A "five-base" synergic ecological remote sensing monitoring method and system

Abstract

The application discloses a kind of " five basic " synergic ecological remote sensing monitoring method and system, belong to synergic control technical field, including design multiple nodes, multi-level interconnection " five basic " synergic monitoring network, establish the monitoring data system with high space-time resolution;R&D demand-driven " five basic " monitoring platform adaptation decision model, multi-task target matching optimization model, formulate the optimal allocation scheme between the available execution platform of to-be-allocated monitoring task and available execution platform.The application forms the closed-loop monitoring of wide-area screening-key locking-precise verification-blindness verification by constructing multi-means synergic linkage monitoring new paradigm, solves the problem of multi-precision, multi-frequency synergic monitoring in the scene such as regional ecological quality monitoring, biodiversity protection, human interference activity control.

Description

Technical Field

[0001] This invention relates to the field of collaborative control technology, and more specifically to a "five-based" collaborative ecological remote sensing monitoring method and system. Background Technology

[0002] With rapid economic and social development, a large amount of important ecological space has been encroached upon, occupied, and destroyed. In the past 40 years, the areas of forests, grasslands, and wetlands have decreased by 35.9%, 21.5%, and 37.5% respectively—the magnitude and speed of this decline are alarming. To promptly grasp the spatiotemporal characteristics of the ecological status of important spaces and effectively detect the damage caused to ecosystems by illegal human activities, comprehensive, high-precision, and short-cycle remote sensing monitoring of the ecological environment is needed to stop destructive behavior as early as possible and avoid ecological loss.

[0003] The existing monitoring system employs a single monitoring platform. Each of the monitoring methods—space-based satellites, aerial drones, and ground-based observations—has its own spatiotemporal coverage blind spots or accuracy limitations. Space-based satellites struggle to achieve both high temporal and spatial resolution; tower-based remote sensing has limited coverage; drone operation time is restricted; and ground-based observations can only cover discrete points, failing to meet the comprehensive, high-precision, and short-cycle requirements of composite monitoring. More critically, the existing monitoring system lacks a unified system architecture and standardized sensing network that defines the heterogeneous collaboration of multiple platforms (space, air, tower, and ground). Each platform operates independently, resulting in dispersed monitoring capabilities and unclear role definitions.

[0004] Secondly, from the discovery of macro-level anomalies to on-site verification at the micro-level, and then to long-term fixed-point monitoring, the various monitoring links are disconnected. Currently, there is a lack of an automated, event-driven, multi-platform collaborative business process and task scheduling strategy, resulting in a breakpoint between problem discovery and problem handling, delayed response, inefficient handling, and difficulty in timely converting clues discovered by remote sensing into evidence for on-site law enforcement.

[0005] Furthermore, in multi-platform, multi-task scenarios, existing technologies lack automatic matching and scheduling methods based on quantitative calculations, making it difficult to optimally allocate limited monitoring resources (such as airborne platforms, drones, and patrol vehicles) to multiple anomaly clues to be verified, thus failing to guarantee global optimization.

[0006] Therefore, how to provide a "five-based" collaborative ecological remote sensing monitoring method and system is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0007] In view of this, the present invention provides a "five-based" collaborative ecological remote sensing monitoring method and system. By constructing a "five-based" collaborative monitoring system architecture, a standardized four-level closed-loop monitoring business process of full-domain screening, key point locking, precise verification, and blind spot verification is realized. Combined with intelligent scheduling strategies, the optimal combination and automatic linkage of multi-platform capabilities are achieved.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: On the one hand, this invention provides a "five-based" collaborative ecological remote sensing monitoring method, including: Construct a collaborative monitoring network integrating space, air, tower, vehicle, and ground systems; Establish a demand-driven "five-based" monitoring platform adaptation decision model to determine the optimal combination of monitoring platforms that meets actual needs; A multi-task target matching optimization model is constructed to determine the optimal allocation scheme between monitoring tasks to be assigned and available execution platforms, thus establishing a closed-loop monitoring system encompassing wide-area screening, key target identification, precise verification, and blind spot validation.

[0009] Preferably, a demand-driven "five-based" monitoring platform adaptation decision model is established to determine the optimal combination of monitoring platforms that meets actual needs, specifically including: Construct a monitoring demand vector and a platform capability benchmark set; For each type of platform, calculate the spatial coverage matching degree, resolution matching degree, and timeliness matching degree separately; The overall matching degree of each platform is calculated based on spatial coverage matching degree, resolution matching degree, and timeliness matching degree. When the overall matching degree of the platform is greater than or equal to the preset value, the platform is selected to perform the monitoring task. When the overall matching degree of all platforms is less than the preset value, the Monte Carlo random combination optimization method is used to randomly combine and evaluate the platform dimensional capabilities to select the optimal platform combination.

[0010] Preferably, a multi-task target matching optimization model is constructed to determine the optimal allocation scheme between the monitoring tasks to be assigned and the available execution platforms, specifically including: Configure weights based on the type of business scenario; A comprehensive score is calculated on the execution capability of each task to be assigned, based on the weights configured according to the business scenario type. Establish a multi-objective matching optimization model: And satisfy the condition that each task can be assigned to at most one platform, and each platform can execute at most K tasks simultaneously. q The constraint that the total execution time of each task and the platform does not exceed the available time window, where x pq For decision variables; Solve the multi-objective matching optimization model and output the optimal allocation scheme.

[0011] The preferred closed-loop monitoring system, which includes broad-area screening, targeted focus, precise verification, and blind spot validation, specifically includes: The target area is detected by space-based satellites according to the first preset cycle. Based on the change detection results, the first-level anomaly patches and the spatial location information corresponding to the first-level anomaly patches are extracted. The spatial location of the primary anomaly patch is spatially matched with the monitoring coverage area of ​​the airborne remote sensing platform. The primary anomaly patches falling within the monitoring coverage area are pushed to the corresponding airborne remote sensing platform as target tasks. The airborne remote sensing platform then scans and identifies the target task area to generate secondary alarm clues. For Level 1 abnormal patches that are not within the monitoring coverage of the airborne remote sensing platform, dispatch aerial drones or mobile patrol and monitoring vehicles to conduct on-site verification. A heat map is constructed by combining the spatial distribution of historical anomalous events, and the deployment of ground monitoring equipment is determined based on the heat value of new event locations in the heat map.

[0012] On the other hand, the present invention also provides a system for implementing the above-mentioned "five-based" collaborative ecological remote sensing monitoring method, comprising: multiple monitoring platforms, a communication module, and a management platform, wherein the management platform is connected to the monitoring platforms through the communication module and is used to acquire the capability parameters and monitoring data of each monitoring platform; The management platform is also used to establish a demand-driven "five-based" monitoring platform adaptation decision model based on the capability parameters of each monitoring platform, and determine the optimal combination of monitoring platforms that meets actual needs; during the execution of multi-level monitoring processes, a multi-task target matching optimization model is constructed to determine the optimal allocation scheme between the monitoring tasks to be assigned and the available execution platforms, forming a closed-loop monitoring system of wide-area screening, key focus locking, precise verification, and blind spot verification.

[0013] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a "five-based" collaborative ecological remote sensing monitoring method and system. By designing a "five-based" collaborative monitoring network, establishing a demand-driven monitoring platform adaptation model, developing a multi-task target matching optimization strategy, and formulating a multi-level collaborative monitoring process, it provides a multi-source, multi-scale monitoring method and practical path of "space-based full-domain perception - tower-based ring locking - flight path verification - UAV dynamic monitoring - ground fixed-point verification". This solves the technical problems in existing ecological remote sensing monitoring, such as unclear multi-platform collaborative monitoring methods and unclear multi-level monitoring processes.

[0014] By designing and implementing a multi-level closed-loop monitoring process encompassing wide-area screening, key focus, and precise verification, an automated linkage from discovery to handling is achieved, driven by anomaly clues. This eliminates information gaps and delays caused by manual handover between stages in traditional monitoring models, significantly improving response efficiency from anomaly discovery to on-site verification. The multi-platform selection decision-making and multi-task scheduling processes are modeled as optimal path selection and multi-objective matching optimization problems. By comprehensively considering multi-dimensional factors such as spatiotemporal matching degree, capability matching degree, and platform availability, the system can automatically solve for the globally optimal allocation scheme under actual constraints, replacing traditional manual experience-based scheduling and greatly improving resource utilization efficiency and the scientific nature of scheduling decisions. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the process provided by the present invention.

[0017] Figure 2 This is a prototype system for collaborative monitoring of the "five basic elements".

[0018] Figure 3 The diagram shows the structure of the monitoring system provided by this invention.

[0019] Figure 4 This is a diagram showing the components of the "Five Basics" collaborative ecological monitoring platform. Detailed Implementation

[0020] 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.

[0021] This invention discloses a "five-based" collaborative ecological remote sensing monitoring method, such as... Figure 1 As shown, it includes: Construct a five-based collaborative monitoring system integrating space, air, tower, vehicle, and ground, establish a monitoring data system that takes into account high spatiotemporal resolution, design a multi-node, multi-level interconnected collaborative monitoring network, and utilize the capability parameters of each monitoring platform to establish a five-based collaborative monitoring data system integrating space, air, tower, vehicle, and ground at any spatiotemporal coordinate.

[0022] This invention defines monitoring methods into five collaborative levels (referred to as "five basic levels") based on spatial dimensions and platform characteristics, and clarifies the system roles of each level, such as... Figure 2 As shown, it specifically includes: Space-based satellites provide continuous full-spectrum data through visible light, multispectral, hyperspectral, and thermal infrared payloads, supporting the monitoring of land surface changes and the retrieval of vegetation parameters (NDVI, GPP, etc.) and environmental parameters (ET, APAR, etc.). Their wide coverage and spatial continuity provide fundamental support for the monitoring of human activities in ecological spaces and ecological assessments, making them a core means of comprehensive monitoring. They are responsible for regular scanning and change detection across the entire region, representing the most direct monitoring method for identifying problems.

[0023] Space-based remote sensing: Deploying ultra-long-range infrared, fog-penetrating cameras, and lidar equipment on platforms such as high towers and tall buildings, it achieves decimeter-level resolution and minute-level frequency monitoring. It possesses 360-degree high-precision observation capabilities, focusing on monitoring ecological damage scenarios such as deforestation and mining in key protected areas and sensitive, vulnerable regions. Space-based remote sensing is responsible for high-frequency, mesoscale continuous monitoring of key areas, effectively compensating for the shortcomings of satellite timeliness and accuracy, and improving the speed of early warning and response to abnormal events.

[0024] Aerial drones: Equipped with sensors such as visible light, multispectral, and lidar, they overcome cloud limitations to perform low-altitude, sub-cloud flight missions, achieving sub-meter level precision monitoring of vegetation types, ecological anomalies, and human activities. Aerial drones are responsible for on-demand, centimeter-level accurate verification of specific targets, filling gaps in satellite and airborne operations, and are a key verification tool for ecological change assessment and regular patrols.

[0025] Mobile patrol and monitoring vehicle: Integrating vehicle-mounted lidar, hyperspectral imager, thermal infrared and other equipment, it conducts three-dimensional monitoring of roadside vegetation, human activities, biodiversity, and other aspects. The mobile patrol and monitoring vehicle is responsible for linear mobile monitoring and on-site verification along the road network, supporting multi-element three-dimensional data acquisition and patrol enforcement linkage, covering near-ground blind spots, and improving the efficiency of large-scale continuous monitoring.

[0026] Ground-based observation equipment: This involves establishing an all-weather sensor network including video probes, spectrometers, and infrared cameras to continuously acquire data on vegetation quality, biodiversity, and meteorological parameters. This equipment is responsible for continuous, round-the-clock monitoring of key locations and verification of product authenticity, supporting the verification of satellite and space-based remote sensing products. It also serves as the "last mile" defense line for identifying hidden ecological problems.

[0027] The "five-based" collaborative monitoring mechanism in this embodiment of the invention quantifies the technical capabilities of each platform through a multi-level spatiotemporal-precision capability matrix and a "five-based" collaborative monitoring capability definition function. Combined with the spatial distribution of deployed monitoring payloads, it clarifies the level of refined monitoring in each region of important ecological spaces, supporting the formulation of monitoring plans for each region, as well as future deployment of monitoring payloads to fill gaps and conduct mobile monitoring. The capability parameters of each monitoring platform are shown in Table 1.

[0028] Table 1. Monitoring Platform Capability Parameters

[0029] To quantify the level of refined regional monitoring after the synergy of the "five basic elements," this invention defines a "five basic elements" synergistic monitoring capability definition function C(x,y,t), representing the optimal monitoring spatial resolution that the system can provide at any spatiotemporal coordinate (x,y,t).

[0030] Where: I[] is an indicator function, which takes the value 1 if the condition is true, otherwise takes the value 0; Ω s For the coverage area of ​​the base layer, Ω a (i) represents the coverage area of ​​the i-th airborne platform (denoted by P). i Center r a (circular region with radius Ω) u (j,t) represents the reachable coverage area of ​​the j-th UAV at time t, Ω m (k,t) represents the effective scan zone of the k-th patrol vehicle along the road at time t; the max function takes the optimal resolution that all available platforms can provide at (x,y,t).

[0031] Based on the monitoring platform's capability functions, a demand-driven monitoring platform adaptation decision model is established, outputting selection suggestions and matching scores for single-platform / multi-platform collaboration.

[0032] (1) Define the monitoring demand vector D as the core independent variable.

[0033] D=(S req ,R req ,T req )

[0034] S req Spatial coverage requirements (including coverage area and spatial form: area / strip / point); R reqRequired spatial resolution (the smaller the value, the higher the precision requirement); T req Timeliness requirements (including monitoring cycle, response time, and effective monitoring time).

[0035] (2) Establish a platform capability benchmark set

[0036] P={P s ,P a ,P u ,P m ,P l}

[0037] Corresponding in order: space-based satellites, tower-based airborne remote sensing, aerial drones, mobile patrol and monitoring vehicles, and ground-based observation equipment. Decision function output.

[0038] (3) Calculation of overall matching degree of each platform

[0039] For each type of platform p∈P, the matching degree is calculated from three dimensions: spatial coverage, resolution, and timeliness, and then weighted to obtain the comprehensive matching degree.

[0040] ① Spatial coverage matching degree M S (p)

[0041] Assess whether the platform's coverage and form meet the requirements. Let Ap be the spatial coverage of platform p. The matching degree is calculated using the following formula:

[0042] Among them, Tianji A s : Regional / basin-level areal coverage (>10000km²); Tower base A a =π×20km 2 ≈1256km 2 (Circular surface); Drone A u =π×30km 2 ≈2827km 2 (Circular accessible area); Patrol vehicle A m Road length × 5km bandwidth (linear strip); Ground A l Point level (<0.1km) 2 ).

[0043] ② Resolution matching degree M R (p)

[0044] We need to assess whether the platform's monitoring accuracy meets the requirements. Let the spatial resolution of platform p be R. p Matching degree calculation formula:

[0045] Among them, Tianji R s: 1m / 16m / 30m; Tower base R a Better than 1m; UAV R u Approximately 0.05–0.1 m; patrol vehicle R m Approximately 0.1 to 1 meter; ground monitoring is not involved.

[0046] ③Timeliness matching degree M T (p)

[0047] We need to assess whether the platform's response speed and monitoring frequency meet the requirements. Let T be the monitoring frequency of platform p. p Matching degree calculation formula:

[0048] Among them, space-based T s : Conducted bi-monthly, ≥6 times per year; Tower Base T a Daily scanning, ≥360 days a year; UAV T u Daily flights, ≥300 days per year; Patrol vehicle T m Patrols as needed, ≥8 times per year; Ground T l Continuous monitoring for ≥360 days per year.

[0049] ④ Calculation of overall matching degree M(p)

[0050] M(p) represents the platform's overall matching degree, calculated using the following formula: M(p) = wS M S (p)+wR M R (p)+wT M T (p) Among them, wS+wR+wT=1 represents the weights of the three dimensions, which can be dynamically adjusted according to task priority.

[0051] (4) Selection of monitoring methods

[0052] When M(p)≥0.85 for a certain platform, that platform is selected to perform the monitoring task.

[0053] When M(p) < 0.85 for all platforms, the Monte Carlo random combination optimization method is used to complete the random combination and evaluation of platform dimension capabilities. That is, in the capability dimension set of the "five basic" platforms, the spatial matching capability, resolution matching capability, and timeliness matching capability of different platforms are randomly selected according to the set constraints to generate a large number of candidate collaborative solutions. The optimal platform combination is selected through quantitative evaluation.

[0054] For each randomly generated platform combination G={p1,p2,...,p5}, the overall matching degree is calculated based on the rule of "optimizing and complementing dimensional capabilities": Spatial coverage matching degree: Take the maximum value of the spatial coverage matching degree of all platforms in the combination, and assign the platform with the strongest coverage capability to undertake the task of full coverage and make up for the spatial shortcomings. S (G)=max{M S (p1),M S (p2),...,M S (p5)}; Resolution matching degree: Take the maximum value of the resolution matching degree of all platforms in the combination, and let the platform with the highest accuracy undertake the detailed verification task and make up for the accuracy deficiency. R (G)=max{M R (p1),M R (p2),...,M R (p5)}; Timeliness Matching Degree: The maximum timeliness matching degree among all platforms in the combination is taken, and the platform with the fastest response time is assigned to the emergency response task to make up for the timeliness deficiency. T (G)=max{M T (p1),M T (p2),...,M T (pn)}; Combined overall matching degree: The weighted calculation is based on the single platform weight system, while introducing a collaborative gain coefficient ΔM (range 0~0.12): the stronger the complementarity of the capabilities between platforms (such as large-range low precision + small-range high precision), the higher the gain value; the combined gain of platforms with homogeneous capabilities is 0.

[0055] M(G)=wS M S (G)+wR M R (G)+wT M T (G)+ΔM

[0056] During the execution of the multi-level monitoring process, the optimal allocation scheme between the monitoring tasks to be assigned and the available execution platforms is determined based on a multi-objective matching optimization model.

[0057] When there are N monitoring tasks {Q} to be assigned simultaneously p |p=1,2,...,N} (including primary clues generated by space-based screening, secondary clues from air-based locking, daily patrol tasks, etc.) and M available execution platforms {Z q When |q=1,2,...,M} (including airborne remote sensing, UAVs, and patrol vehicles), the tasks need to be optimally allocated to each platform. This invention models this as a 0-1 integer programming problem.

[0058] (1) Platform execution capability scoring function

[0059] For each task to be assigned, the platform (Q) p Z q The overall performance score is calculated. The scoring function selects different parameter emphases based on the scenario type:

[0060] in: Spatial matching degree: d is the distance between the platform's current position and the task coordinates, σ d f is the distance attenuation coefficient; res (R q () represents resolution matching degree: R q For platform spatial resolution, R req f is the minimum resolution required for this scene task; avail (Z q Platform availability is determined by a comprehensive consideration of current task load, remaining battery power / battery life, weather conditions, etc., and is rated from 0 to 1. Configurable weights.

[0061] (2) Scenario-specific weight configuration

[0062] Different business scenarios emphasize different aspects of the above scoring factors, and the weight configuration is shown in Table 2.

[0063] Table 2 Weight Configuration for Different Business Scenarios

[0064] (3) Develop a multi-objective matching optimization model

[0065] A multi-objective matching optimization model is established, formalizing the task allocation problem as a 0-1 integer programming problem. The details are as follows:

[0066] Each task can be assigned to a maximum of one platform: ; Each platform can execute a maximum of K concurrently. q One task: ; The total execution time of platform q does not exceed the available time T. q : ; Decision variables: ; Where: x pq x is the decision variable. pq =1 indicates that task p is assigned to platform q; K q t represents the maximum number of concurrent tasks on the platform. pqThe estimated time T is required for platform q to execute task p. q This is the available time window for platform q.

[0067] Furthermore, the specific steps of the multi-level monitoring process include: The target area is detected by space-based satellites according to the first preset cycle. Based on the change detection results, the first-level anomaly patches and the spatial location information corresponding to the first-level anomaly patches are extracted. The spatial location of the primary anomaly patch is spatially matched with the monitoring coverage area of ​​the airborne remote sensing platform. The primary anomaly patches falling within the monitoring coverage area are pushed to the corresponding airborne remote sensing platform as target tasks. The airborne remote sensing platform then scans and identifies the target task area to generate secondary alarm clues. For Level 1 abnormal patches that are not within the monitoring coverage of the airborne remote sensing platform, dispatch aerial drones or mobile patrol and monitoring vehicles to conduct on-site verification. A heat map is constructed by combining the spatial distribution of historical anomalous events, and the deployment of ground monitoring equipment is determined based on the heat value of new event locations in the heat map.

[0068] Specifically, space-based satellites are used to detect changes in the target area according to a first preset cycle. Based on the change detection results, primary anomaly patches and their corresponding spatial location information are extracted, including: Remote sensing images of the target area are acquired using space-based satellites according to a first preset period. Optionally, the first preset period can be set to two months. The remote sensing images from two adjacent periods are input into a weight-sharing dual-branch deep residual network to extract multi-scale semantic feature maps step by step. and Where s=1,2,...,S represents multiple feature levels from shallow details to deep semantics. A bi-branch design with shared weights ensures that the feature representation of the same land cover in two image periods is mapped to the same feature space, guaranteeing semantic consistency in subsequent change comparisons.

[0069] At each feature level, cross-temporal multi-head cross-attention operations are performed on the multi-scale semantic feature maps of two adjacent periods to obtain the change difference feature maps:

[0070] Among them, W Q W K W V The CrossAttn() function calculates the learnable projection matrix as follows: As an attention head dimension, cross-temporal attention uses features from earlier images as queries and features from later images as keys and values. It automatically focuses on semantically related regions between the two images that have spectral or structural differences, and outputs an enhanced feature map of changes. .

[0071] Feature map of changes By fusing features through a Feature Pyramid Network (FPN), a unified variation feature map F with both detail localization and semantic discriminative capabilities is obtained. diff ; Based on the unified change feature map, the change detection head outputs a change probability map P for each pixel. change (x,y); The change probability map is segmented by a threshold to obtain an initial change map patch set; The initial set of changed patches is filtered through multiple levels to obtain first-level abnormal patches and their corresponding spatial location information.

[0072] Furthermore, the multi-level screening includes: First-level set filtering: Remove connected pixels N k Below the first preset threshold N min (Usually 10 pixels) or area A k Below the second preset threshold (200m) 2 The initial change pattern of the patch; Second-level morphological screening: Calculate the compactness index of each initial variation patch. Remove compactness index C k Below the third preset threshold C th (e.g. C) th The initial change patch (=0.3) is used to remove linear pseudo-changes (such as road shadows, cloud edges, striped tillage marks, etc.). k A value close to 1 indicates a shape that is close to a circle, which is a typical morphological feature of human engineering activities or natural patchy changes.

[0073] The third-level semantic filtering involves inputting the spectral and texture features of the retained initial changed patches into a pre-trained random forest change classifier. The classifier outputs the change type (e.g., suspected construction, suspected mining, suspected logging) and its confidence level. Initial changed patches with a confidence level higher than a fourth preset threshold (e.g., 75%) are output as the first-level anomalous patches. The first-level anomalous patches include their center coordinates (x, y, y). p ,y p Area A p Change type label L p Confidence level C p .

[0074] In another embodiment, the spatial location of the primary anomaly patches is spatially matched with the monitoring coverage area of ​​the space-based remote sensing platform. Primary anomaly patches falling within the monitoring coverage area are pushed as target tasks to the corresponding space-based remote sensing platform, which then scans and identifies the target task area to generate secondary alarm clues, including: The coordinates of the primary anomaly patches are spatially matched with the deployment locations and effective monitoring radii of each space-based remote sensing platform. If the primary anomaly patch falls within the coverage area of ​​the space-based platform, its coordinates are pushed as the target area to the corresponding space-based remote sensing platform. Specifically: The coordinates of the primary clue (x p ,y p ) and the deployment locations P of each airborne remote sensing system i and effective monitoring radius r a Perform spatial matching: if ||(x p ,y p )-P i ||≤r a Then the clue will be pushed to the corresponding space-based remote sensing A. i When the same clue falls within the overlapping coverage area of ​​multiple airborne platforms, the one closest to it (i) is selected. =argmin i d(Q p A i ()) is the primary execution platform, with the other platforms serving as alternatives.

[0075] After receiving the target area, the airborne remote sensing system adjusts its orientation to align with the target area, initiates high-frequency staring scanning, and uses a deep learning target detection model to identify, lock onto, and track the behavior of engineering vehicles and human activity targets, generating a secondary alarm clue that includes the target type and movement trajectory.

[0076] Furthermore, for first-level anomaly patches that do not fall within the monitoring coverage area of ​​the space-based remote sensing platform, i.e., for all i, d(Q) p A i )>r a If the airborne platform is unavailable due to weather obstruction, equipment maintenance, or other reasons, the system will execute an alternative verification and scheduling strategy. This includes dispatching aerial drones or mobile patrol and monitoring vehicles for on-site verification, including: Retrieve the deployment coordinates B of all aerial drone base stations j Based on the current availability status, spatial matching is performed between the coordinates of the first-level anomaly patches and the straight-line distances to each UAV base station:

[0077] like (r) uIf the clue is used as the drone monitoring path, it will be included in the drone candidate task list, automatically triggering on-demand drone route planning and on-site verification. Retrieve the current location P of all patrol and monitoring vehicles m (t) and the planned patrol route, calculate the shortest distance from the coordinates of the first-level anomaly patch to the current patrol route of the patrol monitoring vehicle:

[0078] like (w) m If the effective sweeping width of the patrol vehicle is sufficient, the clue will be sent to the corresponding patrol vehicle and incorporated into the daily patrol mission sequence. The patrol vehicle will then travel along the predetermined route to the target area for on-site verification and evidence collection.

[0079] Furthermore, for cases that fall within the drone's monitoring path and are also within the effective sweeping width of the patrol vehicle, the drone base station will be prioritized for deployment.

[0080] For high-priority leads, the system can simultaneously dispatch multiple platforms for coordinated response. For example, when a lead falls within the reach of a drone but requires a rapid response, the drone is triggered first; when roadside evidence collection or on-site law enforcement is needed, patrol vehicles are dispatched simultaneously; the verification results of both are integrated on the same platform to form a complete chain of on-site evidence.

[0081] In another embodiment, to prevent monitoring blind spots, a heat map is constructed based on the spatial distribution of historical anomalies, and the deployment of ground-based monitoring equipment is determined according to the heat value of new event locations in the heat map, including: Using the monitored area as the spatial scope, statistics are compiled for a window T over a past period. history Locations of all verified anomalies within (e.g., 12 months) (x e ,y e Historical anomaly spatial heatmaps are constructed using kernel density estimation. For the newly verified anomaly Q new Query the heat value (Heat(x)) of the anomaly point on the heat map. new ,y new ); If the heat value Heat(x) new ,y new ) greater than or equal to the high heat threshold θ high This indicates that the location is in a hotspot area with repeated historical disturbances, and ground equipment should be deployed at the anomaly point; If the heat value Heat(x) new ,y new Heat(x) is at the high heat threshold new ,y new ) and low heat threshold θ lowIn between, a weighted assessment combining anomaly type and ecological sensitivity is used to determine whether to deploy ground equipment; If the heat value Heat(x) new ,y new Less than the low heat threshold θ low This will be included in the next round of periodic evaluation.

[0082] Once the deployment conditions are met, the spatiotemporal coverage gap analysis method is used to identify existing monitoring blind spots around the verified locations, determine the optimal deployment locations for ground equipment (infrared cameras, video surveillance, etc.), and incorporate them into the collaborative monitoring platform for unified management in a "long-term duty" mode.

[0083] Furthermore, this invention establishes an intelligent scheduling decision-making method based on a triple mapping of monitoring scenario, target object, and platform capabilities. First, it defines three typical business application scenarios and their differentiated monitoring requirements; second, it constructs a unified multi-objective matching optimization model, transforming task allocation into a solvable 0-1 integer programming problem; finally, it realizes the automatic generation and cross-platform push of task instructions, supporting independent execution on a single platform or collaborative linkage across multiple platforms.

[0084] Based on the national ecological protection and restoration supervision needs, the "five basic" collaborative monitoring business application currently supports three major categories of typical scenarios. The specific monitoring objects, timeliness requirements, and accuracy requirements for each scenario are as follows.

[0085] Regional ecological quality monitoring: The monitoring targets include regional-scale ecological parameters such as ecosystem type, vegetation cover (FVC), net primary productivity (NPP), and leaf area index (LAI). Full coverage and regular updates are required, with a spatial resolution of 10m to 250m and a time period of ten days to annually. In this scenario, space-based satellites are responsible for acquiring baseline data across the entire region, airborne remote sensing and UAVs provide high-precision verification data in key areas, and ground equipment provides continuous, fixed-point true values. High-precision regional assessment is achieved through cross-scale push-up.

[0086] Biodiversity conservation: Monitoring targets include wild species types, population dynamics, and habitat quality. Real-time detection, identification, and tracking of key species of concern are required, with spatial resolution ranging from sub-meter to centimeter. In this scenario, airborne remote sensing is used for real-time monitoring and tracking of species activities in key areas; drones perform population statistics along scheduled routes; patrol vehicles collect species information along roadsides during their patrols; and ground-based infrared cameras and acoustic signature devices conduct long-term, fixed-point monitoring.

[0087] Human interference activity control: Monitoring targets include illegal mining, illegal construction, deforestation for cultivation, solid waste dumping, and other human activities. The requirement is to detect anomalies within the shortest possible timeframe and complete verification and evidence collection as quickly as possible, with spatial resolution ranging from meter-level for wide-area screening to centimeter-level for on-site evidence collection. In this scenario, space-based satellites perform comprehensive change detection, airborne remote sensing conducts daily high-frequency scanning and target locking of key areas, drones and patrol vehicles collaborate to complete precise verification, and ground equipment provides 24 / 7 monitoring of high-risk locations.

[0088] Establish a collaborative ecological supervision information platform based on five fundamental elements to achieve automatic generation, intelligent scheduling, and comprehensive analysis and display of monitoring task instructions. The overall system composition is as follows: Figure 4 As shown.

[0089] (1) Automatic generation of task instructions and multi-mode collaborative execution

[0090] For each pair The allocation results are used to automatically generate a standardized task instruction package, which includes the target coordinates (x, y, y). p ,y p The data acquisition specifications include spatial range, monitoring object type and task type, priority level, recommended execution time window, and data acquisition specifications (resolution, band, coverage times, etc.).

[0091] ① Single-platform execution mode: When a task is assigned to only one platform, the instruction is directly pushed to that platform, which then completes the monitoring task independently.

[0092] ② Multi-platform collaborative execution mode: When a task is assigned to multiple platforms, the system generates a collaborative timing scheme based on the arrival time and data acquisition capabilities of each platform.

[0093] ③ Dynamic rescheduling: When the platform status changes during execution (such as equipment failure, weather deterioration, or insertion of higher priority tasks), the platform automatically triggers the rescheduling process, re-solves the matching model for unexecuted tasks, updates the allocation scheme, and pushes update instructions.

[0094] (2) Scheduling strategies for various business application scenarios

[0095] Based on relevant national monitoring requirements and the monitoring capabilities of each platform, the specific scheduling strategies for each business application scenario are shown in Table 3.

[0096] Table 3 Monitoring Paths and Scheduling Strategies for Each Scenario in the "Five Basics" System

[0097] On the other hand, the present invention also provides a system for implementing the above-mentioned "five-based" collaborative ecological remote sensing monitoring method, such as... Figure 3The system includes: multiple monitoring platforms, a communication module, and a management platform. The management platform is connected to the monitoring platforms through the communication module and is used to obtain the capability parameters and monitoring data of each monitoring platform. The management platform is also used to construct the optimal monitoring spatial resolution that the "five-base" collaborative monitoring system (sky-air-tower-vehicle-ground) can provide at any spatiotemporal coordinate based on the acquired capability parameters of each monitoring platform; determine the monitoring platform for the multi-level monitoring process based on the monitoring capability function; and determine the optimal allocation scheme between the monitoring tasks to be assigned and the available execution platforms based on the multi-objective matching optimization model during the execution of the multi-level monitoring process.

[0098] 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 apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0099] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A "five-based" collaborative ecological remote sensing monitoring method, characterized in that, include: Construct a collaborative monitoring network integrating space, air, tower, vehicle, and ground systems; Establish a demand-driven "five-based" monitoring platform adaptation decision model to determine the optimal combination of monitoring platforms that meets actual needs; A multi-task target matching optimization model is constructed to determine the optimal allocation scheme between the monitoring tasks to be assigned and the available execution platforms, and a closed-loop monitoring system is constructed, which includes wide-area screening, key target locking, precise verification, and blind spot verification.

2. The "five-based" collaborative ecological remote sensing monitoring method according to claim 1, characterized in that, Establish a demand-driven "five-based" monitoring platform adaptation decision model to determine the optimal combination of monitoring platforms that meets actual needs, specifically including: Construct a monitoring demand vector and a platform capability benchmark set; For each type of platform, calculate the spatial coverage matching degree, resolution matching degree, and timeliness matching degree separately; The overall matching degree of each platform is calculated based on spatial coverage matching degree, resolution matching degree, and timeliness matching degree. When the overall matching degree of the platform is greater than or equal to the preset value, the platform is selected to perform the monitoring task. When the overall matching degree of all platforms is less than the preset value, the Monte Carlo random combination optimization method is used to randomly combine and evaluate the platform dimensional capabilities to select the optimal platform combination.

3. The "five-based" collaborative ecological remote sensing monitoring method according to claim 1, characterized in that, Construct a multi-task objective matching optimization model to determine the optimal allocation scheme between monitoring tasks to be assigned and available execution platforms, specifically including: Configure weights based on the type of business scenario; A comprehensive score is calculated on the execution capability of each task to be assigned, based on the weights configured according to the business scenario type. Establish a multi-objective matching optimization model: And satisfy the condition that each task can be assigned to at most one platform, and each platform can execute at most K tasks simultaneously. q The constraint that the total execution time of each task and the platform does not exceed the available time window, where x pq For decision variables; Solve the multi-objective matching optimization model and output the optimal allocation scheme.

4. The "five-based" collaborative ecological remote sensing monitoring method according to claim 1, characterized in that, The closed-loop monitoring system, which includes broad-area screening, targeted identification, precise verification, and blind spot validation, specifically includes: The target area is detected by space-based satellites according to the first preset cycle. Based on the change detection results, the first-level anomaly patches and the spatial location information corresponding to the first-level anomaly patches are extracted. The spatial location of the primary anomaly patch is spatially matched with the monitoring coverage area of ​​the airborne remote sensing platform. The primary anomaly patches falling within the monitoring coverage area are pushed to the corresponding airborne remote sensing platform as target tasks. The airborne remote sensing platform then scans and identifies the target task area to generate secondary alarm clues. For Level 1 abnormal patches that are not within the monitoring coverage of the airborne remote sensing platform, dispatch aerial drones or mobile patrol and monitoring vehicles to conduct on-site verification. Heat maps are constructed by combining the spatial distribution of historical anomalous events, and ground monitoring equipment is deployed in hotspots of historical anomalous events.

5. A system for a "five-based" collaborative ecological remote sensing monitoring method as described in any one of claims 1-4, characterized in that, include: The system includes multiple monitoring platforms, a communication module, and a management platform. The management platform is connected to the monitoring platforms via the communication module to acquire monitoring data from each monitoring platform. The management platform is also used to establish a demand-driven "five-based" monitoring platform adaptation decision model based on the capability parameters of each monitoring platform, and determine the optimal combination of monitoring platforms that meets actual needs; during the execution of multi-level monitoring processes, a multi-task target matching optimization model is constructed to determine the optimal allocation scheme between the monitoring tasks to be assigned and the available execution platforms, forming a closed-loop monitoring system of wide-area screening, key focus locking, precise verification, and blind spot verification.

Images