Soil remediation real-time monitoring method utilizing coupling of multispectrum of unmanned aerial vehicle and sensing of internet of things

Abstract

The invention relates to a real-time monitoring method for soil remediation by utilizing multi-spectrum of an unmanned aerial vehicle and sensing coupling of the Internet of Things, and belongs to the technical field of soil environment monitoring and remediation. The method comprises the following steps: constructing a space grid model of a monitoring area; a space-air-ground integrated monitoring network is arranged, image data are obtained through multi-spectral remote sensing of an unmanned aerial vehicle, and soil environment parameters are collected through a ground Internet of Things sensor; preprocessing and fusing the multi-source spatio-temporal data, and establishing a spatio-temporal matching model; constructing an inversion model of the soil heavy metal content, the organic matter content and the pollutant degradation degree based on the fusion data; and the repair efficiency is dynamically calculated and visualized, and real-time evaluation and early warning of the repair process are realized. According to the invention, space-air-ground data collaboration is realized, the real-time performance, accuracy and space coverage of monitoring are remarkably improved, the defects of high cost, low efficiency and limited data dimension of a traditional method are overcome, and whole-process and intelligent decision support is provided for soil remediation.

Description

Technical Field

[0001] This invention relates to the field of soil environmental monitoring and remediation technology, and in particular to a real-time monitoring method for soil remediation processes based on a combination of UAV multispectral remote sensing and Internet of Things sensing technology. This method is especially suitable for dynamic, efficient, and multi-parameter collaborative parameter and assessment in the process of contaminated site remediation. Background Technology

[0002] With rapid industrialization and urbanization, soil pollution has become increasingly prominent, especially pollutants such as heavy metals and organic matter, which pose a serious threat to the soil environment and human health. Soil remediation has become an important part of environmental governance. Currently, traditional soil monitoring methods mainly rely on manual sampling and laboratory analysis, which suffer from problems such as low sampling density, poor timeliness, high cost, and difficulty in achieving large-scale continuous monitoring, failing to meet the needs of real-time, dynamic, and multi-dimensional data in the remediation process.

[0003] In recent years, UAV multispectral remote sensing technology has been gradually applied to agricultural, ecological, and environmental monitoring due to its high efficiency, flexibility, and ability to cover large areas. It can retrieve some physicochemical parameters of soil through multi-band imaging. On the other hand, Internet of Things (IoT) technology, through the deployment of ground sensor networks, can achieve real-time acquisition and transmission of key parameters such as soil temperature, humidity, pH value, and conductivity. However, existing technologies mostly employ single technical means and lack a mechanism for effectively integrating aerial remote sensing and ground sensor data. This results in spatial and temporal fragmentation of monitoring data, making it difficult to comprehensively and accurately reflect the dynamic changes during the soil remediation process.

[0004] Therefore, there is an urgent need in this field for a real-time monitoring method for soil remediation that can integrate UAV multispectral remote sensing and IoT sensing technologies to achieve integrated air-ground and multi-source data collaboration, so as to improve monitoring efficiency and accuracy and provide scientific basis and decision support for the remediation process.

[0005] The information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a real-time monitoring method for soil remediation that integrates UAV multispectral remote sensing and IoT sensing technologies to achieve integrated air-ground and multi-source data collaboration.

[0007] To achieve the above objectives, the present invention provides the following solution:

[0008] A real-time monitoring method for soil remediation utilizing the coupling of UAV multispectral and IoT sensing, characterized by the following steps:

[0009] S1. Construct a spatial grid model of the monitoring area. Based on the topography, area, and pollution distribution characteristics of the contaminated site to be remediated, the monitoring area is divided into several regular or irregular spatial grid units. Each grid unit has a unique spatial coordinate identifier. The side length L of the spatial grid unit is determined jointly by the spatial resolution R of the UAV multispectral image and the effective sensing radius r of the IoT ground sensor. The calculation formula is as follows:

[0010]

[0011] Where k is an adjustment coefficient, ranging from 0.8 to 1.5, used to balance data density and computational efficiency;

[0012] S2, an integrated air-space-ground detection network is deployed for data acquisition:

[0013] S21. Spatial Monitoring: Using an unmanned aerial vehicle (UAV) platform equipped with a multispectral camera, periodic aerial photography is conducted on the monitoring area according to a preset flight path to acquire multispectral image data covering the entire monitoring area; the multispectral camera covers at least five bands: blue light, green light, red light, red edge, and near-infrared; the flight altitude H is determined based on the required ground resolution (GSD), and its calculation formula is as follows:

[0014]

[0015] Where f is the camera focal length and the pixel size is the camera pixel size;

[0016] S22. Ground monitoring: An Internet of Things (IoT) sensor node network is deployed at the nodes or center of the spatial grid unit; each sensor node integrates at least a soil temperature sensor, a soil moisture sensor, a pH sensor, and a conductivity sensor; the sensor node collects soil environmental parameter data at a set time interval Δt and transmits the data to an edge computing gateway or cloud data center via wireless transmission.

[0017] S3. Multi-source spatiotemporal data preprocessing and fusion:

[0018] S31. Perform radiometric calibration, atmospheric correction and geometric correction on the acquired multispectral image data, and calculate the normalized difference in vegetation index (NDVI), normalized difference in water content (NDWI) and soil-adjusted vegetation index (SAVI).

[0019] S32. Perform outlier removal, missing value imputation, and spatiotemporal alignment processing on the soil environmental parameter data collected by IoT sensor nodes.

[0020] S33. Establish a spatiotemporal matching model for air-space-ground data: resample the preprocessed multispectral image data to a resolution that matches the spatial grid cell, and associate the multispectral feature index within each grid cell with ground sensor data within the same spatiotemporal range to form an integrated air-space-ground fusion data unit.

[0021] S4. Inversion and Modeling of Key Parameters for Soil Remediation:

[0022] Based on the fused data units obtained in step S3, an inversion model for key soil parameters is established; the key parameters include soil heavy metal content, organic matter content, and pollutant degradation degree.

[0023] For the inversion of soil heavy metal content [Cd], a method combining multiple linear regression and machine learning was adopted. The inversion model is as follows:

[0024] [Cd] i =α0+α1·NDVI i +α2·NDWI i +a3·pH i +α4·EC i +ε i ;

[0025] Where [Cd]_i represents the predicted heavy metal content of the i-th grid cell, NDVI_i and NDWI_i are the vegetation index and moisture index of the grid cell, respectively, pH_i and EC_i are the measured soil pH and electrical conductivity of the grid cell, respectively, α0~α4 are the model regression coefficients, which are obtained by training with historical sample data, and ε_i is the random error term.

[0026] S5. Dynamic evaluation and visualization of the repair process:

[0027] Based on the spatiotemporal sequence data of key soil parameters obtained from step S4, the remediation efficiency η of each grid cell at different time points is calculated, and the calculation formula is as follows:

[0028]

[0029] Where η_{i,t} represents the repair efficiency of the i-th grid cell at time t, [C]{i,t} represents the target pollutant concentration of the grid cell at time t, and [C]{i,t0} represents the target pollutant concentration of the grid cell at the initial repair time t0;

[0030] Generate spatiotemporal distribution maps of remediation efficiency and trend maps of pollutant concentration changes, and display and provide early warnings in real time through a visualization platform.

[0031] Optionally, in step S1, the division of the spatial grid cells also considers the pollutant migration and diffusion model. For areas with drastic changes in pollutant concentration gradients, an adaptive densification network strategy is adopted, and the formula for calculating the grid side length L' is:

[0032]

[0033] Where β is the encryption strength coefficient, This represents the maximum pollutant concentration gradient modulus for this grid cell.

[0034] Optionally, in step S22, the deployment of the IoT sensor nodes adopts an optimized deployment strategy, the objective function of which is to minimize the total number of nodes N while satisfying full coverage monitoring. Its mathematical model is expressed as:

[0035]

[0036]

[0037] x i ∈{0,1};

[0038] Where M is the total number of potential deployment potentials, x_i is the decision variable, S(r) represents the coverage circle with potential i as the center and radius r, and A represents the entire monitoring area.

[0039] Optionally, in step S3, the spatiotemporal alignment processing employs a time series imputation and spatial interpolation method based on Kriging interpolation to interpolate the non-uniformly acquired ground sensor data onto a unified spatiotemporal grid. Specifically, for any sensor's missing parameter value P(s_i,t_j) at time t_j, spatial interpolation estimation is performed using the observations of neighboring sensors at time t_j.

[0040]

[0041] Where n is the number of adjacent sensors used for interpolation, and λ_k is the Kriging weight coefficient, which is determined by the variogram model.

[0042] Optionally, in step S4, the inversion model for the key soil parameters is further trained and predicted using a random forest regression algorithm, with multispectral feature indices and ground sensor data as input features, and laboratory-measured soil parameters as target values ​​for model training; for each grid cell i, the prediction model for the pollutant degradation degree D_i is:

[0043] D i =RF(NDVIi,NDWI) i ,SAVIi,Tempi,Moisturei,pH iEC i );

[0044] Where RF represents the trained random forest model, and Temp_i and Moisture_i are the soil temperature and humidity measurements of the network unit, respectively.

[0045] Optionally, in step S5, the visualization platform also integrates remediation process simulation and prediction functions. Based on the current remediation efficiency and historical data, it uses a historical sequence prediction algorithm to predict the pollutant concentration change trend of each grid cell in the future, and provides early warning of areas where the remediation effect is not up to standard. For a certain grid cell, the predicted pollutant concentration value [C]{i,t+T} after T days is calculated using the following LSTM model:

[0046] [C] i,t+T =LSTM([C]) i,t [C] i,t-1 ,...,[C] i,t-n η i,t ,Env i );

[0047] Where [C]{i,t},[C]{i,t-1},...,[C]{i,tn} are the pollutant concentration sequences of the grid cell over the past n time points, and Env_i represents the environmental parameter vector of the grid cell.

[0048] Optional, also includes:

[0049] Step S6: Based on the evaluation and prediction results of Step S5, dynamically adjust and optimize the repair strategy; for regions where the repair efficiency η_{i,t} is lower than the preset threshold θ, automatically generate enhanced repair schemes, including increasing the amount of repair agent, adjusting the composition of the repair agent, or changing the frequency and intensity of physical repair operations; the optimization model aims to minimize the total repair cost C_total, and its function is expressed as:

[0050]

[0051] Where c_{\text{material},i} and c_{\text{energy},i} are the unit cost of the repair agent and the unit cost of energy consumption of the i-th grid cell, respectively; m_i and E_i are the amount of repair agent added and the energy consumption, respectively; and η_{i,t+T} is the predicted repair efficiency of the cell after time T.

[0052] Optionally, in step S2, the drone's flight path employs a reinforcement learning-based intelligent path planning algorithm, aiming to maximize the amount of information acquired during each flight and minimize flight energy consumption; its state space includes the drone's current position, battery level, and covered area, while its action space includes flight direction and speed. The reward function R is designed as follows:

[0053] R=w1·∑informationGain-w2·EnergyConsumption;

[0054] InformationGain is the information gain obtained by flying over insufficiently monitored areas;

[0055] EnergyConsumption represents the energy consumption of this flight, and w1 and w2 are weighting coefficients.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] By organically combining UAV multispectral remote sensing with ground-based IoT sensor networks, a three-dimensional, multi-layered monitoring system has been constructed, overcoming the limitations of single technical means in terms of spatiotemporal coverage and data dimensions, and realizing comprehensive and real-time perception of the soil remediation process.

[0058] Based on a spatial grid model and an adaptive densification strategy, the monitoring density can be dynamically adjusted according to the characteristics of pollution distribution. Through multi-source data fusion and advanced interpolation and inversion algorithms, the shortcomings of a single data source are effectively compensated for, and the accuracy and reliability of inversion of key soil parameters (such as heavy metal content and pollutant degradation degree) are significantly improved.

[0059] By calculating repair efficiency and integrating time series prediction models such as LSTM, the current repair effect can be dynamically evaluated and future trends can be accurately predicted, realizing the transformation from "post-event monitoring" to "pre-event early warning" and providing forward-looking guidance for repair decisions.

[0060] Based on monitoring and prediction results, a dynamic optimization model for repair strategies with the goal of minimizing costs was established. This model can automatically identify areas with poor repair results and generate accurate reinforcement repair solutions, which significantly improves repair efficiency and reduces overall repair costs and time.

[0061] By adopting a reinforcement learning-based intelligent path planning strategy for drones and an optimized deployment strategy for IoT nodes, the system can adapt to environmental changes, acquire the maximum amount of information with minimal energy consumption, greatly reduce manual intervention, and improve the overall operational efficiency and intelligence level of the monitoring system. Attached Figure Description

[0062] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0063] Figure 1 This is a schematic diagram of the real-time monitoring method for soil remediation using the coupling of UAV multispectral and IoT sensing, provided in an embodiment of the present invention. Detailed Implementation

[0064] 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.

[0065] The purpose of this invention is to provide a real-time monitoring method for soil remediation that integrates UAV multispectral remote sensing and IoT sensing technologies to achieve integrated air-ground and multi-source data collaboration.

[0066] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0067] Example 1:

[0068] The monitoring area is the site of an abandoned electroplating plant, with a total area of ​​approximately 10,000 square meters (100m × 100m). The main pollutant in the soil is hexavalent chromium, with an average initial concentration of approximately 350 mg / kg, and the highest concentration in some areas exceeding 800 mg / kg. Chemical reduction stabilization technology is planned for remediation, with an estimated remediation period of 90 days.

[0069] The spatial resolution of the UAV multispectral camera is 0.05 meters, and the effective sensing radius of the IoT soil sensor is 2 meters. Taking an adjustment coefficient k = 1.0, the side length of the grid cell is calculated to be L = 1.0 * (0.05 + 2) / 2 ≈ 1.025 meters. For ease of calculation, the monitoring area is divided into 9801 (99 × 99) regular grid cells, each approximately 1 square meter.

[0070] Using a drone equipped with a five-band multispectral camera, the ground resolution (GSD) for the flight mission was set to 0.05 meters, the camera focal length f = 5.5 mm, and the pixel size = 3.4 μm. The calculated flight altitude H = (5.5 * 3.4) / 0.05 ≈ 374 meters. The drone conducted aerial photography every 7 days to collect multispectral image data.

[0071] A total of 105 IoT sensor nodes were deployed at the grid nodes. Each node integrates sensors to measure: soil temperature (range: -40℃~80℃, accuracy: ±0.5℃), soil volumetric moisture content (range: 0~100%, accuracy: ±3%), pH value (range: 0-14, accuracy: ±0.1%), and electrical conductivity EC (range: 0-23dS / m, accuracy: ±2%). The data acquisition interval Δt was set to 2 hours, and the data was transmitted to the cloud data center via LoRaWAN.

[0072] After preprocessing the collected multispectral data, the NDVI, NDWI, and SAVI indices for each grid cell were calculated. Ground sensor data were cleaned and interpolated to ensure data integrity. Using a spatiotemporal matching model, the multispectral indices within each square meter of the grid were correlated with hourly soil temperature, humidity, pH, and EC data from the nearest sensor, forming a fused data unit.

[0073] Laboratory-measured data (hexavalent chromium concentration and organic matter content) from 30 grid cells were collected as training samples. Based on the fused data, a hexavalent chromium concentration inversion model was established. After training, the model accuracy (R²) was [not specified]. 2 The value reached 0.87. The model was applied to the entire field to reproduce the spatial distribution of hexavalent chromium concentration in each grid cell at the initial time t0 before repair, showing that the concentration ranged from 120 to 920 mg / kg.

[0074] After the remediation begins, the concentration of hexavalent chromium across the entire area is inverted every 7 days based on the latest data. For example, on day 30 of the remediation, the inverted concentration of a high-risk grid cell (initial concentration [C]{i,t0}=810mg / kg) is [C]{i,30}=410mg / kg, and its remediation efficiency η_{i,30}=(1-410 / 810)*100%≈49.4%. The visualization platform generates a heatmap of the remediation efficiency, showing that the efficiency in most areas is between 30% and 55%, but the efficiency in a local area in the northwest corner is below 20%, and the platform automatically issues a yellow warning.

[0075] For the warning area (repair efficiency <25%), the system generated an optimized plan: increasing the dosage of the repair agent (ferrous sulfate) from the original 3% to 4.5%, and suggesting an additional stirring operation. After implementing the optimized strategy for 14 days, the repair efficiency in this area increased to 38%, and the warning was lifted.

[0076] This method enables 24 / 7, comprehensive, and real-time dynamic monitoring of the restoration process. Compared to traditional manual sampling inspections (which take 3-5 days each time and only obtain data from a limited number of points), this method shortens the monitoring and evaluation cycle to a few hours, increases the amount of data by two orders of magnitude, and provides accurate and efficient decision support for the restoration project. Ultimately, the site achieved a 100% acceptance rate after restoration.

[0077] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0078] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A real-time monitoring method for soil remediation utilizing the coupling of UAV multispectral sensing and IoT sensing, characterized in that, Includes the following steps: S1. Construct a spatial grid model of the monitoring area. Based on the topography, area, and pollution distribution characteristics of the contaminated site to be remediated, the monitoring area is divided into several regular or irregular spatial grid units. Each grid unit has a unique spatial coordinate identifier. The side length L of the spatial grid unit is determined jointly by the spatial resolution R of the UAV multispectral image and the effective sensing radius r of the IoT ground sensor. The calculation formula is as follows: Where k is an adjustment coefficient, ranging from 0.8 to 1.5, used to balance data density and computational efficiency; S2, an integrated air-space-ground detection network is deployed for data acquisition: S21. Spatial Monitoring: Using an unmanned aerial vehicle (UAV) platform equipped with a multispectral camera, periodic aerial photography is conducted on the monitoring area according to a preset flight path to acquire multispectral image data covering the entire monitoring area; the multispectral camera covers at least five bands: blue light, green light, red light, red edge, and near-infrared; the flight altitude H is determined based on the required ground resolution (GSD), and its calculation formula is as follows: Where f is the camera focal length and the pixel size is the camera pixel size; S22. Ground monitoring: An Internet of Things (IoT) sensor node network is deployed at the nodes or center of the spatial grid unit; each sensor node integrates at least a soil temperature sensor, a soil moisture sensor, a pH sensor, and a conductivity sensor; the sensor node collects soil environmental parameter data at a set time interval Δt and transmits the data to an edge computing gateway or cloud data center via wireless transmission. S3. Multi-source spatiotemporal data preprocessing and fusion: S31. Perform radiometric calibration, atmospheric correction and geometric correction on the acquired multispectral image data, and calculate the normalized difference in vegetation index (NDVI), normalized difference in water content (NDWI) and soil-adjusted vegetation index (SAVI). S32. Perform outlier removal, missing value imputation, and spatiotemporal alignment processing on the soil environmental parameter data collected by IoT sensor nodes. S33. Establish a spatiotemporal matching model for air-space-ground data: resample the preprocessed multispectral image data to a resolution that matches the spatial grid cell, and associate the multispectral feature index within each grid cell with ground sensor data within the same spatiotemporal range to form an integrated air-space-ground fusion data unit. S4. Inversion and Modeling of Key Parameters for Soil Remediation: Based on the fused data units obtained in step S3, an inversion model for key soil parameters is established; the key parameters include soil heavy metal content, organic matter content, and pollutant degradation degree. For the inversion of soil heavy metal content [Cd], a method combining multiple linear regression and machine learning was adopted. The inversion model is as follows: [CD] i =α0+α1·NDVI i +α2·NDWI i +a3·pH i +α4·EC i +e i ; Where [Cd]_i represents the predicted heavy metal content of the i-th grid cell, NDVI_i and NDWI_i are the vegetation index and moisture index of the grid cell, respectively, pH_i and EC_i are the measured soil pH and electrical conductivity of the grid cell, respectively, α0~α4 are the model regression coefficients, which are obtained by training with historical sample data, and ε_i is the random error term. S5. Dynamic evaluation and visualization of the repair process: Based on the spatiotemporal sequence data of key soil parameters obtained from step S4, the remediation efficiency η of each grid cell at different time points is calculated, and the calculation formula is as follows: Where η_{i,t} represents the repair efficiency of the i-th grid cell at time t, [C]{i,t} represents the target pollutant concentration of the grid cell at time t, and [C]{i,t0} represents the target pollutant concentration of the grid cell at the initial repair time t0; Generate spatiotemporal distribution maps of remediation efficiency and trend maps of pollutant concentration changes, and display and provide early warnings in real time through a visualization platform.

2. The method for real-time monitoring of soil remediation using multispectral coupling of UAV and IoT sensing as described in claim 1, characterized in that, In step S1, the division of the spatial grid cells also considers the pollutant migration and diffusion model. For areas with drastic changes in pollutant concentration gradients, an adaptive densification network strategy is adopted. The formula for calculating the grid side length L' is as follows: Where β is the encryption strength coefficient, This represents the maximum pollutant concentration gradient modulus for this grid cell.

3. The method according to claim 1, characterized in that, In step S22, the deployment of the IoT sensor nodes adopts an optimized deployment strategy. The objective function is to minimize the total number of nodes N while ensuring full coverage monitoring. Its mathematical model is expressed as follows: Where M is the total number of potential deployment potentials, x_i is the decision variable, S(r) represents the coverage circle with potential i as the center and radius r, and A represents the entire monitoring area.

4. The method for real-time monitoring of soil remediation using multispectral coupling of UAV and IoT sensing as described in claim 1, characterized in that, In step S3, the spatiotemporal alignment process employs a time series imputation and spatial interpolation method based on Kriging interpolation to interpolate the non-uniformly acquired ground sensor data onto a unified spatiotemporal grid. Specifically, for any sensor's missing parameter value P(s_i,t_j) at time t_j, spatial interpolation estimation is performed using the observations from neighboring sensors at time t_j. Where n is the number of adjacent sensors used for interpolation, and λ_k is the Kriging weight coefficient, which is determined by the variogram model.

5. The method for real-time monitoring of soil remediation using multispectral coupling of UAV and IoT sensing as described in claim 1, characterized in that, In step S4, the inversion model for the key soil parameters is further trained and predicted using a random forest regression algorithm. Multispectral feature indices and ground-sensor data are used as input features, and laboratory-measured soil parameters are used as target values ​​for model training. For each grid cell i, the prediction model for the pollutant degradation degree D_i is as follows: D i =RF(NDVI i ,SEVEN i ,SAVI i ,Temp i ,Moisture i ,pH i ,EC i ); Where RF represents the trained random forest model, and Temp_i and Moisture_i are the soil temperature and humidity measurements of the network unit, respectively.

6. The method for real-time monitoring of soil remediation using multispectral coupling of UAV and IoT sensing as described in claim 1, characterized in that, In step S5, the visualization platform also integrates remediation process simulation and prediction functions. Based on the current remediation efficiency and historical data, it uses a historical sequence prediction algorithm to predict the pollutant concentration change trend of each grid cell in the future, and provides early warning of areas where the remediation effect is not up to standard. For a certain grid cell, the predicted pollutant concentration value [C]{i,t+T} after T days is calculated using the following LSTM model: [C] i,t+T =LSTM([C] i,t ,[C] i,t-1 ,...,[C] i,t-n ;η i,t ,Env i ); Where [C]{i,t},[C]{i,t-1},...,[C]{i,tn} are the pollutant concentration sequences of the grid cell over the past n time points, and Env_i represents the environmental parameter vector of the grid cell.

7. The method for real-time monitoring of soil remediation using multispectral coupling of UAV and IoT sensing as described in claim 1, characterized in that, Also includes: Step S6: Based on the evaluation and prediction results of step S5, dynamically adjust and optimize the repair strategy; For regions where the repair efficiency η_{i,t} is lower than a preset threshold θ, an enhanced repair plan is automatically generated, including increasing the amount of repair agent, adjusting the composition of the repair agent, or changing the frequency and intensity of physical repair operations; the optimization model aims to minimize the total repair cost C_total, and its function is expressed as: Where c_{\text{material},i} and c_{\text{energy},i} are the unit cost of the repair agent and the unit cost of energy consumption of the i-th grid cell, respectively; m_i and E_i are the amount of repair agent added and the energy consumption, respectively; and η_{i,t+T} is the predicted repair efficiency of the cell after time T.

8. The method for real-time monitoring of soil remediation using multispectral coupling of UAV and IoT sensing as described in claim 1, characterized in that, In step S2, the flight path of the UAV adopts an intelligent path planning algorithm based on reinforcement learning, with the goal of maximizing the amount of information acquired in each flight and minimizing flight energy consumption. Its state space includes the drone's current position, battery level, and covered area; its action space includes flight direction and speed; and the reward function R is designed as follows: R=w1·∑informationGain-w2·EnergyConsumption; InformationGain is the information gain obtained by flying over insufficiently monitored areas; EnergyConsumption represents the energy consumption of this flight, and w1 and w2 are weighting coefficients.

9. An electronic device, characterized in that, It includes a processor and a memory, the memory storing computer-readable instructions that, when executed by the processor, perform the method as described in any one of claims 1-8.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it performs the method as described in any one of claims 1-8.

Images