A geological surveying and mapping method and system based on multi-unmanned aerial vehicle cooperation

Abstract

The application relates to the technical field of multi-unmanned aerial vehicle geological surveying and mapping, and discloses a multi-unmanned aerial vehicle cooperative geological surveying and mapping method and system. A distributed topology construction unit of the system constructs a directed graph topology structure according to a multi-unmanned aerial vehicle cluster communication relationship, defines an in-neighborhood and an out-neighborhood node set, and initializes a state observer; a geological data acquisition unit controls the unmanned aerial vehicles to fly according to a preset path, acquires geological deformation, environmental temperature and humidity and soil compactness data, and generates an original data set; a space-time analysis unit analyzes a timestamp to generate a dynamic time axis and mark a geological event node; a dynamic environment correction unit calculates an environmental correction factor, corrects geological deformation data to generate an environment-adaptive data set; and a cooperative state observation unit inputs the adaptive data set into the observer, updates parameters in combination with in-neighborhood node estimated values, and generates a local geological state estimation result. The system can improve the geological surveying and mapping data accuracy and state estimation precision, and adapt to a dynamic operation environment.

Description

Technical Field

[0001] This invention relates to the field of multi-UAV geological mapping technology, specifically to a geological mapping method and system based on multi-UAV collaboration. Background Technology

[0002] In fields such as mineral resource exploration, geological disaster monitoring, and preliminary surveying for engineering construction, the accuracy and efficiency of geological surveying directly affect the scientific nature of decision-making. As surveying demands continue to evolve, traditional geological surveying methods are gradually revealing numerous limitations. Early manual surveying relied on on-site personnel, which not only faced challenges related to accessibility in complex terrains (such as mountains, canyons, and swamps) but also suffered from long work cycles and limited coverage, making it difficult to meet the demands for large-scale, high-efficiency surveying.

[0003] To overcome the limitations of manual surveying, single-UAV surveying technology is gradually being applied to geological work. However, during operation, single UAVs are limited by their endurance and payload capacity, and a single flight can only cover a small area, and they cannot simultaneously and efficiently collect multiple types of geological data. At the same time, single UAVs have weak anti-interference capabilities, and once they encounter signal blockage, airflow disturbances, or other situations, data acquisition is easily interrupted or deviated, making it difficult to ensure the continuity and reliability of surveying data.

[0004] With technological advancements, multi-UAV collaborative mapping solutions have emerged, but existing solutions still have significant shortcomings. In terms of topology, most systems employ static topology designs, which cannot adapt to the dynamic network relationships arising from changes in UAV swarm location and communication distance during flight. This often leads to problems such as confused identification of neighboring nodes and hindered information exchange, impacting collaborative operation efficiency. Furthermore, the lack of clear definition of the ingress and egress sets of each UAV results in inaccurate initialization of the state observer, making subsequent state estimation prone to errors.

[0005] In the data processing stage, existing systems mostly focus only on the collection of geological deformation data, ignoring the impact of environmental parameters such as temperature and humidity, soil density, etc. on the deformation data, and failing to carry out targeted environmental corrections. This results in deviations between the collected raw data and the actual geological conditions. At the same time, the analysis of the timestamps in the raw data is not in-depth enough, and a dynamic timeline is not constructed to mark geological event nodes. This leads to a disconnect between geological data and the time dimension and event process, making it difficult to accurately trace the temporal patterns of geological changes, which is not conducive to subsequent analysis of the causes of geological events and trend prediction.

[0006] In terms of state observation, existing multi-UAV collaborative systems often adopt a centralized data processing mode, relying on a central node to summarize and update the observation data of all UAVs. This mode not only suffers from large data transmission delays and excessive load on the central node, but also lacks fault tolerance. Once the central node fails, the entire observation system will be paralyzed. Furthermore, it does not make full use of the state estimation information of neighboring nodes, relying solely on the data of a single UAV to update the observer parameters, resulting in insufficient accuracy and stability of local geological state estimation results, making it difficult to meet the actual needs of high-precision geological mapping. Summary of the Invention

[0007] The purpose of this invention is to provide a geological mapping method and system based on multi-UAV collaboration to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides a geological mapping system based on multi-UAV collaboration, the system comprising:

[0009] A distributed topology construction unit is used to construct a directed graph topology structure based on the communication relationship of multiple UAV clusters, define the in-neighbor node set and out-neighbor node set of each UAV, and initialize the state observer of each UAV based on the directed graph topology structure.

[0010] The geological data acquisition unit is used to control each UAV to fly along the planned path of the preset sampling points, and to collect geological deformation data, environmental temperature and humidity data and soil density data through the onboard sensors to generate the original sampling dataset.

[0011] The spatiotemporal parsing unit is used to parse and process the timestamp information in the original sampling dataset, generate a dynamic timeline, and mark geological event nodes on the dynamic timeline;

[0012] The dynamic environmental correction unit is used to calculate the environmental correction factor based on the environmental temperature and humidity data and soil density data, combined with the preset material characteristic benchmark value, and to use the environmental correction factor to correct the geological deformation data in real time, thereby generating an environmentally adaptive geological dataset.

[0013] The collaborative state observation unit is used to input the environmental adaptive geological dataset into the state observer of the corresponding UAV, and update the parameters of the state observer through distributed Kalman gain based on the state estimate of the incoming neighbor node set in the directed graph topology to generate local geological state estimation results.

[0014] Preferably, the distributed topology building unit is specifically used for:

[0015] Define the node connection relationship based on the communication link between drones, and assign a unique observer identifier to each drone node;

[0016] The incoming neighbor node set is defined as the set of neighbor nodes that send state estimates to the current UAV, and the number of incoming neighbor nodes is counted.

[0017] The out-neighbor set is defined as the set of neighboring nodes that receive the current UAV state estimate, and the number of neighboring nodes is counted.

[0018] A global topology mapping table is constructed based on the observer identifier, the set of incoming neighbor nodes, and the set of outgoing neighbor nodes.

[0019] Preferably, the geological data acquisition unit specifically includes:

[0020] The sampling point planning module is used to divide the area into grids based on the topographic features of the survey area and generate geological sampling points at the center of each grid.

[0021] The multi-sensor synchronization module is used to control the UAV to simultaneously trigger the 3D laser scanner to collect geological deformation data, the temperature and humidity sensor to collect environmental data, and the soil probe to collect soil compaction data when it arrives at the geological sampling point.

[0022] The data packaging module is used to attach timestamps and geographic location labels to each sample point data to generate the original sampled dataset with spatiotemporal labels.

[0023] Preferably, the dynamic environment correction unit specifically includes:

[0024] The environmental factor calculation module is used to generate an environmental impact index by weighted fusion based on the deviation between the environmental temperature and humidity data and the preset temperature and humidity benchmark values, combined with soil compaction data.

[0025] The correction factor generation module is used to calculate time-dependent environmental correction factors based on the environmental impact index and the preset material aging model.

[0026] The data calibration module is used to apply the time-dependent environmental correction factor to the geological deformation data, eliminate environmental interference errors, and output an environmentally adaptive geological dataset.

[0027] Preferably, the cooperative state observation unit specifically includes:

[0028] The neighborhood state integration module is used to obtain the geological state estimates of neighboring nodes from the state observer of the incoming neighboring node set and generate a neighborhood state consensus matrix.

[0029] The local gain update module is used to calculate the distributed Kalman gain based on the difference between the environmental adaptive geological dataset and the neighborhood state consensus matrix.

[0030] The parameter iteration module is used to update the error covariance matrix and state transition weights of the current state observer using the distributed Kalman gain, and output the local geological state estimation results.

[0031] Preferably, the local gain update module is specifically used for:

[0032] Based on the real-time variance of the environmental adaptive geological dataset and the preset attack probability parameters, calculate the anti-interference Kalman gain coefficient;

[0033] Based on the anti-interference Kalman gain coefficient and the neighborhood state consensus matrix, a state observer weight adjustment instruction is generated.

[0034] The scaling factor of the state transition matrix is ​​dynamically adjusted using the state observer weight adjustment command.

[0035] Preferably, the system further includes:

[0036] The three-dimensional geological reconstruction unit is used to input the local geological state estimation results of each UAV into the geographic information system, perform spatial interpolation processing based on the geographical location labels of the sampling points, and generate a high-resolution three-dimensional geological model.

[0037] The dynamic visualization unit is used to drive the temporal changes of the three-dimensional geological model based on the dynamic time axis, and to distinguish regions with different geological deformation rates through color coding, thereby generating an enhanced geological deformation map.

[0038] Preferably, the system further includes:

[0039] The stability assessment unit is used to delineate risk areas in the three-dimensional geological model and extract geological deformation data from all sampling points within those areas.

[0040] Calculate the mean volatility of the geological deformation data, and mark sampling points whose volatility exceeds the mean volatility as geologically unstable points;

[0041] By integrating the environmental temperature and humidity data and soil density data, a comprehensive geological risk index is generated.

[0042] Preferably, the stability evaluation unit is specifically used for:

[0043] The regional stability level is calculated based on the distribution density of the geologically unstable points and the comprehensive geological risk index.

[0044] When the stability level of the area is lower than a preset threshold, a sampling point encryption command is triggered to the geological data acquisition unit;

[0045] The risk area markers of the three-dimensional geological model are iteratively updated based on the data from newly added sampling points.

[0046] Preferably, the present invention also includes a geological mapping method based on multi-UAV collaboration, the method comprising all the modules and process flow of the above-mentioned geological mapping system based on multi-UAV collaboration.

[0047] Compared with the prior art, the beneficial effects of the present invention are:

[0048] This geological mapping system based on multi-UAV collaboration uses distributed topology building units to dynamically construct a directed graph topology structure based on the real-time communication relationships of multiple UAV clusters, rather than using a fixed static topology. This allows for flexible adaptation to the dynamic network environment caused by UAVs moving during flight and changing communication connections, effectively avoiding the problems of confusing neighbor node identification and poor information exchange under static topology. At the same time, it clearly defines the in-neighbor node set and out-neighbor node set of each UAV, providing a precise network relationship basis for the initialization of the state observer, making the initial parameters of the observer match the actual network position of the UAV and the distribution of neighbor nodes, laying a good foundation for subsequent accurate state estimation.

[0049] The geological data acquisition unit can control each UAV to fly along the pre-planned sampling point path, ensuring that the sampling process has a clear objective and regularity, avoiding sampling omissions or duplications caused by disordered UAV flight, and improving sampling efficiency. At the same time, this unit not only collects geological deformation data, but also simultaneously collects environmental temperature and humidity data and soil density data, generating a raw sampling dataset containing multiple types of parameters. Compared with traditional systems that only collect single deformation data, it can obtain more comprehensive geological mapping-related information, providing rich materials for subsequent multi-dimensional data processing and analysis.

[0050] The spatiotemporal analysis unit performs in-depth analysis of the timestamp information in the original sampled dataset to generate a dynamic timeline, transforming the originally scattered timestamp data into a continuous and ordered temporal framework. At the same time, geological event nodes are marked on the dynamic timeline, directly linking geological data with specific time points and geological events. This effectively solves the problem of geological data being disconnected from time and events in traditional systems. Staff can clearly trace the collection time corresponding to different geological data and the geological events that occurred at that time through the timeline, facilitating accurate analysis of the temporal characteristics of geological changes and clarifying the development of geological events.

[0051] The dynamic environmental correction unit fully utilizes the environmental temperature and humidity data and soil density data acquired by the geological data acquisition unit, combined with preset material property benchmark values, to calculate an environmental correction factor that conforms to the current environmental conditions. This correction factor is then used to correct the geological deformation data in real time, generating an environmentally adaptive geological dataset. This process effectively eliminates the interference of environmental factors on geological deformation data, such as instrument errors caused by temperature and humidity changes, and the impact of soil density differences on deformation perception. This makes the corrected geological data more closely reflect actual geological conditions, avoiding deviations between uncorrected data and reality, and significantly improving the authenticity and reliability of the geological data.

[0052] The collaborative state observation unit inputs the environmentally adaptive geological dataset into the state observer of the corresponding UAV. Instead of relying on data from a single UAV for parameter updates, it fully utilizes the state estimates of the neighboring node set in the directed graph topology, achieving dynamic parameter updates through distributed Kalman gain. This distributed processing mode eliminates the need for a central node; each UAV can autonomously adjust its parameters by combining observation information from its neighbors. This not only reduces data transmission latency and the load on the central node but also provides strong fault tolerance. Even if a UAV or some nodes fail, the remaining nodes can still conduct observations normally based on their own information and that of their remaining neighbors, ensuring overall system stability. Furthermore, incorporating the state estimates of neighboring nodes makes parameter updates more reliable. The generated local geological state estimates comprehensively reflect the geological conditions of the UAV itself and its surrounding areas, avoiding the limitations of single-node data and significantly improving the accuracy and stability of local geological state estimates. This, in turn, promotes comprehensive optimization of the mapping performance of the entire multi-UAV collaborative geological mapping system. Attached Figure Description

[0053] Figure 1 This is a timeline diagram of the geological mapping system based on multi-UAV collaboration described in this invention;

[0054] Figure 2 A flowchart illustrating the workflow of a distributed topology building unit;

[0055] Figure 3 A flowchart of the dynamic environment correction unit;

[0056] Figure 4 This is a flowchart of the local gain update module. Detailed Implementation

[0057] 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.

[0058] Please see Figure 1 This invention provides a geological mapping method and system based on multi-UAV collaboration, the system comprising:

[0059] A directed graph topology is established by a distributed topology construction unit based on the communication links between nodes in a multi-UAV swarm. Each UAV node is assigned a unique observer identifier, and in-neighbor and out-neighbor node sets are defined based on the communication direction. The state observers of each UAV are initialized according to this topology. The geological data acquisition unit controls the UAVs to fly along preset sampling point paths, simultaneously collecting geological deformation, environmental temperature and humidity, and soil density data using an onboard 3D laser scanner, temperature and humidity sensors, and soil probes, generating a raw sampling dataset with timestamps and geographic location labels. The spatiotemporal analysis unit parses the timestamps in the dataset, constructs a dynamic timeline, and labels geological event nodes. The dynamic environmental correction unit calculates environmental correction factors based on environmental data and preset material property benchmarks, performing real-time corrections on the geological deformation data to generate an environmentally adaptive geological dataset. The collaborative state observation unit inputs the corrected data into the state observers, combines it with the state estimates from the in-neighbor node sets, updates the observer parameters using distributed Kalman gain, and outputs local geological state estimation results.

[0060] Example 1: See Figure 2 In practice, the construction of distributed topology and the actual operation of geological data acquisition units involve a series of complex and interrelated technical steps. Taking a geological monitoring project in a mountainous area as an example, the system deployed six UAVs equipped with various types of sensors. In the initial stage of the project, the ground control center dynamically established a communication link map based on the physical location and communication relay capabilities of each UAV. The UAVs were numbered UAV-01 to UAV-06, and each device was assigned a unique observer identifier during network formation. For example, the identifier for UAV-01 was "N01-C band," where "N01" represents the node sequence number and "C band" indicates its primary communication frequency band. This naming method facilitates rapid data identification and routing by the system.

[0061] After the communication link is established, the system begins to define the neighbor relationships of each node. Taking UAV-03 as an example, its wireless module detects UAV-02 and UAV-04 as its stable neighboring nodes. Data flow analysis shows that UAV-02 continuously sends status data packets to UAV-03, while UAV-03 simultaneously transmits data to UAV-04. Therefore, the system adds UAV-02 to the in-neighbor set of UAV-03 and counts the number of nodes in this set as 1; it also adds UAV-04 to the out-neighbor set, with the same count of 1. The neighbor relationships of all nodes are stored in the form of a topology matrix on the central server and synchronized to the local memory of each UAV every 30 seconds to adapt to possible link changes.

[0062] During the geological data acquisition phase, the system first processes the topographic information of the surveyed area. This mountainous region covers approximately 12 square kilometers, and the digital elevation model shows an elevation difference of 800 meters. The algorithm divides the area into 200m x 200m grids based on slope variations. For areas with slopes greater than 25 degrees, the grid is automatically reduced to 100m x 100m to increase sampling density. A sampling point coordinate is generated at the center of each grid. After path planning, six non-intersecting flight routes are generated and assigned to each UAV. During the UAV's arrival at the sampling point, the positioning system corrects positional errors using real-time dynamic differential technology. When the UAV-01 enters a 10-meter radius around the target point, the onboard computing unit triggers a synchronous acquisition command. A 3D laser scanner acquires surface point cloud data at a rate of 2000 points per second, temperature and humidity sensors record the current air temperature and relative humidity, and a soil probe is inserted with constant pressure to a depth of 15 cm below the surface to measure resistance values. All sensor data are stamped with the same timestamp, which is generated by a high-precision BeiDou satellite clock and is in the format of "year-month-day-hour-minute-second-millisecond". The geographic location tag includes latitude and longitude coordinates and altitude value.

[0063] The data packaging module integrates multi-source data into structured data packets. Each data packet contains an index of the point cloud file acquired by the scanner, temperature and humidity values, soil compaction measurements, timestamps, and coordinate information. These data packets are transmitted to the ground station via a data radio. The original sampling dataset is stored in a distributed database in chronological order, with each data record labeled with a corresponding UAV identifier and sampling point number. During topology maintenance, the system continuously monitors the communication link status. When UAV-05 temporarily loses connection with UAV-04 due to terrain obstruction, the topology building unit detects the signal interruption, immediately removes UAV-04 from the ingress neighbor node set of UAV-05, and updates the global topology mapping table. Simultaneously, the path planning module dynamically adjusts the flight path of UAV-05, bringing it closer to the communication relay UAV-03 to restore the link. This dynamic adaptability enables the system to maintain a stable network topology in complex environments.

[0064] Practical problems encountered during data acquisition were also addressed. For example, when the UAV-02's soil probe encountered hard rock, the sensor reported abnormal pressure values. After identifying this anomaly, the system automatically marked the soil data at that sampling point as invalid and increased the sampling time of the laser scanner to compensate for the missing data. All invalid data marking information was recorded in the metadata for reference by the data processing unit. The implementation process reflects the characteristics of multi-system collaborative work. The topology construction unit ensures the reliability of the communication network, the data acquisition unit ensures the integrity and accuracy of the raw data, and the interaction between the two enables the system to adapt to dynamic changes in the field environment. Through precise time synchronization and spatial coordinate annotation, structured input data is provided for the data processing stage.

[0065] Example 2: See Figure 3 During the implementation of the dynamic environmental correction unit, the system receives raw sampling datasets from the geological data acquisition unit. This dataset includes ambient temperature and humidity readings collected by temperature and humidity sensors, as well as soil density measurements collected by soil probes. These data, along with preset material property benchmark values, are input into the environmental factor calculation module. The preset benchmark values ​​are obtained based on a geological material property database of the target monitoring area, which includes the coefficient of thermal expansion, moisture content sensitivity coefficient, and standard density range for different soil and rock types under standard laboratory conditions.

[0066] The environmental factor calculation module first processes temperature and humidity data. The currently collected ambient temperature reading is compared with a standard temperature reference value, and the absolute difference is calculated. This difference is multiplied by a weighting coefficient related to the material's thermal expansion characteristics to generate the temperature influence component. Similarly, the deviation between the ambient humidity reading and the standard humidity reference value, after adjustment for material moisture content sensitivity, generates the humidity influence component. Soil density data is normalized, mapped to a value range of zero to one, and then compared with a standard density reference value in the material property library to obtain the density influence component. These three influence components are integrated using a weighted fusion algorithm to generate a comprehensive environmental impact index. The weighting coefficients are obtained from the property library based on the specific geological material type. Different materials have different sensitivities to changes in temperature, humidity, and density, therefore the weighting allocation varies accordingly. For example, for clay rock layers, the weighting coefficient for humidity changes will be significantly higher than the weighting coefficient for temperature changes.

[0067] The correction factor generation module receives the environmental impact index and inputs it into a preset material aging model. This model considers the cumulative effect of time on environmental impact and employs a combined algorithm of a time decay function and an exponential compensation function. The material aging model simulates the changes in the physical properties of geological materials under long-term environmental exposure, and its output is a time-dependent environmental correction factor. This factor dynamically adjusts as monitoring time progresses, reflecting the cumulative effect of environmental impact.

[0068] The data calibration module receives raw geological deformation data and environmental correction factors to correct the deformation measurements. The calibration process eliminates errors caused by sensor thermal expansion and contraction due to temperature changes, signal attenuation due to moisture infiltration, and wave propagation errors caused by differences in soil density. The calibrated data output is an environmentally adaptive geological dataset, which retains the timestamps and geographic location tags of the original data while adding environmental correction factor values ​​and calibration process metadata. Throughout the environmental correction process, the system calculates the environmental impact index using the following formula:

[0069] ;

[0070] in: This represents the environmental impact index, a dimensionless comprehensive assessment value. This represents the weighting coefficient for the effect of temperature, and its value depends on the thermal expansion characteristics of the material. This indicates the currently collected ambient temperature measurement value, in degrees Celsius. This indicates the standard temperature reference value, in degrees Celsius. The weighting coefficient representing the influence of humidity is determined by the material's moisture content sensitivity. This indicates the currently collected ambient humidity measurement value, expressed as a percentage. This indicates the standard humidity reference value, expressed as a percentage. This represents the weighting coefficient for the influence of density, reflecting the material's sensitivity to changes in density. This indicates the currently collected soil density measurement value, in Newtons per square meter; This represents the maximum possible soil density, expressed in Newtons per square meter (N / m²). The parameters in this formula are dynamically obtained by querying a material property database, ensuring that the calculated environmental impact index accurately reflects the comprehensive impact of current environmental conditions on specific geological materials. All calculations are performed in real-time within the embedded processing unit, guaranteeing the timeliness of environmental corrections. The corrected dataset provides accurate input for co-state estimation, eliminating interference from environmental factors on geological deformation monitoring.

[0071] Example 3: See Figure 4In the implementation of the collaborative state observation unit, the system first needs to acquire the neighborhood state information of the current UAV node. Taking UAV node numbered UAV-03 as an example, this node queries the global topology mapping table and finds that its in-neighbor node set includes two nodes, UAV-02 and UAV-04. The system sends a state request command to these two neighboring nodes via a wireless data link. After receiving the request, the neighboring nodes encapsulate their latest generated geological state estimates, add a timestamp and node identifier, and send them back to UAV-03. The neighborhood state integration module of UAV-03 receives data packets from the two neighboring nodes, performs timestamp consistency verification, and removes expired data with a transmission delay exceeding 100 milliseconds. It parses the state estimates in the data packets, which include the estimation results of geological deformations, the estimation error range, and the confidence index. The module standardizes the estimates of the two nodes to eliminate the dimensional influence caused by the differences in calibration of different sensors, and arranges them in order of node identifier to form a two-dimensional neighborhood state consensus matrix. Each row of the matrix represents the complete state estimation information of a neighboring node, and the column vector contains dimensions such as deformation estimate, error range, and confidence level.

[0072] The local gain update module simultaneously receives the environmentally adaptive geological dataset and the neighborhood state consensus matrix from the dynamic environment correction unit. This module first calculates the difference between the latest deformation measurement value in the environmentally adaptive geological dataset and the estimates from each neighboring node in the consensus matrix, using the Euclidean distance algorithm to assess the consistency between the current measurement value and the neighbor estimates. Simultaneously, the module analyzes the real-time variance characteristics of the environmentally adaptive geological dataset to evaluate the stability and reliability of the current measurement data. Based on these analyses, the module calculates the distributed Kalman gain, which reflects the weight of the current measurement data in the state estimation. During the calculation, the system introduces a preset attack probability parameter to enhance anti-interference capabilities. This parameter is dynamically adjusted based on the quality of historical communication data, primarily considering two factors: first, the recent packet loss rate, obtained by statistically analyzing the proportion of successfully received data packets over a past period; and second, the data anomaly detection results, identified by analyzing the statistical characteristics of data sent by neighboring nodes to identify possible outliers. When the packet loss rate increases or abnormal data is detected, the attack probability parameter increases accordingly, causing the system to reduce its dependence on neighboring node data while increasing the trust weight of local measurement data.

[0073] The parameter iteration module receives a weight adjustment instruction generated by the local gain update module. This instruction includes the scaling factor correction value of the state transition matrix and the newly calculated Kalman gain coefficient. The module updates the error covariance matrix of the state observer, adjusting the weight distribution of each element to make the estimation results more aligned with a reliable data source. The module adjusts the parameters in the state transition matrix, recalibrating the step size and direction of the state prediction based on the scaling factor correction value. These adjustments ensure that the state estimation process considers both the latest information from current measurements and the collaborative observation results from neighboring nodes. After completing the parameter update, the state observer performs a new round of state estimation calculations, which integrates three data sources: the current deformation measurement value after environmental correction, the state estimation consensus from neighboring nodes, and the state estimation result from the previous time step. The output of the calculation contains new local geological state estimation results, stored in a structure containing the deformation estimate, estimation error range, confidence index, and calculation timestamp. This data is stored in a local database for later use and also sent to other UAV nodes in the neighboring node set according to the topology mapping table.

[0074] The collaborative state observation process forms a distributed computing architecture, where each UAV node acts as both a producer of state estimates and a consumer of data from its neighbors. This design enables state coordination among multiple nodes without the need for a central processing unit. Even if some nodes experience communication interruptions or data anomalies, the remaining nodes can still maintain a certain level of state estimation capability through neighborhood information exchange. This distributed characteristic makes the system particularly suitable for long-term geological monitoring in field environments with limited communication conditions.

[0075] Example 4: The 3D geological reconstruction unit receives local geological state estimation results from all UAV nodes. This data is aggregated to the ground processing center via a wireless data transmission network. Each data packet contains the geological deformation estimate processed by the collaborative state observation unit, the corresponding 3D geographic coordinates, the data acquisition timestamp, and the estimation accuracy index. The data processing engine first performs spatiotemporal alignment on the input data, unifying data collected at different times and spatial locations into the same spatiotemporal reference frame. The spatial interpolation algorithm uses an improved Kriging method, which not only considers the spatial relationship of the sampling points but also incorporates the estimation accuracy index of each data point as interpolation weights.

[0076] The Geographic Information System (GIS) platform establishes a 3D mesh model, with the mesh resolution adaptively adjusted based on the sampling point density. The geological deformation value of each mesh node is obtained through spatial interpolation, prioritizing data points with close spatial proximity and high estimation accuracy. For areas with missing data, the system employs a terrain-feature-constrained interpolation method, utilizing slope and aspect information from the digital elevation model to assist in estimating the deformation characteristics of the missing areas. After interpolation, a high-resolution 3D geological model is generated. This model is stored as a vertex array, with each vertex containing latitude and longitude coordinates, elevation, and geological deformation vector data.

[0077] The dynamic visualization unit connects to the dynamic timeline generated by the spatiotemporal analysis unit. This timeline uses geological event nodes as keyframes, dividing continuous monitoring time into multiple time intervals. The visualization engine loads 3D geological models from different time points in chronological order and generates smooth terrain change animations through vertex displacement interpolation algorithms. The color encoding module calculates the corresponding color value based on the deformation rate of each grid vertex. The deformation rate is calculated based on the rate of change of deformation between adjacent time points. The color mapping uses a piecewise linear function, dividing the deformation rate range into multiple intervals. Each interval is mapped to a color spectrum, gradually transitioning from blue (representing low deformation rate) to red (representing high deformation rate).

[0078] The stability assessment unit automatically identifies potential risk areas in a 3D geological model based on multi-source data analysis. The identification algorithm first calculates the topographic curvature distribution of the entire region, marking areas with curvature exceeding a threshold as topographically complex zones. Simultaneously, it analyzes historical deformation data, calculating the cumulative deformation and trend at each grid point. Combining the spatial distribution of topographically complex and anomaly zones, morphological operations are used to generate continuous risk area boundaries. Deformation data sequences from all sampling points within the risk area are extracted, and the deformation fluctuation characteristics of each point are calculated.

[0079] Volatility calculation employs a time-series analysis-based algorithm. The deformation data series for each sampling point is detrended, and the standard deviation of the detrended series is calculated as the volatility indicator. The mean volatility for the entire risk region is obtained through a weighted average, with the weights being the reciprocal of the estimation precision of each sampling point. Sampling points whose volatility exceeds a certain multiple of the regional mean are marked as geologically unstable points; these points are labeled with special symbols in the geological model.

[0080] The comprehensive geological risk index integrates multi-dimensional data, including environmental temperature and humidity data, soil density data, and deformation monitoring data. The weights of each data point are determined using principal component analysis, automatically calculating weight coefficients by analyzing the correlation between each parameter and geological stability in historical data. The risk index is calculated using the following formula:

[0081] ;

[0082] in: This represents the comprehensive geological risk index, a dimensionless assessment value. Representing the The weight coefficients of each evaluation parameter are calculated through principal component analysis. This represents a standardized function that maps the values ​​of its parameters to a range of zero to one. Indicates the first The original values ​​of the evaluation parameters include multiple parameters such as deformation fluctuation rate and temperature and humidity deviation; Indicates the total number of evaluation parameters; This represents the weighting coefficient for the distribution density of unstable points; Indicates the number of geologically unstable points within the risk area; This indicates the total number of sampling points within the risk area. The risk assessment system can dynamically update the risk index; when new monitoring data is input, it recalculates the values ​​of each parameter and updates the risk assessment results. The spatial extent of the risk area also continuously adjusts as new data accumulates, achieving dynamic risk identification and assessment. All risk assessment results are integrated with a 3D geological model for display, providing geological monitoring personnel with an intuitive risk distribution map.

[0083] Example 5: During the implementation of the stability assessment unit, the system conducts an in-depth analysis of the risk areas delineated in the three-dimensional geological model. This unit first extracts geological deformation data sequences from all sampling points within the risk area. These data sequences contain deformation measurements at different time points and their corresponding estimation accuracy indices. In the data preprocessing stage, the deformation sequence of each sampling point is smoothed, and a sliding window algorithm is used to eliminate random noise interference while preserving the true deformation trend characteristics. Volatility is calculated based on the detrended residual sequence, and statistical analysis methods are used to calculate the deformation volatility amplitude of each sampling point. This volatility amplitude reflects the instability of the geological body at that location.

[0084] The mean volatility of risk areas is calculated using a weighted average method, with weighting coefficients determined based on the data quality and observation duration of each sampling point. Sampling points with longer monitoring periods and higher data quality are assigned higher weights, while those with lower data quality are assigned lower weights. The system sets a volatility threshold, typically 1.5 standard deviations from the regional average; sampling points exceeding this threshold are marked as geologically unstable points. These unstable points are prominently displayed in the 3D model, along with their geographic coordinates, volatility values, and marking time.

[0085] The comprehensive geological risk index is calculated by integrating multi-source monitoring data, including environmental temperature and humidity data, soil density data, and deformation monitoring data. Environmental temperature and humidity data provide information on atmospheric and surface geological conditions, soil density data reflects the internal structural characteristics of the geological body, and deformation monitoring data directly characterizes the stability of the geological body. The weight allocation of each data source is dynamically determined using principal component analysis, which automatically calculates the optimal weight coefficients based on the statistical relationship between each parameter and geological stability in historical data. After the risk index is calculated, the system classifies the risk level according to the index value and establishes a correspondence between the risk level and the index value.

[0086] Regional stability level assessment is based on the spatial distribution characteristics of geologically unstable points and a comprehensive geological risk index. The distribution density is calculated using a kernel density estimation method, establishing an influence range centered on each unstable point, and then overlaying these values ​​to obtain a distribution density map of unstable points across the entire risk area. This density value reflects the spatial clustering of unstable points; higher density indicates more significant geological instability in the area. The distribution density value is weighted and fused with the risk index, and mapped to a specific stability level using a pre-defined decision tree rule. The decision tree rule considers multiple scenario modes, including combinations of high-risk index and high distribution density, and combinations of high-risk index and low distribution density, each corresponding to a different stability level determination.

[0087] When the system determines that the stability level of a certain area is lower than a preset safety threshold, it automatically triggers a sampling point encryption command. The command generation process first analyzes the spatial distribution characteristics of the existing sampling points to identify sparsely sampled sub-regions. The generation of encryption point locations employs a spatial optimization algorithm, prioritizing the addition of new sampling points in areas with concentrated unstable points and areas with sparse existing sampling, while maintaining overall distribution uniformity. The number of new sampling points is dynamically calculated based on the area and the degree of deviation in stability level; the greater the deviation, the more points are added. The generated encrypted sampling point set contains detailed geographic coordinate information and sampling priority identifiers, and is transmitted to the geological data acquisition unit via a data interface.

[0088] After receiving the encrypted command, the geological data acquisition unit replans the UAV's flight path. The path planning algorithm considers the geographical location and priority of the new sampling points, combined with the UAV's current position and endurance, to generate the optimal data acquisition path. The UAV flies along the new path, triggering the sensor array at the new sampling point locations to collect data. The data type and accuracy requirements of the collected data remain consistent with those of the original sampling points. The newly acquired data is packaged and transmitted to the data processing center to participate in the data processing flow.

[0089] After new sampling point data is input into the system, the environmental adaptive correction unit first corrects for environmental factors to eliminate the influence of temperature, humidity, and soil conditions on the measurement data. The corrected data is then input into the collaborative state observation unit, which, based on the latest topological relationship network, integrates monitoring data from both new and old sampling points to recalculate the geological state estimates for each location. During the calculation process, newly acquired data, due to its timeliness and specificity, is often assigned a higher weighting coefficient to improve the accuracy of the state estimation.

[0090] The 3D geological reconstruction unit receives the updated state estimation results and uses an incremental update algorithm to correct the original model. This algorithm identifies the model region affected by newly added sampling points, re-interpolates the grid nodes within that region, and simultaneously maintains the model data in unaffected areas. This incremental update method improves model update efficiency and reduces computational resource consumption. After the model update is complete, the boundaries and features of the risk area are adjusted to reflect the geological conditions revealed by the latest monitoring data.

[0091] The stability assessment unit, based on the updated 3D geological model, re-executes the risk zone delineation and unstable point identification processes. The new assessment process considers supplementary information provided by newly added sampling points and recalculates the regional stability level. If the reassessed stability level is still below the safety threshold, the system generates encrypted sampling instructions again, initiating a new round of data acquisition and processing. This iterative process continues until the regional stability level meets safety requirements or the number of iterations reaches the system's set upper limit. Throughout the implementation, the system maintains a complete operation log, including the results of each stability assessment, the content of triggered encrypted instructions, data quality information of newly added sampling points, and state changes after model updates. This log data is used to analyze the system's operational efficiency and the adaptability of the assessment algorithm, providing data support for system optimization. Simultaneously, the system provides a manual intervention interface, allowing professionals to review and adjust automatically generated encrypted instructions and assessment results, ensuring the rationality and reliability of system decisions. This closed-loop feedback mechanism achieves a balance between monitoring accuracy and efficiency, minimizing resource consumption while ensuring monitoring effectiveness through intelligent data acquisition strategy optimization. The system can automatically adjust monitoring strategies according to changes in the geological environment, demonstrating high adaptability and intelligence. This dynamic optimization capability enables the system to undertake long-term geological monitoring tasks, providing reliable technical support for geological disaster early warning and prevention.

[0092] Example 6: In a river and lake mapping project in a certain area of ​​the Hebei section of the Grand Canal, eight UAVs equipped with various mapping sensors were deployed to acquire geospatial data of rivers, waterways, lakes, and surrounding shorelines within the area. This provided mapping support for water system monitoring and ecological base analysis in the protection of the Grand Canal cultural heritage. At the initial stage of the project, the distributed topology construction unit first constructed a directed graph topology based on the communication relationships of the UAV cluster. The ground control center determined the node connection relationships by detecting the signal strength of the communication links between each UAV and assigned a unique observer identifier to each UAV, such as "UAV-H1-Canal Section" and "UAV-H2-Lake Area," with the suffix clearly indicating the mapping area the UAV was responsible for.

[0093] The system defines the ingress and egress node sets for each UAV. Taking UAV-H3, responsible for mapping the main canal channel, as an example, its ingress node set consists of neighboring nodes that can send state estimates to it. Communication detection reveals that UAV-H2 (lake area) and UAV-H4 (canal tributary area) can stably transmit data, so these two nodes are included in UAV-H3's ingress node set, and the total number of ingress and egress nodes is 2. The egress node set consists of nodes that receive UAV-H3's state estimates. UAV-H1 (upstream canal area) and UAV-H5 (downstream canal area) are confirmed to meet the criteria, and their egress node count is also 2. Based on each UAV's observer identifier, ingress and egress node sets, the system constructs a global topology mapping table, which is updated every 25 seconds to adapt to the dynamic changes in communication links during UAV flight. After the topology construction is completed, the state observers of each UAV initialize according to this directed graph topology structure.

[0094] The geological data acquisition unit controls the drone's flight according to the pre-planned sampling point route. The sampling point planning module first divides the area into grids based on the topographic features of the Hebei section of the Grand Canal and the distribution of rivers and lakes. A 100m x 100m grid is used in narrow river areas, while a 200m x 200m grid is used in open lake areas. The center of each grid is the geological sampling point. Additional sampling points are added at river bends and locations with significant shoreline changes to ensure the mapping data accurately reflects the geographical features of key areas. When the drone arrives at a sampling point, the multi-sensor synchronization module triggers various sensors: a 3D laser scanner collects topographic data of the river and lake shorelines for subsequent analysis of shoreline morphology changes; a millimeter-wave radar sensor collects water depth data to obtain underwater topographic information; and an optical camera captures images of the shoreline and surrounding vegetation to aid in ecological indicator analysis. The data packaging module adds a high-precision timestamp (formatted as "YYYY-MM-DD HH:MM:SS.fff") and a geographic location label (containing latitude, longitude and altitude information) generated by the BeiDou satellite positioning system to the multi-source data of each sampling point, generating a raw sampling dataset with spatiotemporal labels, which is then transmitted to the ground processing center in real time via a wireless data transmission link.

[0095] After receiving the raw sampled dataset, the spatiotemporal analysis unit parses and processes the timestamp information. The system extracts the timestamp of each data record, arranges them chronologically, and generates a dynamic timeline. During the construction of the dynamic timeline, historical hydrological data and field survey information are combined to mark geological event nodes, such as water level changes in a section of river due to seasonal factors at a specific time, and morphological adjustments of lake shorelines due to natural erosion. This allows the dynamic timeline to clearly present the changes in the geographical state of rivers and lakes at different points in time.

[0096] The dynamic environmental correction unit processes the environmental temperature and humidity data and soil density data (mainly collected from the shoreline) in the original sampling dataset to correct for geological deformation data (primarily shoreline and underwater topography). The environmental factor calculation module first compares the environmental temperature and humidity data with preset temperature and humidity benchmarks (determined based on multi-year average climate data of the Hebei section of the Grand Canal) to obtain the deviation value. Then, it combines this with the soil density data to generate an environmental impact index through weighted fusion. For example, when a region has a large temperature and humidity deviation and low soil density, the environmental impact index increases accordingly, indicating that environmental factors in that region may have a stronger impact on the topographic data. The correction factor generation module calculates a time-dependent environmental correction factor based on the environmental impact index and a preset material aging model (constructed to account for the natural changes in shoreline soil and rock). This factor adjusts over time to adapt to the degree of environmental impact on the topography at different periods. The data calibration module applies a time-dependent environmental correction factor to the terrain data collected by the 3D laser scanner and millimeter-wave radar sensor, eliminating the interference of sensor measurement errors caused by temperature and humidity changes and soil density differences on terrain perception, and finally generating an environmentally adaptive geological dataset.

[0097] The collaborative state observation unit inputs the environmentally adaptive geological dataset into the state observer of the corresponding UAV. The neighborhood state integration module obtains the geological state estimates of neighboring nodes from the state observers of each UAV entering the neighborhood node set. Taking UAV-H3 as an example, it obtains the terrain state estimates of the lake area and canal tributary area from neighboring nodes UAV-H2 and UAV-H4. After data format unification and validity verification, a neighborhood state consensus matrix is ​​generated. The local gain update module compares the differences between the environmentally adaptive geological dataset and the neighborhood state consensus matrix. Combining the real-time variance of the environmentally adaptive geological dataset (reflecting the stability of the current measurement data) and the preset attack probability parameter (set based on historical data of wireless communication interference in the area), it calculates the distributed Kalman gain. The parameter iteration module uses the distributed Kalman gain to update the error covariance matrix and state transition weights of the current state observer. For example, when the difference between a neighboring node's data and the local data is small, the weight of that neighboring node's data in the state estimation is appropriately increased; otherwise, the weight is decreased. Finally, the local geological state estimation result is output, which is the accurate terrain mapping result of the area responsible for each UAV.

[0098] The 3D geological reconstruction unit inputs the local geological state estimation results from all UAVs into the Geographic Information System (GIS). Based on the geographic location labels of the sampling points, a spatial interpolation algorithm (selecting an appropriate interpolation model based on the terrain features of the Grand Canal area) is used to complete and optimize the data, generating a high-resolution 3D geological model. This model clearly presents the depth changes of river channels, the topographic undulations of lakes and waterways, and the spatial morphology of shorelines. The dynamic visualization unit, based on the dynamic timeline generated by the spatiotemporal analysis unit, drives the 3D geological model to display temporal changes in chronological order. Simultaneously, color coding is used to distinguish the rate of terrain change in different areas; for example, blue represents areas of slow change, and red represents areas of relatively significant change. This generates an enhanced geological deformation map, intuitively presenting the geospatial changes of rivers and lakes in the Hebei section of the Grand Canal.

[0099] The stability assessment unit delineates risk zones in the 3D geological model, primarily including erosion-prone riverbanks and lake water level-sensitive areas. The system extracts geological deformation data (topographic change data) from all sampling points within these risk zones, calculates the mean volatility of the geological deformation data, and marks sampling points with volatility exceeding the mean as geologically unstable points. For example, if the topographic change volatility of a sampling point on a riverbank is significantly higher than the mean, that point is marked as unstable. Subsequently, the system integrates environmental temperature and humidity data and soil density data to comprehensively analyze the influencing factors of each unstable point, generating a comprehensive geological risk index. Based on the distribution density of geologically unstable points and the comprehensive geological risk index, the regional stability level is calculated. When the stability level of a region falls below a preset threshold, the system triggers a sampling point encryption command sent to the geological data acquisition unit. The geological data acquisition unit, based on the encryption command, adds new sampling points in that region, adjusts the UAV flight path, and supplements the collected data. Based on the newly added sampling point data, the risk area markers of the three-dimensional geological model are iteratively updated, making the delineation of risk areas more accurate. This provides detailed and accurate mapping data support for the ecological protection of rivers and lakes in the Hebei section of the Grand Canal and the monitoring of the geographical environment around cultural heritage sites, helping to carry out subsequent work such as shoreline restoration and aquatic ecological maintenance, and adapting to the refined geospatial data requirements in the protection of the Grand Canal cultural heritage.

[0100] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0101] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A geological mapping system based on multi-unmanned aerial vehicle cooperation, characterized in that, The method comprises the following steps: A distributed topology construction unit is configured to construct a directed graph topology according to the communication relationship of the multi-UAV cluster, define the in-neighborhood node set and out-neighborhood node set of each UAV, and initialize the state observer of each UAV based on the directed graph topology; A geological data acquisition unit is configured to control each UAV to fly along a path planned according to a preset sampling point, acquire geological deformation data, environmental temperature and humidity data, and soil density data through a sensor carried by the UAV, and generate an original sampling data set; A space-time analysis unit is configured to analyze the timestamp information in the original sampling data set, generate a dynamic time axis, and label a geological event node on the dynamic time axis; A dynamic environment correction unit is configured to calculate an environmental correction factor according to the environmental temperature and humidity data and the soil density data, in combination with a preset material characteristic benchmark value, correct the geological deformation data in real time by using the environmental correction factor, and generate an environment-adaptive geological data set; A cooperative state observation unit is configured to input the environment-adaptive geological data set into the state observer of the corresponding UAV, update the parameters of the state observer by using a distributed Kalman gain based on the state estimation value of the in-neighborhood node set in the directed graph topology, and generate a local geological state estimation result. The distributed topology construction unit is specifically configured to: define the node connection relationship according to the communication link between the UAVs, assign a unique observer identifier to each UAV node, define the in-neighborhood node set as a neighbor node set that sends a state estimation value to the current UAV, and count the number of in-neighborhood nodes, define the out-neighborhood node set as a neighbor node set that receives the state estimation value of the current UAV, and count the number of out-neighborhood nodes, and construct a global topology relationship mapping table based on the observer identifier, the in-neighborhood node set, and the out-neighborhood node set. The geological data acquisition unit specifically comprises: a sampling point planning module configured to divide a grid according to the terrain features of a surveying and mapping area, and generate a geological sampling point at the center of each grid; a multi-sensor synchronization module configured to control the UAV to synchronously trigger a three-dimensional laser scanner to acquire geological deformation data, a temperature and humidity sensor to acquire environmental data, and a soil probe to acquire soil density data when the UAV arrives at the geological sampling point; 2.The multi-UAV coordinated based geological mapping system of claim 1, wherein, a data packaging module configured to attach a timestamp and a geographic location label to the data of each sampling point, and generate an original sampling data set with space-time labels. The dynamic environment correction unit specifically comprises: an environmental factor calculation module configured to generate an environmental impact index by weighted fusion based on the deviation of the environmental temperature and humidity data from a preset temperature and humidity benchmark value, in combination with the soil density data; a correction factor generation module configured to calculate a time-dependent environmental correction factor according to the environmental impact index and a preset material aging model; 3.The multi-UAV coordinated based geological mapping system of claim 2, wherein, a data calibration module configured to apply the time-dependent environmental correction factor to the geological deformation data to eliminate environmental interference errors, and output an environment-adaptive geological data set. The cooperative state observation unit specifically comprises: ​ ​ 4.The multi-UAV coordinated based geological mapping system of claim 3, wherein, ​ a neighborhood state integration module configured to obtain a neighbor node geological state estimation value from a state observer of the set of inbound neighborhood nodes, and generate a neighborhood state consensus matrix; a local gain update module configured to calculate a distributed Kalman gain based on a difference between the environmental adaptive geological data set and the neighborhood state consensus matrix; a parameter iteration module configured to update an error covariance matrix and a state transition weight of the current state observer using the distributed Kalman gain, and output a local geological state estimation result. 5.The multi-UAV coordinated based geological mapping system of claim 4, wherein, The local gain update module is specifically configured to: calculate an anti-interference Kalman gain coefficient according to a real-time variance of the environmental adaptive geological data set and a preset attack probability parameter; generate a state observer weight adjustment instruction based on the anti-interference Kalman gain coefficient and the neighborhood state consensus matrix; dynamically adjust a scaling factor of a state transition matrix through the state observer weight adjustment instruction. 6.The multi-UAV coordinated based geological mapping system of claim 5, wherein, Further comprising: a three-dimensional geological reconstruction unit configured to input the local geological state estimation result of each unmanned aerial vehicle into a geographic information system, perform spatial interpolation processing according to a sampling point geographic location label, and generate a high-resolution three-dimensional geological model; a dynamic visualization unit configured to drive a time sequence change of the three-dimensional geological model based on the dynamic timeline, and generate an enhanced geological deformation map by color coding different geological deformation rate regions.

7. The multi-UAV coordination based geological mapping system of claim 6, wherein, Further comprising: a stability evaluation unit configured to demarcate a risk region in the three-dimensional geological model, and extract geological deformation data of all sampling points in the region; calculate a mean value of a volatility rate of the geological deformation data, and mark a sampling point whose volatility rate exceeds the mean value as a geological unstable point; fuse the environmental temperature and humidity data and the soil density data to generate a geological risk comprehensive index. 8.The multi-UAV coordinated based geological mapping system of claim 7, wherein, The stability evaluation unit is specifically configured to: calculate a regional stability level according to a distribution density of the geological unstable points and the geological risk comprehensive index; trigger a sampling point encryption instruction to the geological data collection unit when the regional stability level is lower than a preset threshold; iteratively update the risk region label of the three-dimensional geological model based on new sampling point data.

9. A geological mapping method based on multi-unmanned aerial vehicle cooperation, characterized in that, All modules and method processes of the multi-unmanned aerial vehicle cooperative-based geological surveying and mapping system according to any one of claims 1 to 8 are included.

Images