions based on the soil quality and vegetation type; each sub-region is equipped with multiple sensor nodes.

[0160] The data acquisition module is used to acquire environmental monitoring data of the target sub-region, using the environmental monitoring data within the first preset time period as the baseline data, and the real-time monitoring values ​​of multiple environmental parameters at the first target time as the real-time data; the target sub-region is any one of multiple sub-regions;

[0161] The prediction module is used to acquire a global image of the target sub-region, determine vegetation features based on the global image, and obtain prediction data based on the vegetation features and baseline data.

[0162] The anomaly detection module is used to identify abnormal nodes based on real-time data and predicted data.

[0163] The first alarm module is used to acquire multiple local images of the target sub-region if there are abnormal nodes, and to make pest judgment based on the multiple local images to obtain a first judgment result, and to generate and issue a first type of alarm information based on the first judgment result.

[0164] The second alarm module is used to determine whether the vegetation growth environment of the target sub-area is qualified based on real-time data if there are no abnormal nodes, obtain a second judgment result, generate and issue a second type of alarm information based on the second judgment result, and control the sprinkler irrigation equipment based on the second type of alarm information.

[0165] Based on the vegetation growth environment monitoring device provided in this embodiment of the invention, by deploying sensor nodes to monitor environmental data in real time, using technologies such as image processing to quantify vegetation characteristics, and combining prediction models and anomaly detection, it can provide data-driven scientific management solutions, thereby reducing reliance on the experience of management personnel and improving management efficiency and accuracy.

[0166] In one embodiment, the global image includes multispectral images and infrared thermal imaging; vegetation features include normalized vegetation index, leaf area index, and leaf surface temperature; baseline data include soil moisture, soil temperature, soil nutrients, photosynthetically active radiation, air temperature, air humidity, wind speed, and precipitation; and prediction data include predicted values ​​for soil moisture, soil temperature, and soil nutrients.

[0167] The prediction module includes:

[0168] The first time step component module is used to form a target column vector from the monitoring values ​​of multiple environmental parameters at the second target time based on the baseline data; the first preset time period includes multiple monitoring times; the second target time is any one of the multiple monitoring times;

[0169] The first vegetation feature calculation module is used to calculate the normalized vegetation index based on the multispectral image, and to calculate the leaf area index based on the normalized vegetation index.

[0170] The second vegetation feature calculation module obtains the leaf surface temperature based on the infrared thermal imaging of the second target time.

[0171] The second time step component module is used to concatenate the leaf surface temperature, normalized vegetation index, leaf area index and target column vector to obtain the feature column vector.

[0172] The feature determination module is used to concatenate the feature column vectors of multiple monitoring times within a first preset time period to obtain a feature matrix;

[0173] The model prediction module is used to take the feature matrix as input to a pre-trained prediction model to obtain the prediction data for the first target time.

[0174] In one embodiment, the anomaly detection module includes:

[0175] The offset calculation module is used to calculate the offset value of the target parameter based on the predicted value and real-time monitoring value of the target parameter at the first target time; the target parameter is any one of soil moisture, soil temperature, and soil nutrients.

[0176] The multi-time-step offset value statistics module is used to obtain the desired offset vector based on the offset values ​​at different times within the second preset time period.

[0177] The standard deviation calculation module is used to calculate the standard deviation of the offset based on the expected offset vector.

[0178] The abnormal node determination module is used to determine that the sensor node corresponding to the target parameter is abnormal if the offset standard deviation is greater than a first preset threshold.

[0179] In one embodiment, the multiple local images include high-resolution images of vegetation at each node; the first type of alarm information includes pest alarm information and fault alarm information;

[0180] The first alarm module includes:

[0181] The pest detection module is used to take each high-resolution image as input to the pre-trained pest detection model and obtain the first judgment result.

[0182] The pest alarm module is used to generate pest alarm information if the first judgment result is that there are pests.

[0183] The fault alarm module is used to generate fault alarm information if the first judgment result is that there are no pests.

[0184] In one embodiment, the sprinkler irrigation device includes a spraying mode and a drip irrigation mode;

[0185] The second alarm module includes:

[0186] The threshold determination module is used to determine the threshold range of each environmental parameter based on the vegetation type of the target sub-region.

[0187] The parameter alarm module is used to determine that the target environmental parameter is unqualified if the real-time monitoring value of the target environmental parameter exceeds the corresponding threshold range, and to include the judgment result in the second type of alarm information; the target environmental parameter is any one of multiple environmental parameters;

[0188] The humidity control module is used to activate the spraying mode of the sprinkler irrigation equipment if the second type of alarm information contains information about unqualified soil moisture.

[0189] The temperature control module is used to activate the drip irrigation mode of the sprinkler irrigation equipment if the second type of alarm information contains information about unqualified soil temperature.

[0190] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for monitoring vegetation growth environment, characterized in that, The method includes: The park is divided into multiple sub-regions based on soil type and vegetation type; multiple sensor nodes are deployed in each sub-region. Environmental monitoring data of the target sub-region is acquired, and the environmental monitoring data within a first preset time period is used as the baseline data, while the real-time monitoring values ​​of multiple environmental parameters at the first target time are used as the real-time data; the target sub-region is any one of multiple sub-regions; Obtain a global image of the target sub-region, determine vegetation features based on the global image, and obtain predicted data based on the vegetation features and the baseline data. Based on the real-time data and the predicted data, abnormal nodes are identified; If abnormal nodes exist, multiple local images of the target sub-region are acquired, and pest judgment is made based on the multiple local images to obtain a first judgment result. Based on the first judgment result, a first type of alarm information is generated and issued. If there are no abnormal nodes, the vegetation growth environment of the target sub-region is judged to be qualified based on the real-time data, a second judgment result is obtained, a second type of alarm information is generated and issued based on the second judgment result, and the sprinkler equipment is controlled based on the second type of alarm information. The global image includes multispectral images and infrared thermal imaging; the vegetation features include normalized vegetation index, leaf area index, and leaf surface temperature; the baseline data includes soil moisture, soil temperature, soil nutrients, photosynthetically active radiation, air temperature, air humidity, wind speed, and precipitation; the predicted data includes predicted values ​​for soil moisture, soil temperature, and soil nutrients. The step of determining vegetation features based on the global image and obtaining predicted data based on the vegetation features and the baseline data includes: Based on the baseline data, the monitored values ​​of multiple environmental parameters at the second target time are combined into a target column vector; the first preset time period includes multiple monitoring times; the second target time is any one of the multiple monitoring times; Based on the multispectral image, the normalized vegetation index is calculated, and the leaf area index is calculated based on the normalized vegetation index. The surface temperature of the blade is obtained based on the infrared thermal imaging at the second target time. The leaf surface temperature, normalized vegetation index, leaf area index and the target column vector are concatenated to obtain the feature column vector; The feature column vectors of multiple monitoring times within the first preset time period are concatenated to obtain a feature matrix; The feature matrix is ​​used as input to a pre-trained prediction model to obtain prediction data for the first target time. The step of determining abnormal nodes based on the real-time data and the predicted data includes: The offset value of the target parameter is calculated based on the predicted value and real-time monitoring value of the target parameter at the first target time; the target parameter is any one of soil moisture, soil temperature, and soil nutrients. The desired offset vector is obtained based on the offset values ​​at different times within the second preset time period; Calculate the standard deviation of the offset based on the desired offset vector; If the offset standard deviation is greater than the first preset threshold, it is determined that the sensor node corresponding to the target parameter is abnormal; Multiple local images include high-resolution images of vegetation at each node; the first type of alarm information includes pest alarm information and fault alarm information; The step of determining pest infestation based on multiple local images, obtaining a first determination result, and generating and issuing a first type of alarm information based on the first determination result includes: Each high-resolution image is used as input to a pre-trained pest detection model to obtain the first judgment result; If the first judgment result indicates that there is pest infestation, then a pest infestation alarm message is generated; If the first judgment result is no pests, then a fault alarm message is generated; The sprinkler irrigation equipment includes a spray mode and a drip irrigation mode; The step of determining whether the vegetation growth environment of the target sub-region is qualified based on the real-time data, obtaining a second judgment result, generating and issuing a second type of alarm information based on the second judgment result, and controlling the sprinkler irrigation equipment based on the second type of alarm information includes: Based on the vegetation type of the target sub-region, determine the threshold range of each environmental parameter; If the real-time monitoring value of the target environmental parameter exceeds the corresponding threshold range, the target environmental parameter is determined to be unqualified, and the determination result is included in the second type of alarm information; the target environmental parameter is any one of multiple environmental parameters; If the second type of alarm information contains information about unqualified soil moisture, then the spraying mode of the sprinkler irrigation equipment is turned on; If the second type of alarm information contains information about unqualified soil temperature, then the drip irrigation mode of the sprinkler irrigation equipment shall be turned on.

2. A vegetation growth environment monitoring device, characterized in that, The device includes: The zoning module is used to divide the park into multiple sub-regions based on the soil quality and vegetation type; each sub-region is equipped with multiple sensor nodes. The data acquisition module is used to acquire environmental monitoring data of the target sub-region, using environmental monitoring data within a first preset time period as baseline data, and real-time monitoring values ​​of multiple environmental parameters at a first target time as real-time data; the target sub-region is any one of multiple sub-regions; The prediction module is used to acquire a global image of the target sub-region, determine vegetation features based on the global image, and obtain prediction data based on the vegetation features and the baseline data. An anomaly detection module is used to determine abnormal nodes based on the real-time data and the predicted data; The first alarm module is used to acquire multiple local images of the target sub-region if there are abnormal nodes, and to make pest judgment based on the multiple local images to obtain a first judgment result, and to generate and issue a first type of alarm information based on the first judgment result. The second alarm module is used to determine whether the vegetation growth environment of the target sub-region is qualified based on the real-time data if there are no abnormal nodes, obtain a second judgment result, generate and issue a second type of alarm information based on the second judgment result, and control the sprinkler irrigation equipment based on the second type of alarm information. The global image includes multispectral images and infrared thermal imaging; the vegetation features include normalized vegetation index, leaf area index, and leaf surface temperature; the baseline data includes soil moisture, soil temperature, soil nutrients, photosynthetically active radiation, air temperature, air humidity, wind speed, and precipitation; the predicted data includes predicted values ​​for soil moisture, soil temperature, and soil nutrients. The prediction module includes: The first time step component module is used to form a target column vector from the monitoring values ​​of multiple environmental parameters at the second target time based on the reference data; the first preset time period includes multiple monitoring times; the second target time is any one of the multiple monitoring times; The first vegetation feature calculation module is used to calculate the normalized vegetation index based on the multispectral image, and to calculate the leaf area index based on the normalized vegetation index. The second vegetation feature calculation module obtains the leaf surface temperature based on the infrared thermal imaging of the second target time. The second time step component module is used to concatenate the leaf surface temperature, normalized vegetation index, leaf area index and the target column vector to obtain the feature column vector. The feature determination module is used to concatenate the feature column vectors of multiple monitoring times within the first preset time period to obtain a feature matrix; The model prediction module is used to take the feature matrix as input to a pre-trained prediction model to obtain prediction data for the first target time. The anomaly detection module includes: The offset calculation module is used to calculate the offset value of the target parameter based on the predicted value and real-time monitoring value of the target parameter at the first target time; the target parameter is any one of soil moisture, soil temperature, and soil nutrients; The multi-time-step offset value statistics module is used to obtain the desired offset vector based on the offset values ​​at different times within the second preset time period. The standard deviation calculation module is used to calculate the offset standard deviation based on the expected offset vector; An abnormal node determination module is used to determine that the sensor node corresponding to the target parameter is abnormal if the offset standard deviation is greater than a first preset threshold. Multiple local images include high-resolution images of vegetation at each node; the first type of alarm information includes pest alarm information and fault alarm information; The first alarm module includes: The pest detection module is used to take each high-resolution image as input to the pre-trained pest detection model and obtain the first judgment result. The pest alarm module is used to generate pest alarm information if the first judgment result is that there are pests. The fault alarm module is used to generate fault alarm information if the first judgment result is no pests. The sprinkler irrigation equipment includes a spray mode and a drip irrigation mode; The second alarm module includes: The threshold determination module is used to determine the threshold range of each environmental parameter based on the vegetation type of the target sub-region. The parameter alarm module is used to determine that the target environmental parameter is unqualified if the real-time monitoring value of the target environmental parameter exceeds the corresponding threshold range, and to include the determination result in the second type of alarm information; the target environmental parameter is any one of multiple environmental parameters; A humidity control module is used to activate the spraying mode of the sprinkler irrigation equipment if the second type of alarm information contains information about unqualified soil moisture. The temperature control module is used to activate the drip irrigation mode of the sprinkler irrigation equipment if the second type of alarm information contains information about unqualified soil temperature.

Images