Whole-process forest tending inspection evaluation simulation method and system

Abstract

The application provides a whole-process forest tending inspection evaluation simulation method and system, relates to the forestry informatization technical field, and comprises the following steps: constructing a forest development stage knowledge graph, calculating a quality index value and matching a development stage based on ground measurement data, generating a weight coefficient group by using a graph structure connection degree group and a time sequence position parameter, calculating a quality evaluation value and an operation effect value, establishing a time sequence evolution edge, extracting a historical time sequence evolution path to predict a target development stage, and generating a stand regulation scheme. The application realizes dynamic evaluation and decision optimization in the whole process of forest tending, and improves the sustainable management level of forests.

Description

TECHNICAL FIELD

[0001] The present application relates to forestry informatization technology, and in particular to a whole-process forest tending inspection evaluation simulation method and system. BACKGROUND

[0002] Forest tending is an important link in sustainable forest management, and is of great significance for optimizing stand structure, improving forest quality and promoting tree growth.

[0003] The existing forest tending inspection evaluation technology has the following shortcomings: lack of comprehensive expression of semantic association between forest development stages and quality indicators, resulting in failure to fully consider the dynamic change characteristics of stand development in the tending inspection evaluation process, and the evaluation results often focus on static indicators and ignore the key factors in the development stage conversion process. The existing evaluation method usually uses a fixed weight system to calculate the weighted sum of each indicator, and fails to dynamically adjust according to different development stages and stand characteristics, making it difficult to accurately reflect the tending needs and quality performance of the stand at different growth stages. The existing tending decision mainly relies on experience, lacks a scientific decision support system based on historical data and prediction models, and is difficult to achieve scientific prediction of future stand development and planning of optimal tending path, resulting in lack of pertinence and foresight in the implementation of tending measures. SUMMARY

[0004] The embodiments of the present application provide a whole-process forest tending inspection evaluation simulation method and system, which can solve the problems in the prior art.

[0005] In a first aspect, the embodiments of the present application provide a whole-process forest tending inspection evaluation simulation method, comprising:

[0006] Obtaining ground measurement data of a target stand, constructing a forest development stage knowledge graph including development stage entity nodes and quality indicator entity nodes, and establishing semantic association edges between the entity nodes; calculating quality indicator values based on the ground measurement data, matching the current development stage node in the knowledge graph, and traversing the semantic association edges to obtain a set of semantic association paths;

[0007] Counting the number of semantic association edges connected by the quality indicator entity nodes in the set of semantic association paths to obtain a graph structure connection degree group, extracting a time sequence position parameter of the current development stage node, generating a dynamic weight coefficient group based on the graph structure connection degree group and the time sequence position parameter, and weighting and summing the quality indicator values to obtain a quality evaluation value;

[0008] Calculating the operation effect value before and after the tending operation;

[0009] Adding the quality evaluation value and the operation effect value as time identifier attributes to the current development stage node, creating a time sequence evolution edge to the adjacent development stage node, and determining the weight of the time sequence evolution edge by the operation effect value.

[0010] The knowledge graph is used to predict the target development stage node of the historical time sequence evolution path, and the optimal evolution path from the current development stage node to the target development stage node is queried to generate a forest regulation scheme.

[0011] The ground measurement data of the target forest is obtained, a forest development stage knowledge graph is constructed, and semantic association edges are established between the entity nodes; based on the ground measurement data, the quality index value is calculated, and the current development stage node is matched in the knowledge graph, and the semantic association path set is obtained by traversing the semantic association edges.

[0012] The ground measurement data includes diameter at breast height data, tree height data and under-bough height data;

[0013] The development stage entity node set is arranged in the order of the forest growth cycle, and the quality index entity node set includes a diameter class index entity node, a first structure index entity node and a second structure index entity node;

[0014] The semantic association edges are established between the development stage entity nodes and the quality index entity nodes by extracting the knowledge rules of forestry experts, and the semantic association edges carry the applicability weight attribute;

[0015] The diameter class index value is calculated by the diameter at breast height data, the first structure index value is calculated by the ratio of the tree height data to the diameter at breast height data, and the second structure index value is calculated by the ratio of the under-bough height data to the tree height data;

[0016] The diameter class index value, the first structure index value and the second structure index value are matched with the quality index entity nodes respectively, and the current development stage node is obtained by backtracking through the semantic association edges of the matched quality index entity nodes;

[0017] The semantic association edges are traversed from the current development stage node to the quality index entity nodes, and the traversal path is composed of the semantic association path set.

[0018] The dynamic weight coefficient set is generated based on the graph structure connection degree group and the time sequence position parameter, and the quality evaluation value is obtained by weighted sum of the quality index value;

[0019] The single-hop connection degree and the double-hop connection degree of the quality index entity nodes in the semantic association path set are counted, and the graph structure connection degree group is obtained by summing after applying the attenuation factor respectively;

[0020] obtain an evolution trend coefficient by extracting a historical quality evaluation value sequence of the current development stage node and calculating a change rate thereof, modulate the basic time sequence position value by taking the evolution trend coefficient as a modulation factor to obtain a time sequence position parameter;

[0021] multiply the graph structure connection degree group and the time sequence position parameter to obtain a time sequence modulated connection degree group, obtain a synergistic relationship vector through an eigenvector of a covariance matrix of the time sequence modulated connection degree group, and weight and fuse the time sequence modulated connection degree group and the synergistic relationship vector to obtain a dynamic weight coefficient group after normalization;

[0022] weight and combine the quality index value by the dynamic weight coefficient group to obtain an initial quality evaluation value, and adjust the confidence degree of the initial quality evaluation value by the time sequence position parameter to obtain a quality evaluation value.

[0023] The step of modulating the basic time sequence position value by taking the evolution trend coefficient as a modulation factor to obtain a time sequence position parameter comprises:

[0024] calculate a short-term trend coefficient by calculating a first-order difference mean of a short-term window of the historical quality evaluation value sequence, calculate a long-term trend coefficient by calculating a linear fitting slope of a long-term window, and calculate an evolution acceleration coefficient by calculating a second-order difference of a full sequence;

[0025] calculate a deviation degree of the short-term trend coefficient, the long-term trend coefficient, and the evolution acceleration coefficient from a historical mean, fuse a normal trend component with an abnormal trend component after smoothing processing of the abnormal trend component whose deviation degree exceeds a preset deviation threshold to obtain a robust evolution feature vector;

[0026] obtain a modulation sensitivity factor based on a stage attribute parameter of the current development stage node, select a nonlinear modulation function from a preset modulation function library according to the modulation sensitivity factor, and obtain a modulation intensity parameter by a modulus value of the robust evolution feature vector;

[0027] take each component of the robust evolution feature vector as a modulation factor, and modulate the basic time sequence position value in multiple levels according to the selected nonlinear modulation function, and weight and fuse each level of modulation result by the modulation intensity parameter to obtain the time sequence position parameter.

[0028] add the quality evaluation value and the operation effect value to the current development stage node as a time identifier attribute, and create a time sequence evolution edge to an adjacent development stage node, the step comprising:

[0029] obtain ground measurement data before and after the rearing operation, and calculate an operation-before quality evaluation value and an operation-after quality evaluation value respectively, and calculate a difference between the operation-before quality evaluation value and the operation-after quality evaluation value as the operation effect value;

[0030] calculate a quality evolution trend coefficient based on the quality evaluation value of the current development stage node and the historical quality evaluation value sequence, and encapsulate the quality evaluation value, the operation effect value and the quality evolution trend coefficient as time identifier attributes added to the current development stage node;

[0031] identify candidate development stage nodes adjacent to the current development stage node in the development time sequence from the development stage entity node set, calculate a development stage span parameter between each candidate development stage node and the current development stage node, and select adjacent development stage nodes based on the development stage span parameter;

[0032] calculate a basic edge weight value according to the operation effect value and the quality evolution trend coefficient, calculate a historical quality fluctuation coefficient by extracting the historical time identifier attributes of the adjacent development stage nodes, and obtain an adjusted edge weight value by correcting the basic edge weight value through the historical quality fluctuation coefficient;

[0033] create a time evolution edge between the current development stage node and each adjacent development stage node, and assign the adjusted edge weight value to the time evolution edge as an edge weight.

[0034] The step of obtaining an adjusted edge weight value by correcting the basic edge weight value through the historical quality fluctuation coefficient comprises:

[0035] calculate a current quality contribution factor by weightedly fusing the operation effect value and the quality evolution trend coefficient, and calculate the basic edge weight value based on the ratio of the current quality contribution factor to a preset reference factor;

[0036] extract a historical quality evaluation value sequence from the historical time identifier attributes of the adjacent development stage nodes, calculate a historical quality fluctuation amplitude by calculating the standard deviation of the historical quality evaluation value sequence, and calculate the historical quality fluctuation coefficient based on the ratio of the historical quality fluctuation amplitude to the mean value of the historical quality evaluation value;

[0037] construct an edge weight correction function, input the historical quality fluctuation coefficient as a correction factor into the edge weight correction function, correct the basic edge weight value to obtain an adjusted edge weight value, and the adjusted edge weight value and the historical quality fluctuation coefficient are in an inverse relationship.

[0038] The step of extracting a target development stage node of a historical time evolution path in a knowledge graph and querying a forest stand regulation scheme generation method of a nurturing measure node on an optimal evolution path from a current development stage node to the target development stage node comprises:

[0039] extract all historical time evolution paths containing time evolution edges from the knowledge graph, calculate the historical cumulative values of the time evolution edge weights on each historical time evolution path, and take the end point development stage node of the historical time evolution path with the largest historical cumulative value as the target development stage node.

[0040] searching for candidate evolution paths in the knowledge graph with the current development stage node as the starting point and the target development stage node as the end point, calculating candidate cumulative values of time sequence evolution edge weights on each candidate evolution path, and selecting a candidate evolution path with the maximum candidate cumulative value as the optimal evolution path;

[0041] traversing the development stage nodes on the optimal evolution path, extracting the nurturing measure nodes connected by the semantic association edges of each development stage node, and calculating the measure adaptation degree based on the measure attribute parameters of the nurturing measure nodes and the stage attribute parameters of the development stage nodes;

[0042] screening the nurturing measure nodes according to the measure adaptation degree, organizing the nurturing measure nodes according to the time sequence order of the development stage nodes on the optimal evolution path, and generating a forest regulation scheme containing a sequence of nurturing measure nodes and an implementation time sequence.

[0043] In a second aspect of the embodiment of the present application, a whole-process forest nurturing inspection and evaluation simulation system is provided, which comprises:

[0044] A first unit is configured to obtain ground measurement data of a target forest stand, construct a forest development stage knowledge graph comprising development stage entity nodes and quality index entity nodes, establish semantic association edges between the entity nodes, calculate quality index values based on the ground measurement data, match a current development stage node in the knowledge graph, and traverse semantic association path sets;

[0045] A second unit is configured to count the number of semantic association edges connected by the quality index entity nodes in the semantic association path sets to obtain a graph structure connection degree group, extract a time sequence position parameter of the current development stage node, generate a dynamic weight coefficient group based on the graph structure connection degree group and the time sequence position parameter, and weight and sum the quality index values to obtain a quality evaluation value;

[0046] A third unit is configured to calculate operation effect values before and after nurturing operations.

[0047] A fourth unit is configured to add the quality evaluation value and the operation effect value to the current development stage node as a time identifier attribute, create a time sequence evolution edge to an adjacent development stage node, and determine the weight of the time sequence evolution edge by the operation effect value.

[0048] A fifth unit is configured to extract a historical time sequence evolution path in the knowledge graph to predict a target development stage node, query the nurturing measure nodes on the optimal evolution path from the current development stage node to the target development stage node, and generate a forest regulation scheme.

[0049] In a third aspect of the embodiment of the present application,

[0050] An electronic device is provided, which comprises:

[0051] processor;

[0052] Memory used to store processor-executable instructions;

[0053] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0054] Fourth aspect of the present invention,

[0055] A computer-readable storage medium is provided, having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0056] This invention constructs a knowledge graph of forest development stages and establishes semantic associations between development stage entity nodes and quality indicator entity nodes, thereby realizing a systematic expression and evaluation of the forest tending process and solving the problem that static indicators in traditional evaluation methods are difficult to reflect dynamic evolution processes.

[0057] This invention introduces graph structure connectivity and temporal position parameters to generate dynamic weighting coefficients, making quality evaluation more comprehensive and objective. At the same time, by comparing the effect values ​​before and after tending operations, it realizes quantitative evaluation of the tending process and improves the scientificity and accuracy of forest management decisions.

[0058] This invention adds quality evaluation values ​​and operational effectiveness values ​​as time-identifying attributes to a knowledge graph, creates temporal evolution edges, predicts the target development stage through historical evolution paths, and queries the optimal evolution path, providing a precise basis for stand management schemes and realizing intelligent decision support and visualization of the forest tending process. Attached Figure Description

[0059] Figure 1 This is a flowchart illustrating the simulation method for the whole-process forest tending inspection and evaluation in an embodiment of the present invention;

[0060] Figure 2 This is a flowchart for calculating the quality evaluation value based on the graph structure connectivity group and temporal location parameters. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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.

[0062] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0063] Figure 1 This is a flowchart illustrating the simulation method for the whole-process forest tending inspection and evaluation in an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes:

[0064] Obtain ground measurement data of the target forest stand, construct a forest development stage knowledge graph including development stage entity nodes and quality indicator entity nodes, and establish semantic association edges between each entity node; calculate quality indicator values ​​based on ground measurement data, match the current development stage node in the knowledge graph, and traverse the semantic association edges to obtain a set of semantic association paths.

[0065] The number of semantically related edges connecting the entity nodes of the quality index in the set of semantically related paths is used to obtain the graph structure connectivity group. The temporal position parameters of the nodes in the current development stage are extracted. A dynamic weight coefficient group is generated based on the graph structure connectivity group and the temporal position parameters. The quality index values ​​are weighted and summed to obtain the quality evaluation value.

[0066] Calculate the operational effectiveness values ​​before and after the tending operation;

[0067] The quality evaluation value and the work effect value are added as time identifier attributes to the current development stage node, and a temporal evolution edge is created to the adjacent development stage node. The weight of the temporal evolution edge is determined by the work effect value.

[0068] Extract historical time-series evolution paths from the knowledge graph to predict target development stage nodes, and query tending measure nodes on the optimal evolution path from the current development stage node to the target development stage node to generate forest stand regulation schemes.

[0069] In one optional implementation, the steps of acquiring ground measurement data of the target forest stand, constructing a knowledge graph of forest development stages, establishing semantic association edges between entity nodes, calculating quality index values ​​based on ground measurement data, matching the current development stage node in the knowledge graph, and traversing semantic association edges to obtain a set of semantic association paths include:

[0070] The ground measurement data includes diameter at breast height (DBH) data, tree height data, and branch height data;

[0071] The set of entity nodes for developmental stages is arranged in the order of forest growth cycle. The set of entity nodes for quality indicators includes entity nodes for diameter level indicators, entity nodes for first structural indicators, and entity nodes for second structural indicators.

[0072] By extracting knowledge rules from forestry experts, semantic association edges are established between entity nodes in the development stage and entity nodes of quality indicators. The semantic association edges carry applicability weight attributes.

[0073] The diameter-level index value is calculated using the diameter at breast height (DBH) data; the first structural index value is calculated using the ratio of the tree height data to the DBH data; and the second structural index value is calculated using the ratio of the branch height data to the tree height data.

[0074] The diameter index value, the first structural index value, and the second structural index value are matched with the quality index entity nodes respectively. The current development stage node is obtained by backtracking through the semantic association edges of the matched quality index entity nodes.

[0075] Starting from the node at the current development stage, traverse along the semantic association edges to the quality indicator entity node, and form a set of semantic association paths from the traversal paths.

[0076] For example, it is necessary to obtain ground measurement data of the target forest stand, including diameter at breast height (DBH), tree height, and height below the branch. This data is usually obtained through field surveys, such as selecting representative trees in a standard plot for measurement, and recording the DBH (cm), tree height (m), and height below the branch (m) of each tree.

[0077] The knowledge graph contains two types of entity nodes: a set of entity nodes representing developmental stages and a set of entity nodes representing quality indicators. The set of entity nodes representing developmental stages is arranged sequentially according to the forest growth cycle, typically including nodes representing the initial growth stage, middle growth stage, stable growth stage, and mature stable stage. The set of entity nodes representing quality indicators includes entity nodes representing diameter at breast height (DBH), primary structural indicators, and secondary structural indicators. DBH entity nodes can include nodes representing primary, intermediate, advanced, and dominant DBH. Primary structural indicator entity nodes (i.e., the ratio of tree height to DBH) can include nodes representing low, medium, and high structural ratios. Secondary structural indicator entity nodes (i.e., the ratio of branch height to tree height) can include nodes representing low, medium, and high canopy layers.

[0078] These semantic association edges are determined by extracting knowledge rules from forestry experts. Each edge carries an applicability weight attribute, representing the strength of the association. For example, a semantic association edge is established between the initial growth stage and the primary diameter class, with a weight of 0.9; between the mid-growth stage and the intermediate diameter class, with a weight of 0.8; between the stable growth stage and the advanced diameter class, with a weight of 0.85; and between the mature and stable stage and the dominant diameter class, with a weight of 0.75. Similarly, semantic association edges can be established between developmental stages and the first and second structural indicators.

[0079] The diameter at breast height (DBH) index value is directly taken from the average DBH data, such as 25.6 cm. The first structural index value is calculated by the ratio of tree height data to DBH data, such as 18.3 m ÷ 25.6 cm = 0.71. The second structural index value is calculated by the ratio of branch height data to tree height data, such as 6.2 m ÷ 18.3 m = 0.34.

[0080] The calculated quality index values ​​are matched with quality index entity nodes in the knowledge graph. Based on the example above, the diameter class index value of 25.6 cm matches the high-diameter class node; the first structure index value of 0.71 matches the medium structure ratio node; and the second structure index value of 0.34 matches the mid-canopy node. By tracing back along the semantic association edges through these matched quality index entity nodes, the node representing the current developmental stage is identified. In this example, the high-diameter class, medium structure ratio, and mid-canopy nodes ultimately point to a stable growth stage node through semantic association edges, thus determining the current developmental stage of the stand as a stable growth stage.

[0081] After determining the current developmental stage, the system traverses along semantically related edges from that node to the quality indicator entity node, forming a set of semantically related paths. For the aforementioned stable growth stage nodes, the semantically related paths include: stable growth stage → high-level diameter (weight 0.85); stable growth stage → medium structure ratio (weight 0.75); stable growth stage → medium canopy (weight 0.8). These sets of semantically related paths can be used for subsequent analysis and decision support.

[0082] To improve identification accuracy, the weights of semantically related edges can be comprehensively considered. For example, when the index value of a forest stand is located at the boundary between two index nodes, two semantically related paths can be considered simultaneously, and the final developmental stage can be determined based on their weights. For multiple candidate developmental stages with similar weights, other auxiliary indicators can be introduced for further judgment, such as stand density and tree health status.

[0083] This invention constructs a forest knowledge graph that includes entity nodes of developmental stages and entity nodes of quality indicators. It achieves efficient integration and transformation of key data such as stand diameter at breast height (DBH), tree height, and branch height, and can accurately match nodes of the current developmental stage to form a complete set of semantic association paths.

[0084] In one optional implementation, the step of generating a dynamic weighted coefficient set based on the graph structure connectivity set and temporal location parameters, and then weighting and summing the quality index values ​​to obtain the quality evaluation value includes:

[0085] The single-hop and double-hop connectivity of entity nodes in the semantic association path set are statistically analyzed, and the graph structure connectivity groups are obtained by summing the results after applying attenuation factors.

[0086] The basic temporal position value is obtained by determining the sequence position of the current development stage node in the set of development stage entity nodes. The historical quality evaluation value sequence of the current development stage node is extracted and its rate of change is calculated to obtain the evolution trend coefficient. The evolution trend coefficient is used as a modulation factor to modulate the basic temporal position value to obtain the temporal position parameter.

[0087] Multiply the graph structure connectivity group with the temporal position parameter to obtain the temporal modulation connectivity group. Obtain the cooperative relationship vector through the covariance matrix eigenvector of the temporal modulation connectivity group. Weight and fuse the temporal modulation connectivity group and the cooperative relationship vector and normalize them to obtain the dynamic weight coefficient group.

[0088] The initial quality evaluation value is obtained by weighting the quality index values ​​using the dynamic weight coefficient group, and the quality evaluation value is obtained by adjusting the confidence level of the initial quality evaluation value using the time-series location parameter.

[0089] Combination Figure 2 This document describes the flowchart for calculating the quality evaluation value based on graph structure connectivity groups and temporal position parameters. It obtains the quality indicator entity nodes associated with the current developmental stage node in the knowledge graph. These nodes are connected to the developmental stage nodes through semantic association edges. For each quality indicator entity node, its single-hop connectivity is calculated, which is the number of semantic association edges directly connecting it to the quality indicator node. For example, a path-level indicator node directly associated with three developmental stage nodes has a single-hop connectivity of 3. Simultaneously, double-hop connectivity is calculated, which is the number of edges indirectly connected to the quality indicator node through an intermediate node. For example, a first-structure indicator node indirectly associated with five other developmental stage nodes through an intermediate developmental stage node has a double-hop connectivity of 5. A decay factor of 0.8 is applied to the single-hop connectivity, and a decay factor of 0.5 is applied to the double-hop connectivity. The sum of these two values ​​yields the graph structure connectivity value of the quality indicator node. After traversing all quality index nodes, a graph structure connectivity group is formed. This group contains three values: the connectivity of the radix index, the connectivity of the first structural index, and the connectivity of the second structural index. For example, the graph structure connectivity group obtained in a certain traversal is 2.4, 3.1, and 1.9.

[0090] The basic temporal position value is determined from the set of entity nodes at each developmental stage. This set is arranged in order of forest growth cycle, with the index starting from 1 and increasing sequentially. If the current node is in the juvenile forest stage and is ranked second in the set, the basic temporal position value is 2; if it is in the middle-aged forest stage and is ranked fifth, the basic temporal position value is 5. A historical quality assessment value sequence is extracted from the time identifier attribute of the node at the current developmental stage. This sequence records the quality assessment value of the node at different time points, for example, the sequence is 65.3, 68.7, 72.1, 75.8. The rate of change of this sequence is calculated by calculating the difference between two adjacent quality assessment values ​​and then dividing by the time interval. The average of all rates of change is used as the evolutionary trend coefficient. For example, in the above sequence, the rate of change for the first segment is 3.4, divided by the time interval of 1 year, resulting in 3.4; the second segment is 3.4; the third segment is 3.7; and the average is 3.5, which is used as the evolutionary trend coefficient. The evolutionary trend coefficient is normalized to the range of 0.5 to 1.5 and used as the modulation factor. Specifically, the evolutionary trend coefficient is divided by the historical series mean and then multiplied by an adjustment factor of 1.2. If the evolutionary trend coefficient is 3.5 and the historical mean is 70.5, then the modulation factor is 0.05 multiplied by 1.2, which equals 0.06. Adding the baseline offset of 1.0, we get 1.06. Multiplying this modulation factor by the base time series position value yields the time series position parameter. For example, multiplying the base time series position value of 5 by the modulation factor 1.06 yields the time series position parameter 5.3.

[0091] The temporal modulation connectivity set is obtained by multiplying each element in the graph structure connectivity set by the temporal position parameter. For example, if the graph structure connectivity set is 2.4, 3.1, and 1.9, and the temporal position parameter is 5.3, then the temporal modulation connectivity set is 12.72, 16.43, and 10.07. A covariance matrix of the temporal modulation connectivity set is constructed. This matrix is ​​a 3x3 square matrix, and its elements are calculated from the covariances of each element in the temporal modulation connectivity set. A sliding window method is used to calculate the covariance, with a window length of 5 historical time points. For each time point, the covariance of each pair of elements in the temporal modulation connectivity set is calculated. For example, the covariance between the first and second elements is 3.2, the covariance between the first and third elements is 1.8, the covariance between the second and third elements is 2.5, and the variances of the diagonal elements are 4.1, 5.3, and 3.7, respectively. The covariance matrix is ​​decomposed into eigenvalues, and the eigenvectors corresponding to the largest eigenvalues ​​are selected as the co-relation vectors. For example, the eigenvectors corresponding to the largest eigenvalue of 8.6 are 0.52, 0.63, and 0.41.

[0092] The temporal modulation connectivity set and the cooperative relationship vector are weighted and fused. Specifically, each element of the temporal modulation connectivity set is multiplied by the corresponding cooperative relationship vector element, and then multiplied by the fusion weight of 0.6. Simultaneously, each element of the temporal modulation connectivity set is directly multiplied by the retention weight of 0.4. The two parts are then added together to obtain the fused vector. For example, the temporal modulation connectivity sets 12.72, 16.43, and 10.07 are fused with the cooperative relationship vectors 0.52, 0.63, and 0.41 to obtain 10.15, 13.72, and 8.29. The fused vector is then normalized by dividing each element by the sum of all elements. For example, the sum of three elements is 32.16, and after normalization, 0.32, 0.43, and 0.25 are obtained as the dynamic weight coefficient set.

[0093] The quality index values ​​are weighted using a dynamic weighted coefficient set, which includes diameter index values, first structural index values, and second structural index values. For example, the quality index values ​​obtained in a certain measurement are 75.3, 82.1, and 68.7. Multiplying the dynamic weighted coefficient set (0.32, 0.43, and 0.25) by the corresponding quality index values ​​yields 24.1, 35.3, and 17.2, respectively. Adding these three together gives the initial quality evaluation value of 76.6. The initial quality evaluation value is then adjusted for confidence using a time-series location parameter. The adjustment method involves determining a confidence coefficient based on the time-series location parameter, which decreases as the time-series location parameter increases. Specifically, the calculation method is to subtract the difference between the time-series location parameter and the baseline location parameter (3) from the preset baseline confidence level (1.0) and then divide by the adjustment range (10). For example, the confidence coefficient corresponding to the time-series location parameter 5.3 is 1.0 minus 2.3 divided by 10, which equals 0.77. The initial quality evaluation value is multiplied by the confidence coefficient to obtain the quality evaluation value. For example, 76.6 multiplied by 0.77 yields a quality evaluation value of 59.0.

[0094] When the time sequence position parameter is abnormally large, for example, exceeding 1.5 times the total number of developmental stages (10, i.e., 15), it is limited to within 15, and the corresponding lower limit of the confidence coefficient is set to 0.2 to ensure that the quality evaluation value does not excessively decay due to confidence adjustment. When an element in the graph structure connectivity group is 0, for example, a quality indicator node is not associated with any developmental stage node, the element is assigned the minimum connectivity of 0.1 to avoid numerical invalidation in subsequent calculations. When calculating the covariance matrix, if there are fewer than 5 historical time points, the existing time point data is used for calculation, and a correction factor is applied to the covariance value. The correction factor is the actual number of time points divided by the standard number of time points (5). For example, if there are only 3 time points, the correction factor is 0.6, and all covariance values ​​are multiplied by this correction factor. When normalizing the dynamic weight coefficient group, if the sum of the elements is less than 0.1, each element is assigned an equal value of 0.33 to ensure the effectiveness of the weight group. If a quality indicator value is missing when inputting it, the historical average of the indicator will be used to fill the gap. For example, if the value of the second structural indicator is missing, the historical average of 65.8 will be used instead.

[0095] This invention employs a method combining multi-hop connectivity analysis and temporal position dynamic modulation to construct an adaptive weight coefficient generation mechanism. This mechanism can simultaneously consider the structural importance of quality indicators in the knowledge graph and the temporal characteristics of the current developmental stage, thereby achieving an accurate assessment of the contribution of quality indicators. This significantly improves the accuracy and reliability of quality evaluation results, making the evaluation results more consistent with the actual forest stand growth patterns.

[0096] In one optional implementation, the step of modulating the basic temporal position value using the evolution trend coefficient as a modulation factor to obtain the temporal position parameter includes:

[0097] The short-term trend coefficient is obtained by calculating the first-order difference mean of the historical quality evaluation value sequence in the short-term window, the long-term trend coefficient is obtained by calculating the linear fitting slope in the long-term window, and the evolution acceleration coefficient is obtained by calculating the second-order difference of the whole sequence.

[0098] The deviations of the short-term trend coefficient, long-term trend coefficient, and evolution acceleration coefficient from the historical mean are calculated. Abnormal trend components with deviations exceeding a preset deviation threshold are smoothed and then fused with normal trend components to obtain a robust evolution feature vector.

[0099] The modulation sensitivity factor is obtained based on the stage attribute parameters of the current development stage node. A nonlinear modulation function is selected from a preset modulation function library according to the modulation sensitivity factor. The modulation intensity parameter is obtained through the modulus of the robust evolution feature vector.

[0100] The components of the robust evolutionary feature vector are used as modulation factors, and the basic time-series position value is modulated in multiple levels according to the selected nonlinear modulation function. The time-series position parameter is obtained by weighted fusion of the modulation results of each level through the modulation intensity parameter.

[0101] For example, in the calculation of the evolutionary trend coefficient, multi-scale analysis of the historical quality evaluation value sequence requires processing short-term, long-term, and global change characteristics separately. The short-term window is typically set to the most recent 3-5 evaluation periods. A first-order difference sequence is formed by calculating the differences between adjacent evaluation values, and then the arithmetic mean of this sequence is calculated to obtain the short-term trend coefficient. For instance, for the historical quality evaluation value sequence [0.72, 0.75, 0.79, 0.81, 0.78], taking the most recent 4 evaluation periods, the first-order difference sequence [0.03, 0.04, 0.02, -0.03] is calculated, and its arithmetic mean of 0.015 is the short-term trend coefficient. The long-term window typically covers 8-12 evaluation periods. A linear fit is performed on the evaluation value sequence within this window, and the resulting slope value is the long-term trend coefficient. For longer historical sequences, such as [0.65, 0.68, 0.71, 0.72, 0.75, 0.79, 0.81, 0.78, 0.82, 0.84], the slope value obtained by least squares fitting is approximately 0.019, which is the long-term trend coefficient. For full sequence analysis, the second-order difference of the historical quality evaluation values ​​needs to be calculated, i.e., the difference between adjacent first-order differences. After forming a second-order difference sequence, its mean is taken to obtain the evolutionary acceleration coefficient. For the above first-order difference sequence [0.03, 0.04, 0.02, -0.03], the second-order difference is calculated to obtain [0.01, -0.02, -0.05], and its mean of -0.02 is the evolutionary acceleration coefficient, indicating a slowing trend in the rate of change.

[0102] The deviation of the trend coefficient from the historical mean is calculated by obtaining historical datasets for the short-term trend coefficient, long-term trend coefficient, and evolutionary acceleration coefficient, and calculating their respective means and standard deviations. Taking the short-term trend coefficient as an example, if its historical mean is 0.012 and its standard deviation is 0.008, the deviation of the current value of 0.015 from the mean is |(0.015-0.012) / 0.008|=0.375. A preset deviation threshold is typically set between 2.0 and 3.0. If the deviation exceeds this threshold, the trend component is considered abnormal. Abnormal trend components are smoothed using a local weighted average, such as by taking 30% of the current value and 70% of the historical mean and performing a weighted average to obtain the smoothed trend component. The processed abnormal trend component is combined with the normal trend component to form a three-dimensional robust evolutionary feature vector [short_trend, long_trend, accel_coef], such as [0.015, 0.019, -0.02].

[0103] The modulation sensitivity factor is determined based on the characteristic attributes of nodes at the current developmental stage. Each developmental stage node has preset sensitivity parameters for different trend scales; for example, the early growth stage is more sensitive to short-term trends, while the stable growth stage is more sensitive to long-term trends. In practice, stage feature values, such as growth phase position values ​​and stability indices, can be extracted from the attribute parameters of the current developmental stage node and converted into modulation sensitivity factors through a mapping function. For example, for the stable growth stage, its stage attribute parameters include a stability index of 0.75 and a growth phase position value of 0.6. By setting weights of 0.6 and 0.4, a modulation sensitivity factor of 0.69 is obtained through weighted fusion. The selection of functions in the preset modulation function library follows the interval mapping rule: exponential functions are selected in the range of [0-0.3] to emphasize rapid changes; logarithmic functions are selected in the range of [0.3-0.6] to smooth gradual changes; modified sigmoid functions are selected in the range of [0.6-0.8] to balance smooth transitions and critical sensitivity; and piecewise linear functions are selected in the range of [0.8-1.0] to precisely control boundary behavior. Therefore, a modulation sensitivity factor of 0.69 corresponds to selecting a modified sigmoid function as the nonlinear modulation function. The modulation intensity parameter is determined by calculating the magnitude of the robust evolution eigenvector, which is 0.015. 2 +0.019 2 +(-0.02) 2 Then take the square root ≈ 0.031 to represent the overall trend strength.

[0104] Obtaining the baseline temporal position value is typically the relative position of the node at the current developmental stage within the developmental sequence, such as 0.65 (representing the 65th percentile of the total developmental cycle). Multi-level modulation refers to applying three modulation factors—short-term trend, long-term trend, and acceleration—to the baseline temporal position value. In implementation, each modulation level uses the same nonlinear modulation function type but with different parameter configurations. For example, for a selected modified sigmoid function, short-term trend modulation uses parameter a=2.5, long-term trend modulation uses parameter a=3.0, and acceleration modulation uses parameter a=2.0. The short-term trend modulation calculation formula is: baseline position value + S(short-term trend coefficient, short-term modulation parameter) × modulation amplitude factor, where the modulation amplitude factor is a modulation intensity control parameter, typically set between 0.05 and 0.15. Long-term trend and acceleration modulation are performed in a similar manner, yielding modulation results at each of the three levels. Finally, the three-layer modulation results are weighted and fused using the modulation intensity parameter. For example, an intensity parameter of 0.031 indicates that the overall modulation influence is moderate. The three-layer modulation results can be weighted and averaged using weights of [0.25, 0.45, 0.3] to obtain the final time position parameter, such as 0.682. This parameter will be used for subsequent quality evaluation calculations.

[0105] The computational accuracy and boundary handling of the modulation process also require special attention. The effective range of temporal location parameters is usually limited to the interval [0.1, 0.9]. Values ​​below 0.1 are considered early developmental stages, and values ​​above 0.9 are considered late developmental stages. Calculation results outside this range need to be truncated. Modulation calculations use floating-point precision, retaining three decimal places to avoid the accumulation of small floating-point errors. Modulation function parameters can be adaptively adjusted for different forest types; for example, different parameter configuration files can be used for coniferous forests, broadleaf forests, and mixed forests. A parameter self-calibration mechanism should also be included, adjusting the modulation function parameters and weighting coefficients by periodically comparing predicted results with actual observations to improve model accuracy. For data gaps, such as incomplete historical quality assessment sequences, a sliding window filling strategy is used, employing the mean or linear interpolation of available data to fill in the missing points, ensuring the continuity and reliability of evolutionary trend calculations.

[0106] This invention analyzes the short-term and long-term trends and evolutionary accelerations of historical quality assessment sequences, and combines anomaly trend detection and smoothing to achieve a multi-scale robust characterization of developmental trends. Furthermore, it employs a nonlinear modulation function to perform multi-level fine modulation of the basic temporal positions, significantly enhancing the sensitivity of temporal position parameters to the dynamic characteristics of stand development and improving the accuracy and stability of subsequent quality assessments.

[0107] In one optional implementation, the step of adding the quality evaluation value and the job performance value as time-stamped attributes to the current development stage node and creating temporal evolution edges to adjacent development stage nodes includes:

[0108] Ground measurement data before and after the mulching operation were obtained to calculate the quality evaluation value before the operation and the quality evaluation value after the operation. The difference between the quality evaluation value before the operation and the quality evaluation value after the operation was calculated as the operation effect value.

[0109] The quality evolution trend coefficient is calculated based on the quality evaluation value of the current development stage node and the historical quality evaluation value sequence. The quality evaluation value, operation effect value, and quality evolution trend coefficient are encapsulated as a time identifier attribute and added to the current development stage node.

[0110] From the set of entity nodes in the developmental stage, identify candidate developmental stage nodes that are adjacent to the current developmental stage node in terms of developmental time sequence. Calculate the developmental stage span parameter between each candidate developmental stage node and the current developmental stage node. Based on the developmental stage span parameter, select adjacent developmental stage nodes.

[0111] The basic edge weight value is calculated based on the operation effect value and the quality evolution trend coefficient. The historical time identifier attribute of adjacent development stage nodes is extracted to calculate the historical quality fluctuation coefficient. The basic edge weight value is then corrected using the historical quality fluctuation coefficient to obtain the adjusted edge weight value.

[0112] Create temporal evolution edges between the current development stage node and each adjacent development stage node, and assign the adjusted edge weight value to the temporal evolution edge as the edge weight.

[0113] For example, before the tending operation, the diameter distribution data, first structural parameter, and second structural parameter of the forest stand are obtained through ground measurement. Diameter distribution data is obtained by measuring each tree individually, measuring all trees with a diameter at breast height (DBH) greater than 5 cm, recording the DBH value of each tree, and statistically analyzing the number of trees in each diameter class interval. The first structural parameter is obtained by setting up standard quadrats with an area of ​​20 meters by 20 meters. Within the quadrats, the spatial distribution coordinates, crown diameter, and branch height of the trees are measured, and the nearest neighbor distance and crown overlap between individual trees are calculated. The second structural parameter is obtained through vertical stratification surveys, dividing the forest stand vertically into four layers: the surface layer, shrub layer, sub-tree layer, and tree layer. The surface layer height ranges from 0 to 0.5 meters, the shrub layer from 0.5 to 2 meters, the sub-tree layer from 2 to 8 meters, and the tree layer from 8 meters and above. The vegetation cover and biomass percentage of each layer are statistically analyzed. The measured data is input into a quality assessment model to calculate the pre-operation quality assessment value. This model uses a dynamic weighted coefficient set to weight the diameter index value, the first structural index value, and the second structural index value. For example, if a pre-operation measurement yields a diameter index value of 72.3, a first structural index value of 68.5, and a second structural index value of 65.1, the corresponding dynamic weighted coefficient sets are 0.35, 0.42, and 0.23. After weighted combination, the initial quality assessment value is 68.9, and after confidence adjustment, the pre-operation quality assessment value is 66.7.

[0114] During the tending operation, thinning, pruning, and forest clearing were carried out according to the design plan. Thinning was performed based on diameter distribution and spatial distribution, prioritizing the removal of weak, poorly shaped, and overly dense individuals. The thinning intensity was controlled between 15% and 25% of the stand's total number of trees. Pruning was carried out on the remaining trees, removing dead, diseased, and crossing branches below one-third of their height, using smooth cuts to avoid tearing the bark. Forest clearing involved removing fallen trees, litter, and weeds. The cleared biomass was either shredded on-site or removed from the forest. Ground measurements were taken 30 days after the operation, using the same methods as before to ensure data comparability. The obtained diameter distribution data, first structural parameter, and second structural parameter were input into the quality assessment model to calculate the post-operation quality assessment value. For example, after the operation, the measured values ​​were 78.6 for the diameter index, 75.2 for the first structural index, and 71.8 for the second structural index. The corresponding dynamic weighting coefficients were 0.33, 0.44, and 0.23. After weighted combination, the initial quality evaluation value was 75.4. After confidence adjustment, the post-operation quality evaluation value was 73.1.

[0115] The difference between the pre-job quality evaluation value and the post-job quality evaluation value is used to obtain the job effect value. For example, subtracting the pre-job quality evaluation value of 66.7 from the post-job quality evaluation value of 73.1 yields a job effect value of 6.4. A negative job effect value indicates that the job has led to a decrease in quality, and this anomaly needs to be recorded and the job plan review process triggered. The valid range for the job effect value is set between -10 and +20. Values ​​outside this range are considered measurement errors or calculation anomalies and require remeasurement and recalculation.

[0116] Extract the historical quality assessment value sequence from the time stamp attribute of the current development stage node. This sequence is arranged chronologically and contains quality assessment values ​​from multiple past moments. Add the current quality assessment value (73.1) to the end of this sequence to form an updated historical quality assessment value sequence. For example, the updated sequence is 62.8, 66.2, 70.5, 75.3, 66.7, 73.1. Calculate the difference between the current quality assessment value (73.1) and all values ​​in the historical quality assessment value sequence except the last value, resulting in a difference sequence of 10.3, 6.9, 2.6, -2.2, 6.4. Statistically analyze the positive values ​​in the difference sequence, calculating their frequency and average magnitude. A positive value occurring 4 times has an average magnitude of 6.55. Statistically analyze the negative values ​​in the difference sequence, calculating their frequency and average magnitude. A negative value occurring once has an average magnitude of 2.2. Dividing the average amplitude of positive values ​​by the frequency of positive values ​​yields a positive evolution intensity of 1.64, and dividing the average amplitude of negative values ​​by the frequency of negative values ​​yields a negative evolution intensity of 2.2. Dividing the positive evolution intensity by the sum of the positive and negative evolution intensities yields the quality evolution trend coefficient; for example, dividing 1.64 by 3.84 results in a quality evolution trend coefficient of 0.43. When the length of the historical quality evaluation value sequence is less than 3, the quality evolution trend coefficient is set to the default value of 0.5, indicating that the quality evolution is in an uncertain state. A quality evolution trend coefficient greater than 0.5 indicates an overall upward trend in quality, while a coefficient less than 0.5 indicates an overall downward trend in quality.

[0117] The quality evaluation value, job performance value, and quality evolution trend coefficient are encapsulated into a time-stamp attribute structure. This structure includes a timestamp field, a quality evaluation value field, a job performance value field, and a quality evolution trend coefficient field. The quality evaluation value field stores a floating-point number ranging from 0 to 100, with precision to one decimal place. The job performance value field stores a floating-point number ranging from -10 to +20, with precision to one decimal place. The quality evolution trend coefficient field stores a floating-point number ranging from 0 to 1, with precision to two decimal places. This structure is added to the time-stamp attribute list of the current development stage node. The list is managed using a first-in, first-out (FIFO) approach; when the list length exceeds 50 records, the oldest record is deleted to maintain a stable list length. For example, a time-stamp attribute obtained in a certain encapsulation might include a timestamp of November 5, 2025, 14:30:00, a quality evaluation value of 73.1, a job performance value of 6.4, and a quality evolution trend coefficient of 0.43.

[0118] The system queries the set of entity nodes at different developmental stages to find candidate nodes that have a developmental time sequence association with the node at the current developmental stage. This association is determined by the node's developmental sequence number, which starts at 1 and increments to represent the evolutionary order from juvenile to mature forest. Subtracting 1 from the developmental sequence number of the current node yields the predecessor candidate node number, and adding 1 yields the successor candidate node number. For example, if the current node is at the middle-aged forest stage with a developmental sequence number of 5, the predecessor candidate node with a sequence number of 4 corresponds to the late juvenile forest stage, and the successor candidate node with a sequence number of 6 corresponds to the late middle-aged forest stage. The system then retrieves the corresponding node from the set of entity nodes at different developmental stages based on the candidate node number, using a hash index to improve query efficiency.

[0119] The developmental stage span parameter is represented by the absolute value of the difference between the candidate node number and the current node number. For example, when the current node number is 5 and the candidate node number is 4, the span parameter is 1; when the candidate node number is 6, the span parameter is 1; and when the candidate node number is 3, the span parameter is 2. A span parameter threshold of 1 is set, and only candidate nodes with a span parameter less than or equal to the threshold are retained as adjacent developmental stage nodes. This selection rule ensures that temporal evolution edges only connect directly adjacent stages in developmental sequence, avoiding discontinuous evolutionary relationships spanning multiple stages. For example, two adjacent developmental stage nodes are obtained from candidate nodes with numbers 4 and 6.

[0120] Divide the task performance value by 10 to obtain the normalized task performance component, and multiply the quality evolution trend coefficient by 5 to obtain the normalized trend component. Add the two components together and divide by 2 to obtain the basic edge weight value. For example, a task performance value of 6.4 corresponds to a normalized task performance component of 0.64, a quality evolution trend coefficient of 0.43 corresponds to a normalized trend component of 2.15, and the basic edge weight value is 1.40. The basic edge weight value ranges from 0 to 10. When the calculated result exceeds this range, it is truncated: less than 0 is assigned a value of 0.1, and greater than 10 is assigned a value of 10.

[0121] Historical quality evaluation value sequences are extracted from the time-identification attribute lists of adjacent developmental stage nodes, and the standard deviation of these sequences is calculated as a measure of historical quality fluctuation. For example, if the historical quality evaluation value sequence of an adjacent node is 58.3, 61.7, 59.2, 63.5, and 60.8, the standard deviation is calculated to be 2.1. The historical quality fluctuation measure is divided by the average of the historical quality evaluation value sequences to obtain the historical quality fluctuation coefficient. For example, if the average is 60.7, the historical quality fluctuation coefficient is 0.035. When the historical quality fluctuation coefficient is less than 0.05, it indicates stable quality evolution, and the basic edge weight value is multiplied by an enhancement factor of 1.2 to obtain the adjustment edge weight value. When the historical quality fluctuation coefficient is greater than or equal to 0.05 and less than 0.1, it indicates moderate quality evolution fluctuation, and the basic edge weight value remains unchanged as the adjustment edge weight value. When the historical quality fluctuation coefficient is greater than or equal to 0.1, it indicates severe quality evolution fluctuation, and the basic edge weight value is multiplied by a decay factor of 0.8 to obtain the adjustment edge weight value. For example, a historical quality volatility coefficient of 0.035 corresponds to an enhancement factor of 1.2, and the basic edge weight value of 1.40 multiplied by 1.2 yields an adjustment edge weight value of 1.68.

[0122] In the knowledge graph's edge set, directed edges are created from the current developmental stage node to each adjacent developmental stage node. These edges are identified as temporal evolution edges. The edge weight is adjusted and assigned to the edge's weight attribute, stored as a floating-point number with two decimal places. For example, the weight of a temporal evolution edge from node 5 to node 4 is 1.68, and the weight of a temporal evolution edge from node 5 to node 6 is independently calculated based on the historical quality fluctuation coefficients of the corresponding adjacent nodes. After the temporal evolution edge is created, a graph index update operation is triggered, updating the out-degree count of the starting node and the in-degree count of the ending node, and updating the global edge weight distribution statistics of the graph. The edge creation operation is executed using a transactional approach to ensure the atomicity of node attribute updates and edge creation. If the transaction fails, all changes are rolled back to restore graph consistency.

[0123] This invention can construct temporal evolutionary relationships between nodes at different developmental stages based on the effects of tending operations and the evolutionary trends of stand quality, providing a scientific basis for forestry management. This method is applicable to different stand types and developmental stages, and parameters can be adjusted according to specific application scenarios to improve adaptability and accuracy.

[0124] In one optional implementation, the step of obtaining the adjusted edge weight value by correcting the base edge weight value using the historical quality fluctuation coefficient includes:

[0125] The current quality contribution factor is obtained by weighted fusion of the operation effect value and the quality evolution trend coefficient. The basic edge weight value is calculated based on the ratio of the current quality contribution factor to the preset benchmark factor.

[0126] Historical quality evaluation value sequences are extracted from the historical time identifier attributes of adjacent development stage nodes. The standard deviation of the historical quality evaluation value sequences is calculated to obtain the historical quality fluctuation amplitude. The historical quality fluctuation coefficient is calculated based on the ratio of the historical quality fluctuation amplitude to the mean of the historical evaluation values.

[0127] A side weight correction function is constructed, and the historical quality fluctuation coefficient is used as a correction factor to input the side weight correction function to correct the basic side weight value to obtain the adjusted side weight value. The adjusted side weight value is inversely related to the historical quality fluctuation coefficient.

[0128] For example, the current calculated operation effect value and quality evolution trend coefficient are obtained. For instance, a measurement yields an operation effect value of 6.4 and a quality evolution trend coefficient of 0.43. The fusion weight of the operation effect value is set to 0.6, and the fusion weight of the quality evolution trend coefficient is set to 0.4. Multiplying the operation effect value by 0.6 yields 3.84, and multiplying the quality evolution trend coefficient by 10 and then by 0.4 yields 1.72. Adding these two values ​​together gives the current quality contribution factor of 5.56. The current quality contribution factor reflects the overall improvement of stand quality caused by tending operations. A higher value indicates a more significant operation effect and a better quality evolution trend. The valid value range for the current quality contribution factor is 0 to 15. Values ​​less than 0 are truncated to 0.1, and values ​​greater than 15 are truncated to 15.

[0129] The preset baseline factor is set to 5.0, representing the standard quality contribution level of tending operations under ideal conditions, determined based on statistical analysis of a large amount of historical operation data. The quality contribution ratio is obtained by dividing the current quality contribution factor by the preset baseline factor; for example, 5.56 divided by 5.0 yields 1.11. A quality contribution ratio greater than 1 indicates that the current operation is better than the standard level, while a ratio less than 1 indicates that it is lower than the standard level. The basic edge weight value is obtained by multiplying the quality contribution ratio by the baseline edge weight coefficient of 2.0; for example, 1.11 multiplied by 2.0 yields a basic edge weight value of 2.22. The default value of the baseline edge weight coefficient is 2.0, which can be adjusted within the range of 1.5 to 3.0 according to different forest stand types: 2.2 for coniferous forests, 1.8 for broadleaf forests, and 2.0 for mixed forests. The effective range of the basic edge weight value is 0.1 to 10; a value of 0.1 is assigned when the calculated result is less than 0.1, and a value of 10 is assigned when the result is greater than 10.

[0130] The historical quality assessment value sequence is extracted from the time-stamp attribute list of adjacent developmental stage nodes. This list stores the quality assessment values ​​of the node at multiple past moments in chronological order. For example, the historical quality assessment value sequence of an adjacent node is 58.3, 61.7, 59.2, 63.5, 60.8, 62.1, and 64.3. The arithmetic mean of all elements in this sequence is calculated as the historical quality assessment value mean. For example, the sum of the seven values ​​is 429.9, and dividing by 7 gives the mean of 61.41. The squared difference between each quality assessment value and the mean is calculated. For example, the square of 58.3 minus 61.41 is 9.67, and the square of 61.7 minus 61.41 is 0.08. The resulting sequence of squared differences is 9.67, 0.08, 4.88, 4.36, 0.37, 0.48, and 8.36. Summing all the squared differences yields a total of 27.8. Dividing this by the sequence length of 7 gives a variance of 3.97. Taking the square root of this variance gives a standard deviation of 1.99, which is used as the historical quality fluctuation range. The historical quality fluctuation range reflects the dispersion of the quality evaluation value of this developmental stage node over a historical period; a larger value indicates more severe quality fluctuations.

[0131] The historical quality fluctuation coefficient is obtained by dividing the historical quality fluctuation range by the average historical quality evaluation value. For example, 1.99 divided by 61.41 yields 0.032. The historical quality fluctuation coefficient is a dimensionless relative fluctuation index, ranging from 0 to 1. A value close to 0 indicates very stable quality evolution, while a value close to 1 indicates extremely unstable quality evolution. When the length of the historical quality evaluation value sequence is less than 5, the calculated historical quality fluctuation coefficient is adjusted by multiplying it by a correction factor. The correction factor is equal to the actual sequence length divided by the standard sequence length of 5. For example, when the sequence length is 3, the correction factor is 0.6, and the historical quality fluctuation coefficient needs to be multiplied by 0.6. When the average historical quality evaluation value is less than 10, the data is considered abnormal, and the historical quality fluctuation coefficient is set to the default value of 0.1.

[0132] The edge weight correction function employs a piecewise mapping strategy, dividing the historical quality fluctuation coefficient range into three segments for separate processing. When the historical quality fluctuation coefficient is less than 0.05, the quality evolution is considered highly stable, and the correction coefficient is 1.3. When the historical quality fluctuation coefficient is greater than or equal to 0.05 and less than 0.12, the quality evolution fluctuation is considered moderate, and the correction coefficient is calculated using a linear mapping method: 1.3 - [(historical quality fluctuation coefficient - 0.05) / 0.07] × 0.5. For example, the correction coefficient for a historical quality fluctuation coefficient of 0.08 is 1.3 - [(0.08 - 0.05) / 0.07] × 0.5 = 1.09. When the historical quality fluctuation coefficient is greater than or equal to 0.12, the quality evolution fluctuation is considered severe, and the correction coefficient is 0.7. The adjusted edge weight value is obtained by multiplying the basic edge weight value by the corresponding correction coefficient. The correction function ensures that the adjusted edge weight value has an inverse relationship with the historical quality fluctuation coefficient; that is, the greater the historical fluctuation, the smaller the edge weight, reflecting the negative impact of quality instability on the reliability of the evolution path.

[0133] The historical quality fluctuation coefficient of 0.032 is used as a correction factor input to the edge weight correction function. Since 0.032 is less than 0.05, it falls within the highly stable range, so an enhancement coefficient of 1.3 is selected. The base edge weight value of 2.22 is multiplied by the enhancement coefficient of 1.3 to obtain the adjusted edge weight value of 2.89. The precision of the adjusted edge weight value is retained to two decimal places, with a value range of 0.1 to 13. Values ​​outside this range are truncated. The segmentation threshold of the edge weight correction function can be adjusted according to the stand characteristics. For plantations with stable growth environments, the first threshold can be adjusted from 0.05 to 0.08; for naturally succeeding stands, it can be adjusted to 0.03. The adjustable range of the enhancement coefficient is 1.2 to 1.5, and the adjustable range of the attenuation coefficient is 0.6 to 0.8. Under the default configuration, the enhancement coefficient is 1.3 and the attenuation coefficient is 0.7.

[0134] For example, a temporal evolution edge is created from a node in a middle-aged forest stage to nodes in adjacent developmental stages. The node's performance value is 6.4, its quality evolution trend coefficient is 0.43, and its current quality contribution factor is 6.4 × 0.6 + 0.43 × 10 × 0.4 = 5.56. The basic edge weight is (5.56 / 5.0) × 2.0 = 2.22. Querying the node with the adjacent developmental stage number 6, its historical quality evaluation value sequence is extracted as 58.3, 61.7, 59.2, 63.5, 60.8, 62.1, and 64.3. The historical quality evaluation value mean is (58.3 + 61.7 + 59.2 + 63.5 + 60.8 + 62.1 + 64.3) / 7 = 61.41, the standard deviation is 1.99, and the historical quality fluctuation coefficient is 1.99 / 61.41 = 0.032. Since 0.032 is less than 0.05, a correction factor of 1.3 is used, and the adjusted edge weight value is 2.22 × 1.3 = 2.89. This adjusted edge weight value is assigned to the time-series evolution edge pointing from the current node to node 6. For another adjacent node with index 4, its historical quality evaluation value sequence is extracted as 52.6, 54.1, 60.3, 56.8, 53.2, with a mean of (52.6 + 54.1 + 60.3 + 56.8 + 53.2) / 5 = 55.4, a standard deviation of 3.12, and a historical quality fluctuation coefficient of 3.12 / 55.4 = 0.056. Since 0.056 is in the range of 0.05 to 0.12, the correction factor is 1.3 - [(0.056 - 0.05) / 0.07] × 0.5 = 1.26, and the adjusted edge weight value is 2.22 × 1.26 = 2.80.

[0135] When the historical time identifier attribute list of adjacent development stage nodes is empty, it indicates that the node has not yet accumulated historical data. The historical quality fluctuation coefficient is set to the default value of 0.08, and the correction coefficient is calculated using a linear mapping interval. When the basic edge weight value is abnormally large, exceeding 8, a logarithmic transformation is performed on the basic edge weight value before correction. The logarithmically transformed value is multiplied by the correction coefficient and then exponentially restored to obtain the adjusted edge weight value, avoiding extreme weight values ​​from affecting the graph evolution path analysis. The correction function calculation process uses floating-point arithmetic, with intermediate results maintained to four decimal places and the final output precision retained to two decimal places. The edge weight correction operation is executed in the edge creation transaction, maintaining atomicity with node attribute updates and edge insertion operations. If any step fails, all changes are rolled back to ensure graph consistency. The corrected edge weight value is stored in the weight attribute field of the edge object. The field type is double-precision floating-point number, supporting efficient numerical comparison and sorting operations for optimal path search.

[0136] This invention constructs a dual dynamic adjustment mechanism for edge weights, which organically combines the current quality contribution with the historical quality stability. By finely adjusting the basic edge weights through a correction function, the temporal evolution edge weights can comprehensively reflect the immediate effect and long-term stability of nurturing operations, thereby improving the accuracy of temporal relationship expression in the knowledge graph and providing a more reliable weight basis for querying the optimal evolution path.

[0137] In one optional implementation, the steps of extracting historical time-series evolutionary paths from the knowledge graph to predict target developmental stage nodes, and querying tending measure nodes on the optimal evolutionary path from the current developmental stage node to the target developmental stage node to generate a forest stand regulation scheme include:

[0138] Extract all historical temporal evolution paths containing temporal evolution edges from the knowledge graph, calculate the historical cumulative value of the temporal evolution edge weights on each historical temporal evolution path, and take the endpoint development stage node of the historical temporal evolution path with the largest historical cumulative value as the target development stage node;

[0139] Starting from the current developmental stage node and ending at the target developmental stage node, candidate evolutionary paths are searched in the knowledge graph. The cumulative candidate value of the temporal evolutionary edge weights on each candidate evolutionary path is calculated, and the candidate evolutionary path with the largest cumulative candidate value is selected as the optimal evolutionary path.

[0140] Traverse the developmental stage nodes on the optimal evolutionary path, extract the nurturing measure nodes connected by semantic association edges for each developmental stage node, and calculate the measure fit based on the measure attribute parameters of the nurturing measure nodes and the stage attribute parameters of the developmental stage nodes;

[0141] Based on the suitability of the measures, the tending measure nodes are selected, and the tending measure nodes are organized according to the temporal sequence of the developmental stage nodes on the optimal evolution path to generate a forest stand regulation scheme that includes the tending measure node sequence and the implementation time sequence.

[0142] For example, all edge objects identified as temporal evolution edges are selected from the edge set of the knowledge graph. A complete path is constructed along these temporal evolution edges, from the starting developmental stage node to the ending developmental stage node. Path construction employs a depth-first traversal strategy, starting from any developmental stage node in the graph and recursively visiting successor nodes along the direction of the temporal evolution edges until the ending node with no outgoing edges is reached or a cycle is formed. Each path is recorded as a combination of a sequence of developmental stage nodes and a corresponding sequence of temporal evolution edges. For example, a historical temporal evolution path might contain the node sequence 2, 3, 5, and 7, and the edge sequence with edge weights of 1.85, 2.34, and 2.12. The above construction process is repeated, traversing all developmental stage nodes as starting points, to obtain a set of all historical temporal evolution paths in the knowledge graph. The path set is stored in an in-memory index structure, with a dual index built according to path length and starting node number, supporting fast retrieval and filtering operations.

[0143] The historical cumulative value of each historical evolutionary path is calculated by summing the weights of all temporally evolved edges. For example, if the edge weight sequence of the path above is 1.85, 2.34, and 2.12, the historical cumulative value is 1.85 + 2.34 + 2.12 = 6.31. The historical cumulative value reflects the overall quality improvement effect of the evolutionary path over a historical period; a larger value indicates a more significant improvement in stand quality along the path. The historical cumulative value of all paths is calculated by traversing the set of historical temporally evolved paths, and the results are stored in the cumulative value attribute field of the path object. The historical cumulative values ​​are sorted in descending order, and the path ranked first is selected as the optimal historical evolutionary path. The endpoint developmental stage node of this optimal historical evolutionary path is extracted and set as the target developmental stage node. For example, a knowledge graph contains 5 historical evolution paths with historical cumulative values ​​of 6.31, 5.87, 7.42, 6.05, and 5.21, respectively. The path with the maximum value of 7.42 ends at the mature forest stage node with sequence number 8, which is the target development stage node.

[0144] Starting with the current developmental stage node as the search starting point and the target developmental stage node as the search endpoint, a path search operation is performed in the knowledge graph. The path search employs a breadth-first strategy, adding all directly reachable successor nodes via temporal evolution edges to a queue for traversal. Nodes in the queue are processed in ascending order of their node numbers. For each node in the queue, its number is checked against the target node's number. If they are equal, a candidate evolutionary path from the starting point to the target node is found and added to the candidate path set. If they are not equal, the process continues to visit the successor nodes of that node until the queue is empty or all paths have been traversed. The maximum length of a candidate evolutionary path is limited to 10 nodes; paths exceeding this length are considered abnormal evolutions and discarded. For example, searching from the middle-aged forest node (number 5) to the mature forest node (number 8) yields three candidate evolutionary paths: Path 1 contains nodes with numbers 5, 6, and 8; Path 2 contains nodes with numbers 5, 7, and 8; and Path 3 contains nodes with numbers 5, 6, 7, and 8.

[0145] The cumulative sum of all temporal evolutionary edge weights on each candidate evolutionary path is calculated as the candidate cumulative value for that path. For example, the edge weight sequence for path one is 2.89, 3.15, and the candidate cumulative value is 6.04. The edge weight sequence for path two is 2.67, 2.88, and the candidate cumulative value is 5.55. The edge weight sequence for path three is 2.89, 2.42, 2.88, and the candidate cumulative value is 8.19. The candidate cumulative values ​​are sorted in descending order, and the candidate evolutionary path that ranks first after sorting is selected as the optimal evolutionary path. For example, among the three paths above, the path with the largest candidate cumulative value of 8.19 corresponds to path three, which is the optimal evolutionary path. The optimal evolutionary path is stored as an ordered sequence of nodes and corresponding edge weight sequences. The node sequence is 5, 6, 7, 8, and the edge weight sequence is 2.89, 2.42, 2.88.

[0146] Traverse the node sequence of the optimal evolutionary path, processing each developmental stage node sequentially starting from the first node. For the currently processed developmental stage node, query all edge objects in the knowledge graph that originate from that node and have an edge type of semantic association, and extract the endpoint nodes of these edges. Check the type attribute of the endpoint nodes and filter out node objects of type "nurturing measure nodes". For example, the middle-aged forest node at index 5 is connected to three nurturing measure nodes—thinning measure node, pruning measure node, and fertilization measure node—through semantic association edges. Add the filtered nurturing measure nodes to the associated measure set of this developmental stage node. Repeat the above operation until all developmental stage nodes in the node sequence have been processed, obtaining the associated measure set corresponding to each developmental stage node.

[0147] Extract stage attribute parameters from the attribute fields of development stage nodes, including stage type identifier, stand density range, average diameter at breast height (DBH) range, and canopy closure range. For example, the stage attribute parameters for the late-stage middle-aged forest node (serial number 6) are: stage type late-stage middle-aged forest, stand density 1200-1500 trees per hectare, average DBH 18-24 cm, and canopy closure 0.7-0.85. Extract measure attribute parameters from the attribute fields of tending measure nodes, including measure type identifier, applicable stand density range, applicable average DBH range, and operation intensity range. For example, the measure attribute parameters for the thinning measure node are: measure type thinning, applicable stand density 1000-2000 trees per hectare, applicable average DBH 15-30 cm, and operation intensity 15%-30%. Calculate the length of the overlap between the stand density range of the development stage node and the applicable stand density range of the tending measure node, and divide it by the length of the union of the two ranges to obtain the density fit score. For example, the overlapping intervals of 1200 to 1500 and 1000 to 2000 are 1200 to 1500 with a length of 300, and their union is 1000 to 2000 with a length of 1000. The density fit score is 0.3. The average diameter at breast height (DBH) fit score and canopy closure fit score are calculated using the same method. The three fit scores are then weighted and summed with weights of 0.4, 0.4, and 0.2 to obtain the measure fit degree. For example, the measure fit degree corresponding to a density fit score of 0.3, a DBH fit score of 0.6, and a canopy closure fit score of 0.5 is 0.3 × 0.4 + 0.6 × 0.4 + 0.5 × 0.2 = 0.46.

[0148] A threshold of 0.3 is set for the fit of measures. The fit of measures is calculated for each developmental stage node's associated measure set, focusing on the pruning measures nodes. Nodes with a fit greater than or equal to the threshold are retained in the set, while those with a fit less than the threshold are removed. For example, for node 6, the three pruning measures are retained (0.46 fit for intermediate felling, 0.52 fit for pruning, and removed (0.28 fit for fertilization). This results in the effective pruning measure set for each developmental stage node. The pruning measure nodes are organized according to the temporal order of the developmental stage nodes on the optimal evolutionary path. The effective pruning measure set for the first developmental stage node in the node sequence is taken as the first-stage measure group, the set for the second node as the second-stage measure group, and so on, forming a multi-stage measure group sequence. For example, nodes 5, 6, 7, and 8 correspond to four stage measure groups.

[0149] Implementation timing parameters are assigned to each phase of measures, including the start time and duration. The start time is determined based on the position of the developmental stage node in the node sequence and the stand growth cycle. The start time of the first node is set to the current time, and the start time of subsequent nodes is the start time of the previous node plus the typical duration of that developmental stage. For example, the typical duration of the middle-aged forest stage is 8 years, the start time of node number 5 is the current year 2025, and the start time of node number 6 is 2033. The duration is set to 80% of the typical duration of that developmental stage to ensure that the measures are completed within the phase. The phase measure group sequence, the corresponding developmental stage node sequence, the implementation timing parameter sequence, and the candidate cumulative value of the optimal evolution path are encapsulated into a stand control scheme data structure. The stand control scheme includes a scheme identifier field, a target developmental stage field, a path quality score field, a phase measure sequence field, and a timing plan field. The scheme identifier field uniquely identifies the scheme using UUID format. The target developmental stage field stores the sequence number and name of the target developmental stage node, such as sequence number 8, mature forest stage. The path quality score field stores the candidate cumulative value of 8.19 for the optimal evolution path. The stage measure sequence field stores the measure node identifier and measure type for each stage; for example, the first stage includes thinning and pruning. The time series plan field stores the start year and duration of implementation for each stage; for example, the first stage starts in 2025 and lasts for 6 years. After the stand control scheme is generated, it is stored in the scheme repository of the map, supporting query, modification, and execution tracking operations.

[0150] This invention enables intelligent prediction of target developmental stages and automatic search for optimal paths based on historical accumulated experience. Combined with the semantic matching degree assessment of mulching measures and developmental stages, it can screen out the most suitable mulching measures for each developmental stage and organize them into a complete control plan in chronological order. This significantly improves the scientificity and applicability of diameter-based management decisions and provides precise technical support for sustainable forest management.

[0151] A second aspect of the present invention provides a simulation system for the inspection and evaluation of the entire process of forest tending, comprising:

[0152] The first unit is used to acquire ground measurement data of the target forest stand, construct a forest development stage knowledge graph including development stage entity nodes and quality indicator entity nodes, establish semantic association edges between each entity node, calculate quality indicator values ​​based on ground measurement data, match the current development stage node in the knowledge graph, and traverse the semantic association edges to obtain a set of semantic association paths.

[0153] The second unit is used to count the number of semantically related edges connecting the quality index entity nodes in the semantically related path set to obtain the graph structure connectivity group, extract the temporal position parameters of the nodes in the current development stage, generate a dynamic weight coefficient group based on the graph structure connectivity group and the temporal position parameters, and obtain the quality evaluation value by weighted summation of the quality index values.

[0154] The third unit is used to calculate the operational effect values ​​before and after the tending operation;

[0155] The fourth unit is used to add the quality evaluation value and the work effect value as time identifier attributes to the current development stage node, and create temporal evolution edges to adjacent development stage nodes. The weight of the temporal evolution edges is determined by the work effect value.

[0156] The fifth unit is used to extract historical time-series evolution paths from the knowledge graph to predict the target development stage nodes, and to query the tending measures nodes on the optimal evolution path from the current development stage node to the target development stage node to generate forest stand regulation schemes.

[0157] A third aspect of the present invention provides an electronic device, comprising:

[0158] processor;

[0159] Memory used to store processor-executable instructions;

[0160] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0161] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0162] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0163] Finally: The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, and these modifications do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A simulation method for the inspection and evaluation of the entire process of forest tending, characterized in that, include: Obtain ground measurement data of the target forest stand, construct a forest development stage knowledge graph including development stage entity nodes and quality indicator entity nodes, and establish semantic association edges between each entity node; calculate quality indicator values ​​based on ground measurement data, match the current development stage node in the knowledge graph, and traverse the semantic association edges to obtain a set of semantic association paths. The number of semantically related edges connecting the entity nodes of the quality index in the set of semantically related paths is used to obtain the graph structure connectivity group. The temporal position parameters of the nodes in the current development stage are extracted. A dynamic weight coefficient group is generated based on the graph structure connectivity group and the temporal position parameters. The quality index values ​​are weighted and summed to obtain the quality evaluation value. Calculate the operational effectiveness values ​​before and after the tending operation; The quality evaluation value and the work effect value are added as time identifier attributes to the current development stage node, and a temporal evolution edge is created to the adjacent development stage node. The weight of the temporal evolution edge is determined by the work effect value. Extract historical time-series evolution paths from the knowledge graph to predict target development stage nodes, and query tending measure nodes on the optimal evolution path from the current development stage node to the target development stage node to generate forest stand regulation schemes.

2. The method according to claim 1, characterized in that, The steps of acquiring ground measurement data of the target forest stand, constructing a knowledge graph of forest development stages, establishing semantic association edges between entity nodes, calculating quality index values ​​based on ground measurement data, matching the current development stage node in the knowledge graph, and traversing semantic association edges to obtain a set of semantic association paths include: The ground measurement data includes diameter at breast height (DBH) data, tree height data, and branch height data; The set of entity nodes for developmental stages is arranged in the order of forest growth cycle. The set of entity nodes for quality indicators includes entity nodes for diameter level indicators, entity nodes for first structural indicators, and entity nodes for second structural indicators. By extracting knowledge rules from forestry experts, semantic association edges are established between entity nodes in the development stage and entity nodes of quality indicators. The semantic association edges carry applicability weight attributes. The diameter-level index value is calculated using the diameter at breast height (DBH) data; the first structural index value is calculated using the ratio of the tree height data to the DBH data; and the second structural index value is calculated using the ratio of the branch height data to the tree height data. The diameter index value, the first structural index value, and the second structural index value are matched with the quality index entity nodes respectively. The current development stage node is obtained by backtracking through the semantic association edges of the matched quality index entity nodes. Starting from the node at the current development stage, traverse along the semantic association edges to the quality indicator entity node, and form a set of semantic association paths from the traversal paths.

3. The method according to claim 1, characterized in that, The steps for generating a dynamic weighted coefficient set based on the graph structure connectivity group and temporal location parameters, and then weighting and summing the quality index values ​​to obtain the quality evaluation value, include: The single-hop and double-hop connectivity of entity nodes in the semantic association path set are statistically analyzed, and the graph structure connectivity groups are obtained by summing the results after applying attenuation factors. The basic temporal position value is obtained by determining the sequence position of the current development stage node in the set of development stage entity nodes. The historical quality evaluation value sequence of the current development stage node is extracted and its rate of change is calculated to obtain the evolution trend coefficient. The evolution trend coefficient is used as a modulation factor to modulate the basic temporal position value to obtain the temporal position parameter. Multiply the graph structure connectivity group with the temporal position parameter to obtain the temporal modulation connectivity group. Obtain the cooperative relationship vector through the covariance matrix eigenvector of the temporal modulation connectivity group. Weight and fuse the temporal modulation connectivity group and the cooperative relationship vector and normalize them to obtain the dynamic weight coefficient group. The initial quality evaluation value is obtained by weighting the quality index values ​​using the dynamic weight coefficient group, and the quality evaluation value is obtained by adjusting the confidence level of the initial quality evaluation value using the time-series location parameter.

4. The method according to claim 3, characterized in that, The step of modulating the basic temporal position value with the evolution trend coefficient as a modulation factor to obtain the temporal position parameter includes: The short-term trend coefficient is obtained by calculating the first-order difference mean of the historical quality evaluation value sequence in the short-term window, the long-term trend coefficient is obtained by calculating the linear fitting slope in the long-term window, and the evolution acceleration coefficient is obtained by calculating the second-order difference of the whole sequence. The deviations of the short-term trend coefficient, long-term trend coefficient, and evolution acceleration coefficient from the historical average are calculated. Abnormal trend components with deviations exceeding a preset deviation threshold are smoothed and then fused with normal trend components to obtain a robust evolution feature vector. The modulation sensitivity factor is obtained based on the stage attribute parameters of the current development stage node. A nonlinear modulation function is selected from a preset modulation function library according to the modulation sensitivity factor. The modulation intensity parameter is obtained through the modulus of the robust evolution feature vector. The components of the robust evolutionary feature vector are used as modulation factors, and the basic time-series position value is modulated in multiple levels according to the selected nonlinear modulation function. The time-series position parameter is obtained by weighted fusion of the modulation results of each level through the modulation intensity parameter.

5. The method according to claim 1, characterized in that, The steps for adding quality evaluation values ​​and job performance values ​​as time-stamped attributes to the current development stage node and creating temporal evolution edges to adjacent development stage nodes include: The quality evolution trend coefficient is calculated based on the quality evaluation value of the current development stage node and the historical quality evaluation value sequence. The quality evaluation value, operation effect value, and quality evolution trend coefficient are encapsulated as a time identifier attribute and added to the current development stage node. From the set of entity nodes in the developmental stage, identify candidate developmental stage nodes that are adjacent to the current developmental stage node in terms of developmental time sequence. Calculate the developmental stage span parameter between each candidate developmental stage node and the current developmental stage node. Based on the developmental stage span parameter, select adjacent developmental stage nodes. The basic edge weight value is calculated based on the operation effect value and the quality evolution trend coefficient. The historical time identifier attribute of adjacent development stage nodes is extracted to calculate the historical quality fluctuation coefficient. The basic edge weight value is then corrected using the historical quality fluctuation coefficient to obtain the adjusted edge weight value. Create temporal evolution edges between the current development stage node and each adjacent development stage node, and assign the adjusted edge weight value to the temporal evolution edge as the edge weight.

6. The method according to claim 5, characterized in that, The steps for obtaining the adjusted edge weight value by correcting the basic edge weight value using the historical quality fluctuation coefficient include: The current quality contribution factor is obtained by weighted fusion of the operation effect value and the quality evolution trend coefficient. The basic edge weight value is calculated based on the ratio of the current quality contribution factor to the preset benchmark factor. Historical quality evaluation value sequences are extracted from the historical time identifier attributes of adjacent development stage nodes. The standard deviation of the historical quality evaluation value sequences is calculated to obtain the historical quality fluctuation amplitude. The historical quality fluctuation coefficient is calculated based on the ratio of the historical quality fluctuation amplitude to the mean of the historical evaluation values. A side weight correction function is constructed, and the historical quality fluctuation coefficient is used as a correction factor to input the side weight correction function to correct the basic side weight value to obtain the adjusted side weight value. The adjusted side weight value is inversely related to the historical quality fluctuation coefficient.

7. The method according to claim 1, characterized in that, The steps for generating stand management schemes by extracting historical time-series evolution paths from the knowledge graph to predict target developmental stage nodes, and querying tending measure nodes on the optimal evolution path from the current developmental stage node to the target developmental stage node, include: Extract all historical temporal evolution paths containing temporal evolution edges from the knowledge graph, calculate the historical cumulative value of the temporal evolution edge weights on each historical temporal evolution path, and take the endpoint development stage node of the historical temporal evolution path with the largest historical cumulative value as the target development stage node; Starting from the current developmental stage node and ending at the target developmental stage node, candidate evolutionary paths are searched in the knowledge graph. The cumulative candidate value of the temporal evolutionary edge weights on each candidate evolutionary path is calculated, and the candidate evolutionary path with the largest cumulative candidate value is selected as the optimal evolutionary path. Traverse the developmental stage nodes on the optimal evolutionary path, extract the nurturing measure nodes connected by semantic association edges for each developmental stage node, and calculate the measure fit based on the measure attribute parameters of the nurturing measure nodes and the stage attribute parameters of the developmental stage nodes; Based on the suitability of the measures, the tending measure nodes are selected, and the tending measure nodes are organized according to the temporal sequence of the developmental stage nodes on the optimal evolution path to generate a forest stand regulation scheme that includes the tending measure node sequence and the implementation time sequence.

8. A simulation system for the inspection and evaluation of the entire process of forest tending, used to implement the method described in any one of claims 1-7, characterized in that, include: The first unit is used to acquire ground measurement data of the target forest stand, construct a forest development stage knowledge graph including development stage entity nodes and quality indicator entity nodes, establish semantic association edges between each entity node, calculate quality indicator values ​​based on ground measurement data, match the current development stage node in the knowledge graph, and traverse the semantic association edges to obtain a set of semantic association paths. The second unit is used to count the number of semantically related edges connecting the quality index entity nodes in the semantically related path set to obtain the graph structure connectivity group, extract the temporal position parameters of the nodes in the current development stage, generate a dynamic weight coefficient group based on the graph structure connectivity group and the temporal position parameters, and obtain the quality evaluation value by weighted summation of the quality index values. The third unit is used to calculate the operational effect values ​​before and after the tending operation; The fourth unit is used to add the quality evaluation value and the work effect value as time identifier attributes to the current development stage node, and create temporal evolution edges to adjacent development stage nodes. The weight of the temporal evolution edges is determined by the work effect value. The fifth unit is used to extract historical time-series evolution paths from the knowledge graph to predict the target development stage nodes, and to query the tending measures nodes on the optimal evolution path from the current development stage node to the target development stage node to generate forest stand regulation schemes.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images