Underway greenhouse gas monitoring system based on autonomous exploration

Abstract

The invention relates to a navigation type greenhouse gas monitoring system based on autonomous exploration. The navigation type greenhouse gas monitoring system comprises an unmanned ship body, an autonomous sensing and dynamic planning module integrated on the unmanned ship body and an integrated greenhouse gas collecting and monitoring module. The integrated greenhouse gas collecting and monitoring module is used for working in a sailing type continuous pre-sampling mode to obtain real-time pre-sampling data when the unmanned ship body carries out autonomous cruise scanning in a water area to be detected, and is also used for obtaining real-time pre-sampling data when the unmanned ship body arrives at any optimized sampling point according to the optimal sailing path. Automatically switching to a fixed-point closed accurate measurement mode for measurement; the autonomous sensing and dynamic planning module is used for generating dynamic space distribution information of greenhouse gas based on the real-time pre-sampling data, acquiring environment obstacle information in real time, and generating an optimized sampling point position sequence and an optimal navigation path for the unmanned ship body to arrive at each optimized sampling point based on the environment obstacle information and the dynamic space distribution information; according to the invention, the monitoring efficiency and the monitoring accuracy are improved.

Description

Technical Field

[0001] This application relates to the field of environmental monitoring technology, and in particular to a mobile greenhouse gas monitoring system based on autonomous exploration. Background Technology

[0002] Greenhouse gas emissions and escapes are among the core drivers of global climate change. The escape of greenhouse gases from the water-air interface is a crucial component of the global carbon cycle, and its accurate monitoring plays a vital supporting role in climate change research and emission reduction policy formulation. With the upgrading of environmental monitoring needs, the monitoring of greenhouse gas escapes from the water-air interface is gradually developing towards automation, intelligence, and large-scale coverage. However, existing monitoring technologies still struggle to meet the requirements of high precision and high efficiency, and many problems urgently need to be addressed.

[0003] Currently, greenhouse gas monitoring at the water-air interface in water bodies mainly adopts the traditional "preset path + fixed location" approach. The core of this approach is to manually determine sampling points within the monitoring area through preliminary surveys. After pre-setting a navigation path, monitoring equipment or a simple unmanned surface vessel (USV) platform is manually operated to reach the fixed locations along the pre-set route to complete sampling and monitoring. Because the sampling point setting in this model relies entirely on manual pre-setting, it cannot be dynamically adjusted according to the actual distribution characteristics of dissolved greenhouse gas concentrations in the water body. Therefore, it is prone to problems such as missed detections in high-value anomaly areas and redundant sampling in uniformly distributed areas. This results in poor representativeness of the monitoring data, making it difficult to truly reflect the overall situation of regional emissions and meet the precise monitoring needs of complex water environments. Consequently, it cannot provide high-quality monitoring data support for climate change research and emission reduction policy formulation.

[0004] Therefore, there is an urgent need for a mobile greenhouse gas monitoring system based on autonomous exploration. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] In view of the above-mentioned shortcomings and deficiencies of the existing technology, this application provides a mobile greenhouse gas monitoring system based on autonomous exploration, which solves the technical problems of low monitoring efficiency, poor data representativeness and environmental adaptability of the existing mobile greenhouse gas monitoring technology.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, the main technical solutions adopted in this application include:

[0009] This application provides a mobile greenhouse gas monitoring system based on autonomous exploration, comprising:

[0010] The unmanned vessel body, the autonomous perception and dynamic planning module integrated into the unmanned vessel body, and the integrated greenhouse gas collection and monitoring module;

[0011] The integrated greenhouse gas acquisition and monitoring module is used to acquire real-time pre-sampling data of greenhouse gases in the water-air interface by operating in a continuous pre-sampling mode while the unmanned vessel is autonomously cruising and scanning the water area to be measured.

[0012] The autonomous perception and dynamic planning module is used to generate dynamic spatial distribution information of greenhouse gases in the water-air interface based on the real-time pre-sampling data. It is also used to acquire environmental obstacle information in real time. Based on the environmental obstacle information and the dynamic spatial distribution information, it dynamically generates an optimized sampling point sequence and the optimal navigation path for the unmanned vessel to reach each optimized sampling point in real time.

[0013] The integrated greenhouse gas acquisition and monitoring module is also used to automatically switch to a fixed-point closed-loop precision measurement mode when the unmanned vessel arrives at any of the optimized sampling points according to the optimal navigation path, and to measure the greenhouse gases released at the water-air interface.

[0014] Among them, the generation of the optimized sampling point sequence and the optimal navigation path, as well as the triggering of the fixed-point closed-loop precision measurement mode, are all related to the real-time pre-sampling data obtained by the mobile continuous pre-sampling mode.

[0015] Optionally, in some embodiments of this application, the autonomous perception and dynamic planning module includes:

[0016] An environment and gas correlation sensing unit is used to generate dynamic spatial distribution information of greenhouse gases in the water-air interface based on the real-time pre-sampling data, and to acquire environmental obstacle information in real time. Based on the environmental obstacle information and the dynamic spatial distribution information, an environment-gas correlation map is generated.

[0017] The sampling point decision unit is used to identify the spatial distribution characteristics of greenhouse gas concentrations based on the environment-gas correlation map, and generate an optimized sampling point sequence containing point coordinates and priorities according to a preset monitoring strategy.

[0018] An adaptive path planning unit is used to generate an optimal navigation path based on the optimized sampling point sequence, the navigation performance parameters of the unmanned vessel, and environmental obstacle information.

[0019] Optionally, in some embodiments of this application, the environment and gas correlation sensing unit includes:

[0020] The spatiotemporal alignment and fusion subunit is used to perform spatiotemporal alignment on the real-time presampled data and the environmental obstacle information based on a unified time and space reference system, and generate spatiotemporally aligned real-time presampled data and environmental obstacle information.

[0021] The spatial distribution field reconstruction subunit is used to process the spatiotemporally aligned real-time pre-sampled data through a spatial interpolation algorithm to generate a spatially continuously distributed enhanced gas concentration distribution field as dynamic spatial distribution information of greenhouse gases in the water-air interface.

[0022] The multidimensional correlation map generation subunit is used to spatially overlay and correlate the enhanced gas concentration distribution field with the spatiotemporally aligned environmental obstacle information to generate an environment-gas correlation map that characterizes the coupling relationship between navigable areas, environmental features and greenhouse gas concentration distribution.

[0023] Optionally, in some embodiments of this application, the sampling point decision unit includes:

[0024] The distribution feature extraction subunit is used to analyze the environment-gas correlation map, extract high concentration clusters and gradient abrupt change zones as spatial distribution features, and divide the water area to be measured into multiple independent candidate monitoring areas based on the spatial boundaries of the extracted spatial distribution features.

[0025] The monitoring priority ranking subunit is used to configure basic weights for different types of spatial distribution features based on a preset evaluation system, configure sub-item weights for each candidate monitoring area, generate a comprehensive priority score for each area through weighted calculation, and generate a monitoring priority sequence based on the score.

[0026] The initial point generation subunit is used to generate an initial sampling point set according to the monitoring priority sequence and the pre-set layout rules. The pre-set layout rules are: to densify the points in high-priority areas, to linearly distribute the points along the gradient change zone in medium-priority areas, and to distribute the points in a uniform grid in low-priority areas.

[0027] The point optimization and filtering subunit is used to simplify and adjust based on the endurance constraints of the unmanned vessel body, and obtain an optimized sampling point sequence containing point coordinates and priorities.

[0028] Optionally, in some embodiments of this application, the preset evaluation system in the monitoring priority ranking subunit includes:

[0029] The feature type weight library pre-stores the basic weights corresponding to different spatial distribution feature types. Among them, the weights configured for high concentration clusters are greater than those configured for gradient abrupt change zones.

[0030] Regional attribute quantification rules are used to calculate a set of quantifiable attribute parameters for each candidate monitoring region, wherein the attribute parameters include at least the region area.

[0031] Optionally, in some embodiments of this application, the adaptive path planning unit includes:

[0032] The global path generation subunit is used to generate a cost map based on the optimized sampling point sequence containing point coordinates and priorities, the navigation performance parameters of the unmanned vessel, and the environmental obstacle information.

[0033] The path optimization subunit is used on the cost map to calculate the optimal navigation path by using a motion planning algorithm, with the goal of minimizing the overall navigation cost and maximizing the attractiveness to high-priority points, and to generate the optimal navigation path connecting the current unmanned vessel position with subsequent points in the optimized sampling point sequence.

[0034] Optionally, in some embodiments of this application, the global path generation subunit is specifically used for:

[0035] An initial grid map is constructed, and the base value of each cell or node is determined by the environmental obstacle information, wherein areas occupied by obstacles are assigned a value of being prohibited from passage.

[0036] Based on the optimized sampling point sequence containing point coordinates and priorities, a corresponding attraction field is generated for each optimized sampling point on the initial map; the attraction field is then superimposed on the base cost value of the initial map to generate the cost map.

[0037] Optionally, in some embodiments of this application, the path optimization subunit is specifically used for:

[0038] A comprehensive objective function for path optimization is constructed. On the cost map, the A* algorithm is run to find candidate paths that optimize the comprehensive objective function. The candidate paths that optimize the comprehensive objective function are then smoothed to obtain the optimal navigation route.

[0039] Optionally, in some embodiments of this application, the comprehensive objective function consists of two parts: the total navigation cost of the path and the total monitoring value gain of the path.

[0040] The total navigation cost of the route is calculated by accumulating the basic travel costs corresponding to each location the route passes through on the cost map.

[0041] The total monitoring value of the path is calculated cumulatively based on the proximity of the path to each point in the optimized sampling point sequence and the priority of each point.

[0042] The optimization objective of the comprehensive objective function is to minimize the total cost of navigation and maximize the total monitoring value, while satisfying the constraints of navigation endurance.

[0043] Optionally, in some embodiments of this application, the system further includes:

[0044] The task management and storage module integrated into the unmanned vessel body is used to receive and store the environmental-gas correlation map, optimized sampling point sequence and optimal navigation path from the autonomous perception and dynamic planning module, as well as all pre-sampling data and fixed-point precise measurement data from the integrated greenhouse gas collection and monitoring module.

[0045] (III) Beneficial Effects

[0046] The beneficial effects of this application are that the mobile greenhouse gas monitoring system based on autonomous exploration, by adopting a collaborative working mode of mobile continuous pre-sampling and fixed-point closed-loop precise measurement, and by associating real-time pre-sampling data with optimized sampling point sequences, optimal navigation path generation, and precise measurement mode triggering, combined with the autonomous perception and dynamic planning capabilities of unmanned vessels, can dynamically and accurately grasp the spatial distribution characteristics of greenhouse gases escaping at the water-air interface compared to existing technologies. This improves the targeting of sampling points and the accuracy of measurement data, while also increasing the automation and operational efficiency of greenhouse gas monitoring in water areas, reducing the operational difficulty and labor costs of monitoring complex water areas, and achieving the technical effect of efficient, accurate, and autonomous monitoring of greenhouse gases escaping at the water-air interface in water areas. Attached Figure Description

[0047] Figure 1 This is a structural diagram of a mobile greenhouse gas monitoring system based on autonomous exploration, according to an embodiment of this application.

[0048] Figure 2 This is a structural diagram of the autonomous sensing and dynamic planning module of a mobile greenhouse gas monitoring system based on autonomous exploration, according to an embodiment of this application.

[0049] Figure 3 This is a schematic diagram of the cost map in the global path generation subunit in this embodiment of the application. Detailed Implementation

[0050] To better explain and facilitate understanding of this application, the following detailed description of the application is provided in conjunction with the accompanying drawings and specific embodiments.

[0051] In the context of addressing global climate change, monitoring of fugitive greenhouse gases (such as methane and carbon dioxide) at the water-air interface in aquatic waters has become a crucial aspect of ecological and environmental assessment. Currently, greenhouse gas monitoring in aquatic waters primarily employs manual fixed-point sampling or fixed-path patrol monitoring, which has significant technical limitations: manual sampling is inefficient and has limited coverage, failing to reflect the spatial heterogeneity of greenhouse gas distribution; fixed-path patrols cannot dynamically adjust sampling points based on real-time monitoring data, easily leading to omissions of key high-emission areas and insufficient obstacle avoidance capabilities in complex waters; furthermore, pre-sampling and precise measurement in existing monitoring systems are often independent processes with poor data correlation, making it difficult to ensure a synergistic improvement in monitoring accuracy and efficiency. Therefore, how to achieve efficient, autonomous, and accurate monitoring of fugitive greenhouse gases at the water-air interface in aquatic waters, balancing broad coverage and monitoring accuracy while reducing the operational difficulty in complex waters, has become an urgent technical challenge.

[0052] Based on this, this application proposes a mobile greenhouse gas monitoring system based on autonomous exploration. Through the collaborative design of mobile continuous pre-sampling and fixed-point closed-loop precise measurement, combined with real-time data-driven dynamic path planning, it effectively solves the above-mentioned technical pain points and realizes autonomous, precise and efficient monitoring of greenhouse gases in water areas, providing reliable data support for ecological and environmental assessment.

[0053] To better understand the above technical solutions, exemplary embodiments of this application will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application can be understood more clearly and thoroughly, and that the scope of this application can be fully conveyed to those skilled in the art.

[0054] A* Algorithm: A classic heuristic path search algorithm. Its core is to guide the search direction through an evaluation function of "actual cost + estimated cost". It considers both the actual travel cost from the starting point to the current node and the estimated potential cost from the current node to the target point. It prioritizes exploring the path with the lowest total cost, which can significantly reduce invalid searches while ensuring that the optimal path is found. It is widely applicable to dynamic path planning scenarios of unmanned ships, drones and other equipment, and is especially suitable for the core requirement of generating the optimal travel path in this application by combining environmental obstacles and sampling point requirements.

[0055] Figure 1 This is a schematic diagram of a mobile greenhouse gas monitoring system based on autonomous exploration, according to an embodiment of this application. Figure 1 As shown, the mobile greenhouse gas monitoring system includes:

[0056] The unmanned vessel body, the autonomous perception and dynamic planning module integrated into the unmanned vessel body, and the integrated greenhouse gas collection and monitoring module;

[0057] The integrated greenhouse gas acquisition and monitoring module is used to acquire real-time pre-sampling data of greenhouse gases in the water-air interface by operating in a continuous pre-sampling mode while the unmanned vessel is autonomously cruising and scanning the water area to be measured.

[0058] The autonomous perception and dynamic planning module is used to generate dynamic spatial distribution information of greenhouse gases in the water-air interface based on real-time pre-sampling data. It is also used to acquire environmental obstacle information in real time. Based on the environmental obstacle information and dynamic spatial distribution information, it dynamically generates an optimized sampling point sequence and the optimal navigation path for the unmanned vessel to reach each optimized sampling point in real time.

[0059] The integrated greenhouse gas collection and monitoring module is also used to automatically switch to a fixed-point closed-loop precision measurement mode when the unmanned vessel arrives at any optimized sampling point according to the optimal navigation path, in order to measure the greenhouse gases released at the water-air interface.

[0060] Among them, the generation of optimized sampling point sequence and optimal navigation path, as well as the triggering of fixed-point closed-loop precision measurement mode, are all related to the real-time pre-sampling data obtained by the mobile continuous pre-sampling mode.

[0061] In the specific implementation process, the unmanned vessel is a small intelligent surface navigation platform adapted to autonomous navigation and fixed-point measurement in the waters to be measured. It is equipped with a dual-propulsion power system and attitude stabilization unit, which can achieve low-speed precise navigation and parking positioning. The hull has an independent power supply module and data transmission unit built in, and reserves standardized power supply interfaces and data communication interfaces for the autonomous perception and dynamic planning module and the integrated greenhouse gas collection and monitoring module. This enables the stable integration of each module, power supply and real-time data interaction. At the same time, it has the water adaptability to resist wind and waves and prevent grounding, ensuring the stability of the navigation and measurement process.

[0062] Furthermore, the autonomous perception and dynamic planning module integrates multiple sensors, including, for example, lidar for high-precision near-range obstacle detection, a stereo camera for visual recognition, millimeter-wave radar suitable for harsh weather conditions, and a GNSS integrated navigation system providing high-precision pose information. It also integrates a high-performance embedded computing unit, enabling environmental perception, synchronous positioning and map building, and real-time path planning. The integrated greenhouse gas acquisition and monitoring module consists of a gas acquisition subsystem composed of a liftable floating gas sampling chamber, a gas pump, solenoid valves, and a flow controller, as well as a greenhouse gas analyzer based on high-precision optical cavity ring-down spectroscopy technology. The sampling chamber is raised and lowered by a servo motor, thereby automatically and intelligently switching between two modes: "mobile continuous pre-sampling" and "fixed-point closed precision measurement".

[0063] The aforementioned autonomous exploration-based mobile greenhouse gas monitoring system achieves precise navigation and positioning through standardized integrated design of the unmanned vessel and real-time data interaction between various modules, combined with dual thrusters and attitude stabilization units. In addition, the servo motor controls the sampling cabin to automatically switch monitoring modes, which greatly improves the efficiency of automated monitoring and reduces labor costs.

[0064] Specifically, in this embodiment, the autonomous perception and dynamic planning module includes:

[0065] An environment and gas correlation sensing unit is used to generate dynamic spatial distribution information of greenhouse gases in the water-air interface based on the real-time pre-sampling data, and to acquire environmental obstacle information in real time. Based on the environmental obstacle information and the dynamic spatial distribution information, an environment-gas correlation map is generated.

[0066] The environmental and gas correlation sensing unit includes:

[0067] The spatiotemporal alignment and fusion subunit is used to perform spatiotemporal alignment on the real-time presampled data and the environmental obstacle information based on a unified time and space reference system, and generate spatiotemporally aligned real-time presampled data and environmental obstacle information.

[0068] Specifically, this subunit incorporates a high-precision clock source to timestamp all input data uniformly. At the spatial level, it utilizes the pose information provided by the unmanned surface vessel's (USV) GNSS navigation system as a spatial reference frame. Through pre-calibrated sensor extrinsic parameters (position and attitude offset), the acquisition locations of lidar point clouds, visual detection results, millimeter-wave radar targets, and real-time pre-sampled data are all transformed into a navigation coordinate system centered on the USV itself. This process ensures that data points from any sensor are strictly aligned in both spatial and temporal dimensions. Ultimately, this subunit outputs two parallel and spatiotemporally synchronized data streams: spatiotemporally aligned real-time pre-sampled data and spatiotemporally aligned environmental obstacle information.

[0069] The spatial distribution field reconstruction subunit is used to process the spatiotemporally aligned real-time pre-sampled data through a spatial interpolation algorithm to generate a spatially continuously distributed enhanced gas concentration distribution field as dynamic spatial distribution information of greenhouse gases in the water-air interface.

[0070] In the specific implementation process, the spatial distribution field reconstruction subunit receives a real-time pre-sampled data sequence after spatiotemporal alignment. This data is spatially non-uniformly distributed. The spatial distribution field reconstruction subunit incorporates or invokes geospatial interpolation algorithms, such as the statistically based Kriging interpolation method. This method first calculates the spatial variability function between known sampling points to quantify the statistical relationship of concentration variation with distance, and then uses this function to perform optimal unbiased estimation of the concentration in unsampled areas. By traversing the regular grid of the entire water body to be measured, the algorithm assigns a predicted concentration value to each grid cell, thereby generating a spatially continuous, digital enhanced gas concentration distribution field. This distribution field not only visually displays high and low concentration areas in the form of a heatmap, but it is also high-precision raster data itself, which can be directly used for spatial analysis. With the continuous input of new data points, the distribution field is dynamically refreshed in the form of a sliding window or incremental updates, truly reflecting the spatial dynamic changes of gas dispersion.

[0071] The multidimensional correlation map generation subunit is used to spatially overlay and correlate the enhanced gas concentration distribution field with the spatiotemporally aligned environmental obstacle information to generate an environment-gas correlation map that characterizes the coupling relationship between navigable areas, environmental features and greenhouse gas concentration distribution.

[0072] Specifically, the workflow of the multidimensional correlation map generation subunit is as follows: First, data import and preprocessing: Import the enhanced gas concentration distribution field generated by the spatial distribution field reconstruction subunit and the spatiotemporally aligned environmental obstacle information, respectively. Classify and label the environmental obstacle information, such as labeling it as "non-navigable - shallow waters," "navigable - open waters," and "interference source - dense aquatic plant area." Second, spatial overlay and fusion: Using Geographic Information System (GIS) spatial overlay technology, overlay the enhanced gas concentration distribution field and the labeled environmental obstacle information in the same spatial coordinate system to form a layer containing concentration information and environmental... The first step involves a composite data layer of environmental information; the second step is correlation analysis, which uses spatial correlation rule mining algorithms to analyze the correspondence between different greenhouse gas concentration levels and various environmental characteristics, while also marking the limitation range of non-navigable areas on sampling operations; the third step is map generation, which visualizes the correlation analysis results and generates an environmental-gas correlation map containing a three-dimensional coupling relationship of "navigable area range - environmental obstacle type - greenhouse gas concentration gradient". Different colors are used in the map to distinguish concentration levels and different symbols are used to mark obstacle types, providing direct and accurate decision-making basis for subsequent optimization of sampling point selection and optimal navigation path planning.

[0073] Among them, the spatial association rule mining algorithm was used to analyze the correspondence between different greenhouse gas concentration levels and various environmental features. At the same time, the spatial association rule mining algorithm in the restricted area of ​​non-navigable areas for sampling operations was identified as the spatial Apriori algorithm. Combined with the composite data layer formed by spatial overlay and fusion (containing labeled environmental obstacle information and gas concentration distribution information) mentioned above, the specific analysis process of this algorithm can be summarized as follows: the two types of core information in the composite layer are transformed into discretized spatial attributes that the algorithm can process. For greenhouse gas concentration, the continuous enhanced gas concentration distribution field needs to be discretized into 3-5 levels (such as high concentration area, medium concentration area, and low concentration area) according to a preset threshold, and each level corresponds to a set of spatial regions. For environmental features, the labeled environmental obstacle information (such as shallow water, open water, dense aquatic plant area, etc.) is divided into discrete spatial regions according to type, and each environmental type also corresponds to a set of spatial regions. Subsequently, spatial relationships and candidate itemsets are defined, with "spatial coexistence" as the core relationship, meaning that two types of spatial regions overlap or are adjacent. Discretized combinations of "concentration level - environmental characteristics" are then used as candidate itemsets, such as {shallow areas, high-concentration areas}, {dense aquatic plant areas, high-concentration areas}, etc. Next, iterative screening is used to identify "concentration level - environmental characteristics" combinations with a high probability of coexistence. Specifically, a "minimum support" threshold is first set, representing the lower limit of the frequency of a candidate itemset appearing in the entire tested water area. Then, all candidate itemsets are traversed, and their spatial frequency is counted. Itemets with a frequency not lower than the minimum support are retained as frequent itemsets. Based on this, association rules are generated and filtered. Each frequent itemset is broken down into "premise-conclusion" rule forms (e.g., {shoal} → {high concentration area}). A "minimum confidence" threshold (e.g., 70%) is then set to measure the reliability of the rules. The confidence level is calculated by dividing the "frequency of both premise and conclusion existing" by the "frequency of premise existing alone". Finally, rules with a confidence level not lower than the minimum confidence level are retained, which are the discovered effective associations. The filtered association rules are then mapped back to the spatial layer to clarify the coupling patterns between different concentration levels and environmental characteristics, providing a core basis for generating environment-gas association maps. For example, key information such as "shoal edges are high-concentration gas emission hotspots" can be accurately marked in the map.

[0074] The environmental and gas correlation sensing unit of this application deeply integrates discrete and heterogeneous raw data into an environmental-gas correlation map with clear spatial semantics through multi-level processing of spatiotemporal alignment, spatial interpolation, and correlation rule mining. It not only intuitively presents greenhouse gas distribution hotspots, but also profoundly reveals the coupling law between concentration fields and environmental elements. This upgrades path planning from obstacle avoidance navigation based on geometric information to active exploration based on spatial cognition, realizing an intelligent leap from "blind uniform sampling" to "precise targeted detection" in the monitoring process, significantly improving monitoring efficiency and the scientific value of the data.

[0075] The autonomous perception and dynamic programming module of this application also includes:

[0076] The sampling point decision unit is used to identify the spatial distribution characteristics of greenhouse gas concentrations based on the environment-gas correlation map, and generate an optimized sampling point sequence containing point coordinates and priorities according to the preset monitoring strategy.

[0077] The sampling point decision unit includes:

[0078] The distribution feature extraction subunit is used to analyze the environment-gas correlation spectrum, extract high concentration clusters and gradient abrupt change zones as spatial distribution features, and divide the water area to be measured into multiple independent candidate monitoring areas based on the spatial boundaries of the extracted spatial distribution features.

[0079] In the specific implementation process, the distribution feature extraction subunit serves as a fundamental link in the decision-making process. Its core task is to analyze and extract key spatial distribution features from the environment-gas correlation map. Specifically, this subunit first performs in-depth analysis of the enhanced gas concentration distribution layer in the map. Using the DBSCAN clustering algorithm, it identifies high-concentration clusters—precisely capturing continuously distributed high-concentration areas by setting preset cluster radius of 0.5-1m and minimum sample size of 3, avoiding misclassification of isolated high-concentration points as core monitoring areas. Simultaneously, the Sobel edge detection algorithm is used to extract concentration gradient abrupt change zones. By setting reasonable gradient thresholds to filter environmental noise interference, it accurately identifies boundary regions where concentrations change significantly, using both types of regions as core spatial distribution features. Based on this, the subunit conducts connectivity analysis based on the extracted spatial distribution feature boundaries. Simultaneously, it combines this with non-navigable areas marked in the map, such as shoals and floating debris accumulation areas, to eliminate the fragmenting effect of obstacle areas on feature regions. Ultimately, the water area to be monitored is divided into multiple independent candidate monitoring areas with clear boundaries, no overlap, and each bound to a corresponding feature type, providing a foundation for subsequent priority ranking.

[0080] The monitoring priority ranking subunit is used to configure basic weights for different types of spatial distribution characteristics based on a preset evaluation system, and to configure sub-item weights for each candidate monitoring area. It generates a comprehensive priority score for each area through weighted calculation and generates a monitoring priority sequence based on the score.

[0081] The pre-defined evaluation system in the monitoring priority ranking sub-unit includes:

[0082] The feature type weight library pre-stores the basic weights corresponding to different spatial distribution feature types. Among them, the weights configured for high concentration value clusters are greater than those configured for gradient abrupt change zones. For example, the basic weights for high concentration value clusters are set to 0.6-0.7, and the basic weights for gradient abrupt change zones are set to 0.3-0.4. The core monitoring direction is clarified by the difference in basic weights.

[0083] Regional attribute quantification rules are used to calculate a set of quantifiable attribute parameters for each candidate monitoring region. The attribute parameters include at least the region area.

[0084] Optionally, in addition to the explicitly required area, auxiliary attributes such as the average concentration within the area, the distance to the current position of the unmanned vessel, and the percentage of navigable areas can also be included. Each attribute parameter is converted into a sub-item weight in the range of 0-0.3 through a preset mapping rule. For example, the larger the area and the higher the average concentration, the higher the corresponding sub-item weight. Finally, this sub-unit obtains the comprehensive priority score for each candidate monitoring area through a weighted calculation method of "comprehensive priority score = basic weight × 0.7 + Σ (sub-item weight × corresponding coefficient)", and generates a three-level monitoring priority sequence of high, medium, and low according to the score from high to low.

[0085] The initial point generation sub-unit is used to generate an initial set of sampling points based on the monitoring priority sequence and the pre-set layout rules. The pre-set layout rules are: to densify the points in high-priority areas, to lay out the points linearly along the gradient abrupt change zone in medium-priority areas, and to lay out the points uniformly in the grid in low-priority areas.

[0086] Specifically, for example, for high-priority areas (mainly high-concentration clusters), a denser sampling strategy is adopted, with grid points spaced at 2-3m intervals to ensure complete coverage of the core and edge areas of high-concentration regions and accurately capture details of concentration changes. For medium-priority areas (mainly areas associated with gradient abrupt change zones), linear sampling is conducted along the extension direction of the gradient abrupt change zone, with a point spacing of 4-5m, focusing on covering key nodes with drastic concentration changes and accurately identifying patterns in concentration gradient changes. For low-priority areas (areas with uniformly distributed ordinary concentrations), a uniform grid spacing of 10-15m is used to control the number of sampling points while ensuring basic monitoring coverage and avoiding resource waste. It is worth noting that during the sampling process, this subunit simultaneously verifies the navigability of each initial sampling point, automatically eliminating points located in non-navigable areas based on obstacle markers in the environment-gas correlation map, ensuring the feasibility of the initial sampling point set.

[0087] The point optimization and filtering sub-unit is used to simplify and adjust the unmanned vessel's endurance constraints, and obtain an optimized sampling point sequence containing point coordinates and priorities.

[0088] This subunit first acquires core parameters of the unmanned surface vessel (USV), such as remaining range, energy consumption per unit distance, and current location. Combined with the distribution of points in the initial sampling set, it calculates the total voyage required to complete all sampling (based on the sum of the shortest paths between points). If the total voyage exceeds the USV's range threshold, a simplification and optimization process is initiated: the optimization priority follows the principle of "preserving high-value areas first, then adjusting medium- and low-value areas." Sampling points in high-priority areas are retained first, and only redundant points with excessively close spacing (e.g., less than 1.5m) are merged. Secondly, linear points in medium-priority areas are selectively removed, retaining only key nodes with abrupt concentration changes. Finally, uniform points in low-priority areas are sparsified, with the spacing between points appropriately increased. During the simplification process, this subunit strictly controls the process to ensure that each candidate monitoring area retains at least one core sampling point to avoid monitoring blind spots. The final output is an optimized sampling point sequence containing point coordinates, priority levels, and suggested monitoring order, providing accurate input for subsequent optimal USV navigation path planning.

[0089] In summary, the refined design of the sampling point decision unit has significant beneficial effects: First, through precise algorithm selection and connectivity analysis of the distribution feature extraction subunit, the core distribution characteristics of greenhouse gases are accurately identified and candidate areas are scientifically divided, effectively avoiding the omission of high-value monitoring areas or misjudgment of invalid areas, laying the foundation for subsequent accurate sampling. Second, the multi-dimensional quantitative evaluation system of the monitoring priority ranking subunit, through the synergistic weighting of basic weights and sub-item weights, clarifies the monitoring priority of each area, ensuring that high-emission and high-value areas are covered first, and enhancing the supporting value of monitoring data for emission assessment. Third, the differentiated deployment rules combined with navigability verification achieve precise point deployment of "high-priority encryption, medium-priority targeting, and low-priority balancing," ensuring comprehensive monitoring while avoiding resource waste caused by point redundancy. Fourth, the point optimization and screening subunit, combined with the simplification and adjustment of endurance constraints, makes the final sampling point sequence adaptable to the actual operation capabilities of unmanned vessels, taking into account both the scientific nature of monitoring and practical feasibility, effectively improving the overall efficiency and data reliability of water greenhouse gas monitoring, and reducing the cost of invalid navigation and sampling.

[0090] The autonomous perception and dynamic programming module of this application also includes:

[0091] The adaptive path planning unit is used to generate the optimal navigation path based on the optimized sampling point sequence, the navigation performance parameters of the unmanned vessel, and environmental obstacle information.

[0092] The adaptive path planning unit includes:

[0093] The global path generation subunit is used to generate a cost map based on an optimized sampling point sequence containing point coordinates and priorities, the navigation performance parameters of the unmanned vessel, and environmental obstacle information.

[0094] The global path generation sub-unit is specifically used for:

[0095] See Figure 3 An initial grid map is constructed, and the base value of each cell or node is determined by environmental obstacle information, where areas occupied by obstacles are assigned a value indicating that passage is prohibited.

[0096] like Figure 3 As shown, the cost of unobstructed open water is set to 1 (basic passage cost), the cost of semi-obstructed areas such as shoals and floating objects is set to 5-8 (increased passage cost), and non-navigable areas such as reefs and dikes are directly assigned an infinite cost (marked as no passage). Figure 3 The symbol “∞” is used to represent it.

[0097] Based on an optimized sampling point sequence containing point coordinates and priorities, a corresponding attraction field is generated for each optimized sampling point on the initial map, where the optimized sampling point is... Figure 3 The red origin is shown in the center, and concentric circles around each sampling point represent the diffusion effect of the attraction field; the attraction field is superimposed on the base cost value of the initial map to generate the cost map.

[0098] like Figure 3 As shown, high-priority sampling points are represented by gradient rings from dark red to light red, indicating an attraction intensity of 8-10; medium-priority points are represented by gradient rings from orange to yellow, indicating 4-6; and low-priority points are represented by gradient rings from light green to white, indicating 2-3. The attraction intensity decreases exponentially with distance (the closer to the sampling point, the stronger the attraction), ensuring that path planning prioritizes high-value locations. Subsequently, a cost map is generated by fusion, superimposing the attraction field cost of each sampling point with the base cost of the initial grid map to calculate the final cost map that integrates "environmental barrier cost" and "monitoring value attraction." The total cost of each grid cell in this map reflects both the ease of passage and the value weight of the area for monitoring operations.

[0099] The path optimization subunit is used to calculate the optimal navigation path on the cost map by using a motion planning algorithm, with the goal of minimizing the overall navigation cost and maximizing the attractiveness to high-priority points. This generates the optimal navigation path that connects the current unmanned vessel position with subsequent points in the optimized sampling point sequence.

[0100] The path optimization subunit is specifically used for:

[0101] Construct a comprehensive objective function for path optimization, run the A* algorithm on the cost map to find candidate paths that optimize the comprehensive objective function, and smooth the candidate paths that optimize the comprehensive objective function to obtain the optimal navigation route.

[0102] The overall objective function consists of two parts: the total navigation cost of the path and the total monitoring value gain of the path.

[0103] The total cost of the route is calculated by accumulating the basic travel costs corresponding to each location the route passes through on the cost map;

[0104] The total monitoring value of the path is calculated cumulatively based on the proximity of the path to each point in the optimized sampling point sequence and the priority of each point.

[0105] The optimization objective of the comprehensive objective function is to minimize the total cost of navigation and maximize the total monitoring value while satisfying the constraints of navigation endurance.

[0106] In the specific implementation process, the path optimization sub-unit, based on the aforementioned cost map, selects the optimal navigation path through scientific objective function construction and algorithmic calculation. Its specific implementation logic is as follows:

[0107] A comprehensive objective function for path optimization is constructed, which precisely balances navigation cost and monitoring value. It consists of two core components: "total navigation cost" and "total monitoring value benefit," with weighting coefficients adjusting their priorities. Based on the monitoring task requirements, the monitoring value weight is set to 0.6, and the navigation cost weight is set to 0.4. The total navigation cost is calculated by summing the total value corresponding to all grids traversed by the path. The total monitoring value benefit is calculated by weighting the path based on its proximity to each optimized sampling point (closer distance yields higher benefit), combined with the priority weight of each point. If the path precisely covers the sampling point coordinates, the full benefit is obtained; if the distance exceeds a preset threshold (e.g., 3m), the benefit is zero. Secondly, path optimization and selection are performed. This sub-unit runs the A* algorithm on the cost map to search for paths. The algorithm's heuristic function is Euclidean distance. Through a comprehensive evaluation of "actual navigation cost (total cost of the path already traveled) + estimated cost (straight-line distance cost from the current position to the target point)," it quickly identifies potential optimal candidate paths. Simultaneously, it automatically avoids prohibited areas with infinite cost values ​​to ensure path safety. Finally, path smoothing is performed. Since the initial candidate paths generated by the A* algorithm may have many inflection points, which do not conform to the physical navigation characteristics of the unmanned surface vessel (USV), a Bezier curve smoothing algorithm is used to optimize the candidate paths. This eliminates sharp inflection points and adjusts the path curvature, ensuring that the optimized path conforms to the USV's turning radius limitations, improving navigation stability and energy efficiency. After the above processes, the final output is the optimal navigation path connecting the current USV position with all subsequent points in the optimized sampling point sequence. The path information includes a detailed coordinate sequence, suggested speed, and turning node prompts.

[0108] The adaptive path planning unit in this application constructs a dynamic cost map that integrates the cost of environmental obstacles and the attractiveness of monitoring value. Based on a comprehensive objective function that weighs the cost of navigation and the benefits of monitoring, it performs intelligent path optimization. This transforms the unmanned vessel from a simple point-to-point navigation tool into an intelligent agent capable of making autonomous scientific judgments in complex aquatic environments. Ultimately, it achieves the dynamic generation of an optimal operational path that efficiently avoids obstacles and prioritizes coverage of key monitoring areas while ensuring navigation safety. This minimizes the path cost of monitoring operations and maximizes the data collection value of a single voyage.

[0109] The mobile greenhouse gas monitoring system based on autonomous exploration, as described in this application embodiment, also includes:

[0110] The mission management and storage module integrated into the unmanned vessel body is used to receive and store environmental-gas correlation maps, optimized sampling point sequences and optimal navigation paths from the autonomous perception and dynamic planning module, as well as all pre-sampling data and fixed-point precise measurement data from the integrated greenhouse gas collection and monitoring module.

[0111] This application presents a mobile greenhouse gas monitoring system based on autonomous exploration. By combining mobile continuous pre-sampling with fixed-point closed-loop precision measurement, and intelligently generating dynamic spatial distribution information, environment-gas correlation maps, optimized sampling point sequences, and optimal navigation paths based on real-time pre-sampling data, a complete autonomous closed-loop system of "perception-cognition-decision-action" is constructed. This system realizes intelligent and adaptive monitoring of greenhouse gas distribution in water areas from macroscopic scanning to key precision measurement. Thus, while significantly improving monitoring efficiency and data spatial representativeness, it effectively overcomes the technical bottlenecks of traditional methods, such as high cost, strong blindness, and difficulty in capturing spatial heterogeneity.

[0112] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0113] In this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0114] In this application, unless otherwise expressly specified and limited, "above" or "below" the second feature can mean that the first and second features are in direct contact, or that they are in indirect contact through an intermediate medium. Furthermore, "above," "on top of," and "over" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0115] In the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0116] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make modifications, alterations, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A mobile greenhouse gas monitoring system based on autonomous exploration, characterized in that, include: The unmanned vessel body, the autonomous perception and dynamic planning module integrated into the unmanned vessel body, and the integrated greenhouse gas collection and monitoring module; The integrated greenhouse gas acquisition and monitoring module is used to acquire real-time pre-sampling data of greenhouse gases in the water-air interface by operating in a continuous pre-sampling mode while the unmanned vessel is autonomously cruising and scanning the water area to be measured. The autonomous perception and dynamic planning module is used to generate dynamic spatial distribution information of greenhouse gases in the water-air interface based on the real-time pre-sampling data. It is also used to acquire environmental obstacle information in real time. Based on the environmental obstacle information and the dynamic spatial distribution information, it dynamically generates an optimized sampling point sequence and the optimal navigation path for the unmanned vessel to reach each optimized sampling point in real time. The integrated greenhouse gas acquisition and monitoring module is also used to automatically switch to a fixed-point closed-loop precision measurement mode when the unmanned vessel arrives at any of the optimized sampling points according to the optimal navigation path, and to measure the greenhouse gases released at the water-air interface. Among them, the generation of the optimized sampling point sequence and the optimal navigation path, as well as the triggering of the fixed-point closed-loop precision measurement mode, are all related to the real-time pre-sampling data obtained by the mobile continuous pre-sampling mode.

2. The mobile greenhouse gas monitoring system based on autonomous exploration according to claim 1, characterized in that, The autonomous perception and dynamic programming module includes: An environment and gas correlation sensing unit is used to generate dynamic spatial distribution information of greenhouse gases in the water-air interface based on the real-time pre-sampling data, and to acquire environmental obstacle information in real time. Based on the environmental obstacle information and the dynamic spatial distribution information, an environment-gas correlation map is generated. The sampling point decision unit is used to identify the spatial distribution characteristics of greenhouse gas concentrations based on the environment-gas correlation map, and generate an optimized sampling point sequence containing point coordinates and priorities according to a preset monitoring strategy. An adaptive path planning unit is used to generate an optimal navigation path based on the optimized sampling point sequence, the navigation performance parameters of the unmanned vessel, and environmental obstacle information.

3. The mobile greenhouse gas monitoring system based on autonomous exploration according to claim 2, characterized in that, The environment and gas correlation sensing unit includes: The spatiotemporal alignment and fusion subunit is used to perform spatiotemporal alignment on the real-time presampled data and the environmental obstacle information based on a unified time and space reference system, and generate spatiotemporally aligned real-time presampled data and environmental obstacle information. The spatial distribution field reconstruction subunit is used to process the spatiotemporally aligned real-time pre-sampled data through a spatial interpolation algorithm to generate a spatially continuously distributed enhanced gas concentration distribution field as dynamic spatial distribution information of greenhouse gases in the water-air interface. The multidimensional correlation map generation subunit is used to spatially overlay and correlate the enhanced gas concentration distribution field with the spatiotemporally aligned environmental obstacle information to generate an environment-gas correlation map that characterizes the coupling relationship between navigable areas, environmental features and greenhouse gas concentration distribution.

4. The mobile greenhouse gas monitoring system based on autonomous exploration according to claim 2, characterized in that, The sampling point decision unit includes: The distribution feature extraction subunit is used to analyze the environment-gas correlation map, extract high concentration clusters and gradient abrupt change zones as spatial distribution features, and divide the water area to be measured into multiple independent candidate monitoring areas based on the spatial boundaries of the extracted spatial distribution features. The monitoring priority ranking subunit is used to configure basic weights for different types of spatial distribution features based on a preset evaluation system, configure sub-item weights for each candidate monitoring area, generate a comprehensive priority score for each area through weighted calculation, and generate a monitoring priority sequence based on the score. The initial point generation subunit is used to generate an initial sampling point set according to the monitoring priority sequence and the pre-set layout rules. The pre-set layout rules are: to densify the points in high-priority areas, to linearly distribute the points along the gradient change zone in medium-priority areas, and to distribute the points in a uniform grid in low-priority areas. The point optimization and filtering subunit is used to simplify and adjust based on the endurance constraints of the unmanned vessel body, and obtain an optimized sampling point sequence containing point coordinates and priorities.

5. The mobile greenhouse gas monitoring system based on autonomous exploration according to claim 4, characterized in that, The preset evaluation system in the monitoring priority ranking subunit includes: The feature type weight library pre-stores the basic weights corresponding to different spatial distribution feature types. Among them, the weights configured for high concentration clusters are greater than those configured for gradient abrupt change zones. Regional attribute quantification rules are used to calculate a set of quantifiable attribute parameters for each candidate monitoring region, wherein the attribute parameters include at least the region area.

6. The mobile greenhouse gas monitoring system based on autonomous exploration according to claim 2, characterized in that, The adaptive path planning unit includes: The global path generation subunit is used to generate a cost map based on the optimized sampling point sequence containing point coordinates and priorities, the navigation performance parameters of the unmanned vessel, and the environmental obstacle information. The path optimization subunit is used on the cost map to calculate the optimal navigation path by using a motion planning algorithm, with the goal of minimizing the overall navigation cost and maximizing the attractiveness to high-priority points, and to generate the optimal navigation path connecting the current unmanned vessel position with subsequent points in the optimized sampling point sequence.

7. The mobile greenhouse gas monitoring system based on autonomous exploration according to claim 6, characterized in that, The global path generation subunit is specifically used for: An initial grid map is constructed, and the base value of each cell or node is determined by the environmental obstacle information, wherein areas occupied by obstacles are assigned a value of being prohibited from passage. Based on the optimized sampling point sequence containing point coordinates and priorities, a corresponding attraction field is generated on the initial map for each optimized sampling point; The attraction field is superimposed on the base cost value of the initial map to generate the cost map.

8. The mobile greenhouse gas monitoring system based on autonomous exploration according to claim 6, characterized in that, The path optimization subunit is specifically used for: A comprehensive objective function for path optimization is constructed. On the cost map, the A* algorithm is run to find candidate paths that optimize the comprehensive objective function. The candidate paths that optimize the comprehensive objective function are then smoothed to obtain the optimal navigation route.

9. The mobile greenhouse gas monitoring system based on autonomous exploration according to claim 8, characterized in that, The comprehensive objective function consists of two parts: the total navigation cost of the path and the total monitoring value gain of the path. The total navigation cost of the route is calculated by accumulating the basic travel costs corresponding to each location the route passes through on the cost map. The total monitoring value of the path is calculated cumulatively based on the proximity of the path to each point in the optimized sampling point sequence and the priority of each point. The optimization objective of the comprehensive objective function is to minimize the total cost of navigation and maximize the total monitoring value, while satisfying the constraints of navigation endurance.

10. The mobile greenhouse gas monitoring system based on autonomous exploration according to claim 1, characterized in that, The system also includes: The task management and storage module integrated into the unmanned vessel body is used to receive and store the environmental-gas correlation map, optimized sampling point sequence and optimal navigation path from the autonomous perception and dynamic planning module, as well as all pre-sampling data and fixed-point precise measurement data from the integrated greenhouse gas collection and monitoring module.

Images