Ancient tree health state monitoring method based on digital twinning

Abstract

The invention discloses an ancient tree health state monitoring method based on digital twinning, and relates to the field of ancient tree health monitoring, and the method comprises the steps: collecting point cloud data and texture data of an ancient tree, and constructing an initial three-dimensional model of the ancient tree; a micro ion selective electrode array is deployed in an ancient tree root distribution area, Ca2 and K + ion flow velocity changes are detected through a constant potential rectifier, and the mycorrhizal symbiosis index is calculated in combination with soil temperature and humidity data; uploading the mycorrhizal symbiosis index to a digital twinborn body, and displaying the active state of a root system area in an initial three-dimensional model of the ancient tree in a three-dimensional thermodynamic diagram form; a 16-channel MOX gas sensor electronic nose array is installed on the surface of a trunk of an ancient tree, and volatile organic compound (VOC) concentration data of the ancient tree are collected. According to the invention, a micro electrode array with a 45-degree inclination angle is adopted to non-destructively monitor the ion flow velocity of a root system through the mycorrhizal symbiosis index detection unit, and a three-dimensional thermodynamic diagram is generated in combination with an inverse distance weighted interpolation algorithm, so that dynamic visualization of underground mycorrhizal activity is realized.

Description

Technical Field

[0001] This invention relates to the field of ancient tree health monitoring, and in particular to a method for monitoring the health status of ancient trees based on digital twins. Background Technology

[0002] In the field of ancient tree health monitoring, existing technologies have gradually evolved from traditional manual detection to digitalization and intelligence. By deploying equipment such as soil moisture sensors and trunk tilt meters, and combining them with Internet of Things (IoT) technology to achieve real-time data collection and analysis, this technology can monitor changes in the growth environment parameters and physical state of ancient trees and visualize them through a cloud platform, significantly improving monitoring efficiency and data coverage. However, this solution mainly relies on the monitoring of single environmental parameters and macroscopic physical indicators, lacks quantitative assessment of the microscopic physiological state of ancient trees, and fails to achieve accurate mapping of health status in three-dimensional space.

[0003] The shortcomings of existing technologies lie in their limited monitoring dimensions, which are limited to environmental and physical indicators and cannot fully reflect the physiological health status of ancient trees. Mycorrhizal symbiotic activity, as a key biomarker of ancient tree health, shows a significant correlation between dynamic changes and disease occurrence. However, existing technologies have not effectively integrated the synergistic analysis of mycorrhizal activity detection and disease early warning. Traditional methods often present data in the form of two-dimensional charts or simple three-dimensional models, which cannot intuitively show the spatial distribution characteristics of problems such as internal decay and root degradation in ancient trees. This results in a lack of accurate spatial positioning basis for maintenance decisions. The application of digital twin technology in the health monitoring of ancient trees is still in the initial exploratory stage. Existing solutions mostly focus on geometric modeling and static data display, without fully combining the dynamic fusion and intelligent diagnosis of multimodal physiological data. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a digital twin-based method for monitoring the health status of ancient trees, which solves the problems of existing technologies in quantifying the physiological health status of ancient trees and lacking accurate spatial visualization.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for monitoring the health status of ancient trees based on digital twins, which includes collecting point cloud data and texture data of ancient trees and constructing an initial three-dimensional model of the ancient trees;

[0008] A miniature ion-selective electrode array was deployed in the root distribution area of ​​ancient trees, and Ca was detected using a potentiostat. 2 and K + The mycorrhizal symbiosis index was calculated by combining changes in ion flow rate with soil temperature and humidity data.

[0009] The mycorrhizal symbiosis index is uploaded to the digital twin, and the activity status of the root system area is displayed in the form of a three-dimensional heat map in the initial three-dimensional model of the ancient tree.

[0010] A 16-channel MOX gas sensor electronic nose array was installed on the surface of the ancient tree trunk to collect data on the concentration of volatile organic compounds (VOCs) in the ancient tree.

[0011] VOC concentration data is input into the graph convolutional network (GCN) model, matched with the VOC feature patterns of ancient tree diseases in the ancient tree disease knowledge graph, to obtain disease types and probability distributions. Disease types and probability distributions are detected, and their locations are marked in a digital twin to obtain three-dimensional early warning markers.

[0012] The mycorrhizal symbiosis index and three-dimensional early warning markers are integrated into the initial three-dimensional model of the ancient tree to generate a health status report of the ancient tree.

[0013] As a preferred embodiment of the digital twin-based ancient tree health status monitoring method of the present invention, the method includes the following steps: collecting point cloud data and texture data of the ancient tree to construct an initial three-dimensional model of the ancient tree.

[0014] The tree trunk and main branches are scanned using LiDAR to obtain point cloud data, and hyperspectral images are acquired using a hyperspectral camera to obtain texture data.

[0015] Point cloud data is registered using the ICP algorithm to generate a triangular mesh model. Hyperspectral data is then mapped to a texture map of the triangular mesh model to obtain an initial 3D model of the ancient tree with multispectral texture.

[0016] As a preferred embodiment of the digital twin-based ancient tree health status monitoring method of the present invention, a micro-ion-selective electrode array is deployed in the root distribution area of ​​the ancient tree, and Ca is detected by a potentiostat. 2 and K + The mycorrhizal symbiosis index is calculated by analyzing changes in ion flow rate in conjunction with soil temperature and humidity data, including the following steps.

[0017] Micro-ion electrodes were deployed in a ring around the tree trunk, and inserted into the soil at a 45° angle to obtain the spatial coordinate information of the electrodes.

[0018] Ca was collected using a potentiostat. 2+ and K + The current signal is used to obtain the raw current-time series data. Wavelet denoising is performed on the current data to generate the ion activity time series curve.

[0019] By replacing the ion activity time series curve with ion flow rate according to the Nernst equation, we obtain Ca 2+ and K + Spatial distribution of velocity field;

[0020] Soil volumetric water content and temperature were collected using a Decagon 5TE sensor to calculate the mycorrhizal symbiosis index.

[0021] As a preferred embodiment of the digital twin-based ancient tree health status monitoring method of the present invention, the method includes the following steps: uploading the mycorrhizal symbiosis index to the digital twin, and displaying the root system activity status in the form of a three-dimensional heat map in the initial three-dimensional model of the ancient tree.

[0022] Based on the mycorrhizal symbiosis index, the inverse distance weighting method was used to complete the missing electrode data and obtain a complete electrode dataset.

[0023] Align the electrode coordinates with the root mesh vertex space in the initial 3D model of the ancient tree to generate a vertex list with MSI attributes;

[0024] Map the vertex list of MSI attributes to the HSL color space to obtain a vertex color encoding array. Render the vertex colors using UnityShader Graph to obtain a 3D heatmap with transparency adjustment. Add ray detection feet to the 3D heatmap area to obtain an interactive digital twin.

[0025] As a preferred embodiment of the digital twin-based ancient tree health status monitoring method of the present invention, the method includes: installing a 16-channel MOX gas sensor electronic nose array on the surface of the ancient tree trunk to collect VOC concentration data, comprising the following steps.

[0026] A 16-channel MOX gas sensor electronic nose is used. The MOX gas sensor electronic nose is spirally arranged on the trunk, connected to the canopy solar panel equipped with a supercapacitor for energy storage, and driven by constant current to obtain the original resistance value.

[0027] The original resistance values ​​were converted into VOC concentration data of ancient trees.

[0028] As a preferred embodiment of the digital twin-based ancient tree health status monitoring method of the present invention, the method includes the following steps: inputting VOC concentration data into a graph convolutional network (GCN) model, matching the VOC feature patterns of ancient tree diseases in the ancient tree disease knowledge graph, and obtaining the disease type and probability distribution.

[0029] The VOC concentration data of ancient trees were normalized to obtain normalized VOC concentration data of ancient trees, and feature extraction was performed to obtain the time-series features of VOC concentration.

[0030] Based on the temporal characteristics of VOC concentration and knowledge graph, a disease association graph is generated, and the adjacency matrix and node feature matrix are obtained.

[0031] The node feature matrix is ​​aggregated by 3 layers of GATv2 graph convolution to obtain the high-order graph embedding vector. The cosine similarity between the high-order graph embedding vector and the VOC feature pattern of ancient tree diseases in the knowledge graph is calculated.

[0032] The Softmax layer based on the GCN model transforms the high-order graph embedding vector into a disease probability distribution.

[0033] As a preferred embodiment of the ancient tree health status monitoring method based on digital twins described in this invention, the method includes the following steps: detecting disease types and probability distributions, marking their locations on the digital twin, and obtaining three-dimensional early warning markers.

[0034] The rendering template parameters are obtained by retrieving the 3D annotation rules from the knowledge graph based on the disease type.

[0035] By combining the VOC concentration data of ancient trees with the three-dimensional heat map, the three-dimensional coordinates of the disease were determined, and a list of vertex IDs of the abnormal area was obtained.

[0036] Textures are dynamically generated based on the probability distribution of diseases, resulting in PNG sequence frames with transparency. Pulse frequencies are set according to the probability to obtain Unity particle parameters. Textures, lighting effects, and risk icons are then overlaid on the marked positions in the digital twin to obtain 3D early warning markers.

[0037] As a preferred embodiment of the digital twin-based ancient tree health status monitoring method of the present invention, the method involves: fusing the mycorrhizal symbiosis index and three-dimensional early warning markers into the initial three-dimensional model of the ancient tree to generate an ancient tree health status report, including the following steps.

[0038] The ICP algorithm is used to align the vertices of the 3D heatmap with the initial 3D model of the ancient tree, output the vertex attribute table of spatial matching, and evaluate the health of local areas of the ancient tree.

[0039] The vertex-level health score is partitioned according to the ancient tree structure to obtain the aggregated regional health index;

[0040] Based on the health status score, an HSL gradient is generated and its transparency is superimposed to obtain an RGBA vertex color array;

[0041] Mechanical markings were applied to ancient trees to add fracture risk icons, and risk marker coordinates were obtained.

[0042] By integrating regional health indicators, risk-marked coordinates, and disease probability distribution, a health status report of ancient trees is obtained.

[0043] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the ancient tree health status monitoring method based on digital twins as described in the first aspect of the present invention.

[0044] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the method for monitoring the health status of ancient trees based on digital twins as described in the first aspect of the present invention.

[0045] The beneficial effects of this invention are as follows: The mycorrhizal symbiosis index detection system uses a 45° tilted microelectrode array to non-destructively monitor root ion flow rate. Combined with an inverse distance weighted interpolation algorithm, a three-dimensional heat map is generated, achieving dynamic visualization of underground mycorrhizal activity and providing precise spatial guidance for root maintenance. A multi-channel electronic nose array collects volatile organic compound (VOC) fingerprints, and graph convolutional networks and knowledge graph matching are used to achieve early disease identification. Furthermore, a probability-driven three-dimensional early warning marker is innovatively developed, linking disease threat level with spatial location. This collaborative application constructs a full-dimensional digital twin model of underground-above-ground health status. Through spatiotemporal correlation analysis of mycorrhizal activity and VOC characteristics, a technological leap from experience-based maintenance to data-driven precision intervention is achieved, significantly improving the early warning capability for ancient tree diseases and the scientific nature of maintenance decisions. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a flowchart of a method for monitoring the health status of ancient trees based on digital twins.

[0048] Figure 2 This is a schematic diagram of a three-dimensional early warning marker.

[0049] Figure 3 This is a schematic diagram of a health status report for ancient trees.

[0050] Figure 4 A schematic diagram for calculating the mycorrhizal symbiosis index. Detailed Implementation

[0051] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0052] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0053] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0054] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a method for monitoring the health status of ancient trees based on digital twins, including the following steps:

[0055] S1. Collect point cloud data and texture data of ancient trees to construct an initial 3D model of the ancient trees.

[0056] S1.1 Use LiDAR to scan the trunk and main branches to obtain point cloud data, and use a hyperspectral camera to acquire hyperspectral images to obtain texture data.

[0057] Furthermore, the ancient tree was scanned at multiple sites using a RIEGL VZ-400i lidar with a point pitch of 0.5mm to obtain point cloud data containing three-dimensional coordinates and reflection intensity. At the same time, a HySpex VNIR-1800 hyperspectral camera was used to simultaneously acquire reflectivity image data in the 400-1000nm band to obtain texture data.

[0058] S1.2 Register the point cloud data using the ICP algorithm to generate a triangular mesh model, and map the hyperspectral data into a texture map of the triangular mesh model to obtain an initial 3D model of the ancient tree with multispectral texture.

[0059] Furthermore, the iterative nearest-point algorithm is used to register the multi-site point cloud data acquired by the lidar scan. By minimizing the distance between corresponding points, the rotation matrix and translation vector are optimized to achieve spatial alignment of the point cloud data. Based on the registered point cloud data, the Poisson surface reconstruction algorithm is applied to generate a closed triangular mesh model. The surface reconstruction parameters are set, including reconstruction depth and sampling interval. The reflectivity image data of the 400-1000nm band acquired by the hyperspectral camera is reduced to the visible light RGB channel. The texture coordinates are calculated by the least squares conformal mapping algorithm, and the hyperspectral texture is accurately mapped to the surface of the triangular mesh model to obtain the initial three-dimensional model of the ancient tree with multispectral texture.

[0060] S2. Deploy a micro-ion-selective electrode array in the root distribution area of ​​ancient trees and detect Ca using a potentiostat.2 and K + Changes in ion flow rate were combined with soil temperature and humidity data to calculate the mycorrhizal symbiosis index.

[0061] S2.1. Deploy micro-ion electrodes in a ring with the tree trunk as the center radius, insert them into the soil at a 45° angle, and obtain the spatial coordinate information of the electrodes.

[0062] Furthermore, with the base of the tree trunk as the center, eight micro ion-selective electrodes were evenly distributed on a circumference with a radius of 1.5 meters. Each electrode was inserted into the soil layer at three depth gradients (30 cm, 50 cm, and 80 cm) with a 45-degree tilt angle. The three-dimensional spatial coordinates of each electrode were accurately measured using a Trimble R12 real-time dynamic measuring instrument, with the measurement error controlled within 5 mm. The electrode insertion process avoided the main root system with a diameter greater than 2 cm. After insertion, the potential was calibrated on-site using a standard solution. Finally, an electrode spatial coordinate information table containing the electrode number, spatial coordinates, and calibration parameters was obtained.

[0063] S2.2, Use a potentiostat to collect Ca 2+ and K + The current signal is used to obtain the original current-time series data. Wavelet denoising is performed on the current data to generate the ion activity time series curve.

[0064] Furthermore, a PalmSens4 potentiostat was used to continuously acquire Ca at a sampling frequency of 1Hz. 2+ The current signal from the / K+ ion selective electrode was sampled for 5 minutes at a time to obtain raw current-time series data containing timestamps and current values. The raw current data was decomposed using the Daubechies 5 wavelet basis in three levels, and high-frequency noise components were filtered out using a hard thresholding method. The threshold was set to √(2lnN) times the noise standard deviation (N is the number of sampling points). The reconstructed signal generated a smooth ion activity time-series curve, and the curve data included Ca... 2+ and K + The activity variation trend at three soil depths of 30 cm, 50 cm, and 80 cm was analyzed. The noise reduction process effectively eliminated environmental electromagnetic interference and sensor background noise, preserved the true ion migration signal characteristics, and generated ion activity time series curves.

[0065] S2.3. Based on the Nernst equation, replace the ion activity time series curve with ion flow rate to obtain Ca. 2+ And the K+ spatial distribution velocity field.

[0066] Furthermore, based on the Nernst equation, the wavelet-denoised ion activity time-series curve is converted into ion flow rate, and for each electrode point, Ca... 2+ and K +The ion flow rates were obtained by applying the modified Nernst-Planck equation to the data. Spatial interpolation was then used to transform the flow rate data at discrete electrode points into a continuous three-dimensional flow field, establishing a system including X / Y / Z coordinates and Ca... 2 + / K + The table shows the spatial distribution of flow velocity values, and the table shows a strict correspondence between ion flow velocity and electrode spatial coordinates.

[0067] S2.4. Use the Decagon 5TE sensor to collect soil volumetric water content and temperature, and calculate the mycorrhizal symbiosis index.

[0068] Specifically, the expression is,

[0069]

[0070] Wherein, MSI is the mycorrhizal symbiosis index, J K For potassium ion flow rate, J Ca θ represents the calcium ion flow rate, θ represents the soil volumetric water content, T represents the soil temperature, K represents potassium ions, and Ca represents calcium ions.

[0071] S3. Upload the mycorrhizal symbiosis index to the digital twin and display the activity status of the root system area in the form of a three-dimensional heat map in the initial three-dimensional model of the ancient tree.

[0072] S3.1 Based on the mycorrhizal symbiotic index, the inverse distance weighting method is used to complete the missing electrode data to obtain a complete electrode dataset.

[0073] Specifically, the expression is,

[0074]

[0075] Among them, MSI interp MSI is the estimated value of the interpolation point. i Let d be the measured MSI value of the i-th electrode. i Let be the distance from the point to be interpolated to the i-th electrode, where i is the electrode index.

[0076] S3.2 Align the electrode coordinates with the root mesh vertex space in the initial 3D model of the ancient tree to generate a vertex list with MSI attributes.

[0077] Furthermore, an iterative nearest-point algorithm is used to register the spatial coordinates of the micro ion-selective electrode with the root mesh vertices of the initial 3D model of the ancient tree. By minimizing the distance between the electrode measurement point and the nearest mesh vertex, the rigid body transformation parameters are optimized, and the registration error is controlled within 2 mm. For each root mesh vertex, the mycorrhizal symbiosis index interpolation is obtained based on inverse distance weighted interpolation, with the interpolation radius limited to 1 meter. An attribute list containing vertex number, 3D coordinates, and mycorrhizal symbiosis index value is generated. The list data strictly corresponds to the topology of the 3D model in GLTF 2.0 format.

[0078] It should be noted that the motor coordinates are obtained by deploying a micro ion-selective electrode array in a ring around the root distribution area with the trunk as the center, according to the ring distribution pattern (including radial distance and angular interval) and the installation parameters of inserting it into the soil at a 45° tilt angle, combined with the three-dimensional spatial position data of each electrode obtained by RTK positioning technology, and then registering and aligning it with the coordinate system of the initial three-dimensional model of the ancient tree.

[0079] S3.3 Map the vertex list of the MSI attribute to the HSL color space to obtain the vertex color encoding array. Render the vertex colors using Unity Shader Graph to obtain a 3D heatmap with transparency adjustment. Add ray detection feet to the 3D heatmap area to obtain an interactive digital twin.

[0080] Furthermore, the list of vertices with mycorrhizal symbiosis index attributes is mapped to color values ​​using the HSL color space conversion algorithm. The hue parameter H varies linearly in the range of 0-120 degrees according to the mycorrhizal symbiosis index value (0 degrees for red to 120 degrees for green), the saturation S is fixed at 100%, and the lightness L is fixed at 50%. A custom shader is created in Unity Shader Graph to receive vertex color data and apply an opacity blending algorithm to generate a 3D heatmap with a gradient of opacity. A Unity Raycast ray detection component is added to the 3D heatmap mesh. When the user interacts, it returns the ID of the hit vertex and queries the associated mycorrhizal symbiosis index data. The specific values ​​and health status assessment results are displayed on the UI interface, resulting in an interactive digital twin.

[0081] S4. Install a 16-channel MOX gas sensor electronic nose array on the surface of the ancient tree trunk to collect data on the concentration of volatile organic compounds (VOCs) in the ancient tree.

[0082] The S4.1 uses a 16-channel MOX gas sensor electronic nose, which is spirally arranged on the trunk and connected to a canopy solar panel equipped with a supercapacitor for energy storage. It is driven by a constant current to obtain the original resistance value.

[0083] Furthermore, a 16-channel MOX gas sensor electronic nose array is spirally arranged within a height range of 1.3-1.7 meters on the trunk surface of the ancient tree. Adjacent sensors are spaced 20 centimeters apart and cover four main directions. Each MOX gas sensor electronic nose is connected to a 10-watt solar panel installed in the canopy layer via a waterproof connector. A supercapacitor bank provides continuous power supply. A constant current drive of 1.8 volt heating voltage and 5 volt sensing voltage is applied to the MOX gas sensor electronic nose. The resistance changes of the semiconductor sensitive film of each channel are collected in real time through a 16-bit analog-to-digital converter, and the raw resistance value sequence containing timestamps and channel numbers is output. During the acquisition process, the ambient temperature and humidity parameters are automatically monitored. Data recording is paused when the relative humidity exceeds 90% or the temperature exceeds the range of 0-50 degrees Celsius. The raw resistance value sequence is transmitted to the edge computing node via the LoRaWAN wireless transmission protocol, with a transmission cycle of 2 hours per session and a sampling duration of 5 minutes per session. The spiral arrangement of the MOX gas sensor electronic nose ensures full spatial coverage detection of the volatile organic compound release characteristics on the trunk surface to obtain the raw resistance values.

[0084] S4.2 Convert the original resistance value into VOC concentration data of ancient trees.

[0085] Specifically, the expression is,

[0086]

[0087] Among them, C m Y represents the concentration data of volatile organic compounds (VOCs) from ancient trees in the m-th MOX gas sensor channel. m Rm is the calibration coefficient for the m-th MOX gas sensor channel, R0 is the resistance value in the target gas of the MOX gas sensor, and Rm is the resistance value in the target gas of the MOX gas sensor. s α is the reference resistance in clean air for the MOX gas sensor. m Let be the nonlinear response index of the m-th MOX gas sensor channel, where m is the MOX gas sensor channel index.

[0088] S5. Input the VOC concentration data into the graph convolutional network (GCN) model, match the VOC feature patterns of ancient tree diseases in the ancient tree disease knowledge graph, and obtain the disease types and probability distributions.

[0089] S5.1 Normalize the VOC concentration data of ancient trees to obtain normalized VOC concentration data of ancient trees, and extract features to obtain the time-series features of VOC concentration.

[0090] Furthermore, the minimum-maximum normalization method was used to process the VOC concentration data of ancient trees collected by the electronic nose of the 16-channel MOX gas sensor. The normalization formula is as follows. Three types of time-series features were extracted from the normalized VOC concentration data of ancient trees. The mean calculated by the sliding window reflects the concentration baseline level, the slope of the first-order difference sequence represents the trend of change, and the main frequency component obtained by the fast Fourier transform indicates the periodic release characteristics, thus obtaining the time-series features of VOC concentration.

[0091] S5.2. Generate a disease association graph based on the time-series features of VOC concentration and knowledge graph, and obtain the adjacency matrix and node feature matrix.

[0092] Furthermore, a heterogeneous graph structure was constructed, comprising 16 MOX gas sensor electronic nose nodes and 327 disease nodes. The sensor node features are 48-dimensional VOC concentration time-series feature vectors (mean / slope / dominant frequency), while the disease node features are predefined one-hot encodings in the knowledge graph. Two types of edge relationships were defined, and the MOX gas sensor edge weights were calculated inversely proportional to the spatial distance. The MOX gas sensor-disease edge weights adopted the conditional probabilities stored in the knowledge graph. The node connection relationships were encoded through an adjacency matrix, and the node feature matrix integrated the time-series features and category features.

[0093] S5.3. Aggregate the node feature matrix through 3-layer GATv2 graph convolution to obtain the high-order graph embedding vector, and calculate the cosine similarity between the high-order graph embedding vector and the VOC feature pattern of ancient tree diseases in the knowledge graph.

[0094] Specifically, the expression is,

[0095]

[0096] Where S is the similarity score, f GCN f is the high-order feature vector output by GCN. 图谱 This refers to the VOC feature pattern vectors pre-stored in the knowledge graph.

[0097] S5.4 The Softmax layer based on the GCN model transforms the high-order graph embedding vector into a disease probability distribution.

[0098] Furthermore, the high-order graph generated by three layers of GATv2 graph convolution is embedded into the fully connected layer as a vector input, and the normalization exponent is calculated using the Softmax function.

[0099]

[0100] Where P(y) is the conditional probability of the disease category y to be predicted, h is the higher-order graph embedding vector, and W y W is the weight vector of the disease category y to be predicted.g Let y be the weight vector for the g-th disease, where G is the total number of disease categories, y is the disease category label to be predicted, and k is the disease category index.

[0101] It should be noted that the output is the probability distribution vector of each disease node. The vector dimension corresponds completely to the disease types in the ancient tree disease knowledge graph. The transformation process retains the topological features learned by graph convolution.

[0102] S6. Detect the types and probability distribution of diseases, mark their locations in the digital twin, and obtain three-dimensional early warning markers.

[0103] S6.1. Based on the disease type, retrieve the three-dimensional annotation rules from the knowledge graph to obtain the rendering template parameters.

[0104] Furthermore, based on the disease types with a probability value greater than 85% in the disease probability distribution, the predefined disease-rendering rule mapping table in the ancient tree disease knowledge graph is queried to obtain the corresponding 3D annotation parameter combination. For canker disease, the reddish-brown crack normal map path, pulse warning light effect frequency parameter and transparency curve are extracted. For trunk borer, the 3D model file of the insect tunnel and the yellow warning particle parameter are called. The rendering template parameters include three types of data: texture map reference path, dynamic light effect control parameter and risk level annotation style.

[0105] S6.2 Combine the VOC concentration data of ancient trees with the three-dimensional heat map to determine the three-dimensional coordinates of the disease and obtain a list of vertex IDs of the abnormal area.

[0106] Furthermore, the VOC concentration data of ancient trees collected by the electronic nose of the 16-channel MOX gas sensor is mapped onto the surface of the initial three-dimensional model of the ancient trees using a spatial interpolation algorithm, generating a three-dimensional gradient field of VOC concentration distribution. The spatial coordinates of the VOC concentration gradient field and the three-dimensional heat map of mycorrhizal symbiosis index are compared vertex by vertex. When the VOC concentration exceeds the disease characteristic threshold in the knowledge graph and the mycorrhizal symbiosis index at the corresponding position is <0.4, the vertex is marked as an abnormal region. The vertex numbers and three-dimensional coordinates of all abnormal regions are extracted to generate a list of abnormal region vertex IDs containing vertex ID, coordinate position, associated VOC type and concentration value.

[0107] S6.3 Dynamically generate textures according to the probability distribution of diseases to obtain PNG sequence frames with transparency. Set the pulse frequency according to the probability to obtain Unity particle parameters. Overlay textures, lighting effects and risk icons on the marked positions of the digital twin to obtain three-dimensional early warning markers.

[0108] Furthermore, based on the probability distribution values ​​of diseases, predefined texture generation rules in the knowledge graph are invoked to generate semi-transparent yellow crack textures for disease areas with a probability value between 70% and 90%, and reddish-brown pulse textures for disease areas with a probability value > 90%. Particle units are created in the Unity engine, and particle emission parameters are set according to the disease type. Ulcer diseases correspond to ring pulse particles, and borers correspond to insect tunnel trajectory particles. The texture maps and particle effects are matched with the coordinates in the vertex ID list of abnormal areas, and superimposed and rendered at the corresponding mesh vertex positions of the digital twin. At the same time, a mechanical risk icon is added (a fracture warning symbol is displayed when the stress σ in the finite element analysis is > 50MPa), resulting in a three-dimensional early warning marker.

[0109] S7. Integrate the mycorrhizal symbiosis index and three-dimensional early warning markers into the initial three-dimensional model of the ancient tree to generate a health status report of the ancient tree.

[0110] S7.1 Align the vertices of the 3D heatmap with the initial 3D model of the ancient tree using the ICP algorithm, output the vertex attribute table of spatial matching, and evaluate the health status of local areas of the ancient tree.

[0111] Specifically, the expression is,

[0112] H v =w1·MSI v +w2·(1-P disease ) v ;

[0113] Among them, H v The health status score for vertex v, MSI v w1 is the mycorrhizal symbiosis index of vertex v, w2 is the MSI weight coefficient, and v is the inverse weight coefficient of disease probability.

[0114] S7.2. The vertex-level health score is partitioned according to the ancient tree structure to obtain the aggregated regional health index.

[0115] Furthermore, based on the grid labels of the initial 3D model of the ancient tree, the regions are divided (root region, trunk region, and main branch region). When aggregating the root region, grid vertices with a depth greater than 1.5 meters are excluded, and the trunk region is limited to a grid range with an elevation of 0.3-3 meters. This generates an aggregated region health index table that includes the average health score, the lowest health score, and the percentage of abnormal vertices for each region.

[0116] S7.3. Generate an HSL gradient based on the health status score and overlay the transparency to obtain an RGBA vertex color array.

[0117] Furthermore, the vertex-level health score is mapped to color values ​​using the HSL color space conversion algorithm. The hue changes linearly in the range of 0-120 degrees, the saturation is fixed at 100%, and the brightness is fixed at 50%. This generates a vertex color array containing four RGBA channels. The vertex order of the output array is completely consistent with the mesh vertex index of the initial 3D model of the ancient tree, resulting in an RGBA vertex color array.

[0118] S7.4. Add fracture risk icons to the ancient trees by mechanical marking, and obtain the risk marking coordinates.

[0119] Furthermore, based on point cloud data, geometric morphology data of the trunk and main branches of the ancient tree are obtained through a 3D laser scanner. The 3D point cloud data is then imported into finite element analysis software to establish a mechanical model of the ancient tree. Material parameters are set, such as an elastic modulus of 1000 MPa and a Poisson's ratio of 0.3. Wind loads are applied to the mechanical model of the ancient tree, such as a wind speed of 20 m / s, to obtain the stress distribution and displacement of various parts of the ancient tree. High-risk areas are determined based on the stress results. When the maximum principal stress of a certain part exceeds the tensile strength of the material, such as 5 MPa, it is determined to be a fracture risk area. The center coordinates of the fracture risk area are located in the 3D point cloud data. A circular area with a radius of 20 cm is generated with this coordinate as the center point as the risk label coordinate. Finally, a fracture risk icon is superimposed and displayed at the risk label coordinate position to obtain the risk label coordinate.

[0120] S7.5 Integrate the health indicators, risk marker coordinates, and disease probability distribution of the aggregation area to obtain a health status report of ancient trees.

[0121] Furthermore, numerical scores of health indicators for the aggregated area are extracted, such as a score of 0-10 for trunk decay and 0-5 for sparse foliage. Risk-marked coordinates are mapped onto a 3D point cloud model of the ancient tree, and the number and density of high-risk areas are statistically analyzed; for example, three fracture risk areas are detected on a single ancient tree. Disease probability distribution data is read, and the probability of occurrence for each disease type is extracted, such as 70% for decay and 30% for insect infestation. A weighted method is used, with 40% weight for the aggregated area health indicators, 30% for the risk-marked coordinates, and 30% for the disease probability distribution, to calculate the comprehensive health index of the ancient tree. Based on the comprehensive health index, health levels are classified, for example, 0-3 for dangerous, 4-6 for sub-healthy, and 7-10 for healthy. Finally, a health status report of the ancient tree is generated, including text descriptions and visual illustrations. The text records specific indicator values ​​and assessment conclusions, while the illustrations mark risk locations and disease distribution heatmaps.

[0122] This embodiment also provides a computer device applicable to the ancient tree health status monitoring method based on digital twins, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the ancient tree health status monitoring method based on digital twins as proposed in the above embodiment.

[0123] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0124] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for monitoring the health status of ancient trees based on digital twins as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0125] In summary, this invention achieves dynamic visualization of underground mycorrhizal activity by using a mycorrhizal symbiosis index detection system with a 45° tilted microelectrode array for non-destructive monitoring of root ion flow rate, combined with an inverse distance weighted interpolation algorithm to generate a three-dimensional heat map, providing precise spatial guidance for root maintenance. It also collects volatile organic compound (VOC) fingerprints using a multi-channel electronic nose array, utilizes graph convolutional networks and knowledge graph matching to achieve early disease identification, and innovatively develops probability-driven three-dimensional early warning markers that correlate disease threat levels with spatial location. This collaborative application constructs a full-dimensional digital twin model of underground-above-ground health status. Through spatiotemporal correlation analysis of mycorrhizal activity and VOC characteristics, it achieves a technological leap from experience-based maintenance to data-driven precision intervention, significantly improving the early warning capability for ancient tree diseases and the scientific nature of maintenance decisions.

[0126] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for monitoring the health status of ancient trees based on digital twins, characterized in that: include, Collect point cloud data and texture data of ancient trees to construct an initial 3D model of the ancient trees; A miniature ion-selective electrode array was deployed in the root distribution area of ​​ancient trees, and Ca was detected using a potentiostat. 2 The mycorrhizal symbiosis index was calculated by combining changes in K+ ion flow rate with soil temperature and humidity data. The mycorrhizal symbiosis index is uploaded to the digital twin, and the activity status of the root system area is displayed in the form of a three-dimensional heat map in the initial three-dimensional model of the ancient tree. A 16-channel MOX gas sensor electronic nose array was installed on the surface of the ancient tree trunk to collect data on the concentration of volatile organic compounds (VOCs) in the ancient tree. VOC concentration data is input into the graph convolutional network (GCN) model, matched with the VOC feature patterns of ancient tree diseases in the ancient tree disease knowledge graph, to obtain disease types and probability distributions. Disease types and probability distributions are detected, and their locations are marked in a digital twin to obtain three-dimensional early warning markers. The mycorrhizal symbiosis index and three-dimensional early warning markers are integrated into the initial three-dimensional model of the ancient tree to generate a health status report of the ancient tree.

2. The method for monitoring the health status of ancient trees based on digital twins as described in claim 1, characterized in that: The process of collecting point cloud and texture data of ancient trees and constructing an initial 3D model of the ancient trees includes the following steps. The tree trunk and main branches are scanned using LiDAR to obtain point cloud data, and hyperspectral images are acquired using a hyperspectral camera to obtain texture data. Point cloud data is registered using the ICP algorithm to generate a triangular mesh model. Hyperspectral data is then mapped to a texture map of the triangular mesh model to obtain an initial 3D model of the ancient tree with multispectral texture.

3. The method for monitoring the health status of ancient trees based on digital twins as described in claim 2, characterized in that: A miniature ion-selective electrode array was deployed in the root distribution area of ​​ancient trees, and Ca was detected using a potentiostat. 2 The mycorrhizal symbiosis index is calculated based on changes in K+ ion flux and soil temperature and humidity data, including the following steps. Micro-ion electrodes were deployed in a ring around the tree trunk, and inserted into the soil at a 45° angle to obtain the spatial coordinate information of the electrodes. Ca was collected using a potentiostat. 2+ and K + The current signal is used to obtain the raw current-time series data. Wavelet denoising is performed on the current data to generate the ion activity time series curve. By replacing the ion activity time series curve with ion flow rate according to the Nernst equation, we obtain Ca 2+ and K + Spatial distribution of velocity field; The Decagon 5TE sensor was used to collect soil volumetric water content and temperature, and the mycorrhizal symbiosis index was calculated; specifically.

4. The method for monitoring the health status of ancient trees based on digital twins as described in claim 3, characterized in that: The mycorrhizal symbiosis index is uploaded to a digital twin, and the root system activity status is displayed in the form of a 3D heatmap in the initial 3D model of the ancient tree. This includes the following steps. Based on the mycorrhizal symbiosis index, the inverse distance weighting method was used to complete the missing electrode data and obtain a complete electrode dataset. Align the electrode coordinates with the root mesh vertex space in the initial 3D model of the ancient tree to generate a vertex list with MSI attributes; Map the vertex list of MSI attributes to the HSL color space to obtain a vertex color encoding array. Render the vertex colors using UnityShader Graph to obtain a 3D heatmap with transparency adjustment. Add ray detection feet to the 3D heatmap area to obtain an interactive digital twin.

5. The method for monitoring the health status of ancient trees based on digital twins as described in claim 4, characterized in that: A 16-channel MOX gas sensor electronic nose array was installed on the surface of the ancient tree trunk to collect data on the concentration of volatile organic compounds (VOCs) in the ancient tree. Includes the following steps, A 16-channel MOX gas sensor electronic nose is used. The MOX gas sensor electronic nose is spirally arranged on the trunk, connected to the canopy solar panel equipped with a supercapacitor for energy storage, and driven by constant current to obtain the original resistance value. The original resistance values ​​were converted into VOC concentration data of ancient trees.

6. The method for monitoring the health status of ancient trees based on digital twins as described in claim 5, characterized in that: VOC concentration data is input into a graph convolutional network (GCN) model, and matched with VOC feature patterns of ancient tree diseases in the ancient tree disease knowledge graph to obtain disease types and probability distributions. This includes the following steps: The VOC concentration data of ancient trees were normalized to obtain normalized VOC concentration data of ancient trees, and feature extraction was performed to obtain the time-series features of VOC concentration. Based on the temporal characteristics of VOC concentration and knowledge graph, a disease association graph is generated, and the adjacency matrix and node feature matrix are obtained. The node feature matrix is ​​aggregated by 3 layers of GATv2 graph convolution to obtain the high-order graph embedding vector. The cosine similarity between the high-order graph embedding vector and the VOC feature pattern of ancient tree diseases in the knowledge graph is calculated. The Softmax layer based on the GCN model transforms the high-order graph embedding vector into a disease probability distribution.

7. The method for monitoring the health status of ancient trees based on digital twins as described in claim 6, characterized in that: The detection of disease types and probability distributions, location marking on a digital twin, and the generation of three-dimensional early warning markers include the following steps. The rendering template parameters are obtained by retrieving the 3D annotation rules from the knowledge graph based on the disease type. By combining the VOC concentration data of ancient trees with the three-dimensional heat map, the three-dimensional coordinates of the disease were determined, and a list of vertex IDs of the abnormal area was obtained. Textures are dynamically generated based on the probability distribution of diseases, resulting in PNG sequence frames with transparency. Pulse frequencies are set according to the probability to obtain Unity particle parameters. Textures, lighting effects, and risk icons are then overlaid on the marked positions in the digital twin to obtain 3D early warning markers.

8. The method for monitoring the health status of ancient trees based on digital twins as described in claim 7, characterized in that: The mycorrhizal symbiosis index and three-dimensional early warning markers are integrated into the initial three-dimensional model of the ancient tree to generate a health status report, including the following steps. The ICP algorithm is used to align the vertices of the 3D heatmap with the initial 3D model of the ancient tree, output the vertex attribute table of spatial matching, and evaluate the health status of local areas of the ancient tree. The vertex-level health score is partitioned according to the ancient tree structure to obtain the aggregated regional health index; Based on the health status score, an HSL gradient is generated and its transparency is superimposed to obtain an RGBA vertex color array; Mechanical markings were applied to ancient trees to add fracture risk icons, and risk marker coordinates were obtained. By integrating regional health indicators, risk-marked coordinates, and disease probability distribution, a health status report of ancient trees is obtained.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the ancient tree health status monitoring method based on digital twins as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the ancient tree health status monitoring method based on digital twins as described in any one of claims 1 to 8.

Images