Tree growth prediction method, device and equipment and storage medium

Abstract

The invention discloses a tree growth prediction method, device and equipment and a storage medium, and the method comprises the steps: obtaining the related information of the tree growth, and the related information of the tree growth comprises basic information, growth environment information and growth state information; respectively carrying out tree classification and knowledge graph construction according to the growth vigor related information, and generating a tree initial information base; taking the tree initial information base as an initial training set, and performing sequential tree growth prediction training on the initial MHSA-LSTM model in combination with multiple pieces of real-time growth information of trees in different growth cycles to obtain an optimized MHSA-LSTM model; and adopting the optimized MHSA-LSTM model to predict the growth vigor of the tree in the future time period, and obtaining a prediction result of the growth vigor of the target tree. According to the method and the device, the technical problems that the interaction relationship among various different influence factors is not considered in the prior art, the model generalization ability is limited, various prediction requirements cannot be met, and the prediction result is lack of accuracy and reliability can be solved.

Description

Technical Field

[0001] This application relates to the technical field of target prediction, and particularly to a method, device, equipment and storage medium for predicting the growth trend of trees. Background Art

[0002] The accurate prediction of the growth trend of trees is crucial for ensuring the stable operation and efficient maintenance of the power system. Under complex natural environments and changing geographical conditions, the power department needs to accurately grasp the growth trend of trees to ensure the safe and stable operation and efficient maintenance of power lines. Modern time series prediction models provide strong support for predicting the growth trend of trees. By querying different professional tree information websites and databases, the power department can obtain information data on the growth states of the same type of trees corresponding to different periods. These information provide a comprehensive background for the growth of trees for the power department and help them more accurately predict the future growth trend of trees.

[0003] However, the existing technology mainly relies on qualitative analysis and empirical judgment of the tree growth environment, and there is a lack of integration means for multi-source data on the growth trend of trees. Coupled with the fact that tree growth is affected by multiple factors, the interaction relationship between different influencing factors cannot be captured in the prediction process, resulting in inaccurate prediction results. In addition, the generalization ability of the model used for predicting the growth trend is limited and cannot meet the prediction requirements for different tree species in different regions. Summary of the Invention

[0004] This application provides a method, device, equipment and storage medium for predicting the growth trend of trees, which is used to solve the technical problems that the existing technology does not consider the interaction relationship between multiple different influencing factors, and the generalization ability of the model is limited and cannot meet various prediction requirements, resulting in inaccurate and unreliable prediction results.

[0005] In view of this, the first aspect of this application provides a method for predicting the growth trend of trees, including:

[0006] Obtain information related to the growth trend of trees, where the information related to the growth trend includes basic information, growth environment information and growth state information;

[0007] Classify trees and construct a knowledge graph respectively according to the information related to the growth trend, and generate an initial information database of trees;

[0008] Use the initial information database of trees as an initial training set, and combine multiple real-time growth trend information of trees in different growth periods to perform time-series prediction training on the initial MHSA-LSTM model to obtain an optimized MHSA-LSTM model;

[0009] The optimized MHSA-LSTM model is used to predict the growth trend of trees in a future time period, and the predicted result of the target tree growth trend is obtained.

[0010] Preferably, the tree classification and knowledge graph construction are respectively carried out according to the growth trend-related information, and an initial tree information library is generated, including:

[0011] Tree classification is carried out according to the basic information in the growth trend-related information to obtain tree categories;

[0012] Based on the growth trend-related information, an entity relationship based on tree individuals, environmental factors, time factors and expert knowledge is constructed to generate a knowledge graph;

[0013] The tree categories and the knowledge graph are associated and integrated to generate an initial tree information library.

[0014] Preferably, taking the initial tree information library as an initial training set, and combining multiple real-time growth trend information of different growth cycles of trees to perform temporal tree growth trend prediction training on the initial MHSA-LSTM model to obtain an optimized MHSA-LSTM model, including:

[0015] Taking the initial tree information library as an initial training set and inputting it into the initial MHSA-LSTM model for growth trend prediction training in the initial growth stage to obtain initial prediction data;

[0016] Combining the real-time growth trend information of the next growth cycle of the tree and the initial prediction data to perform temporal tree growth trend prediction training on the initial MHSA-LSTM model of the next growth cycle to obtain an optimized MHSA-LSTM model.

[0017] Preferably, before taking the initial tree information library as an initial training set, and combining multiple real-time growth trend information of different growth cycles of trees to perform temporal tree growth trend prediction training on the initial MHSA-LSTM model to obtain an optimized MHSA-LSTM model, it also includes:

[0018] Combining the multi-head self-attention mechanism and the LSTM network to construct an MHSA-LSTM model to obtain an initial MHSA-LSTM model.

[0019] Preferably, taking the initial tree information library as an initial training set, and combining multiple real-time growth trend information of different growth cycles of trees to perform temporal tree growth trend prediction training on the initial MHSA-LSTM model to obtain an optimized MHSA-LSTM model, it also includes:

[0020] Adding the real-time growth trend information to the knowledge graph for information update, and generating an updated tree information library.

[0021] The second aspect of the present application provides a device for predicting the growth trend of trees, including:

[0022] An information acquisition unit for acquiring information related to the growth trend of trees, where the growth trend related information includes basic information, growth environment information, and growth status information;

[0023] An information analysis unit for classifying trees and constructing a knowledge graph based on the growth trend related information respectively, and generating an initial information database of trees;

[0024] A model training unit for using the initial information database of trees as an initial training set, and combining multiple real-time growth trend information of trees in different growth cycles to perform time-series tree growth trend prediction training on the initial MHSA-LSTM model to obtain an optimized MHSA-LSTM model;

[0025] A growth trend prediction unit for using the optimized MHSA-LSTM model to predict the growth trend of trees in a future time period to obtain a target tree growth trend prediction result.

[0026] Preferably, the information analysis unit is specifically used for:

[0027] Classifying trees according to the basic information in the growth trend related information to obtain tree categories;

[0028] Constructing entity relationships based on tree individuals, environmental factors, time factors, and expert knowledge according to the growth trend related information to generate a knowledge graph;

[0029] Associating and integrating the tree categories and the knowledge graph to generate an initial information database of trees.

[0030] Preferably, the model training unit is specifically used for:

[0031] Using the initial information database of trees as an initial training set and inputting it into the initial MHSA-LSTM model for growth trend prediction training in the initial growth stage to obtain initial prediction data;

[0032] Combining the real-time growth trend information of the next growth cycle of the tree and the initial prediction data to perform time-series tree growth trend prediction training on the initial MHSA-LSTM model of the next growth cycle to obtain an optimized MHSA-LSTM model.

[0033] The third aspect of the present application provides a device for predicting the growth trend of trees, and the device includes a processor and a memory;

[0034] The memory is used to store program codes and transmit the program codes to the processor;

[0035] The processor is used to execute the tree growth prediction method described in the first aspect according to the instructions in the program code.

[0036] The fourth aspect of this application provides a computer-readable storage medium, which is used to store program code, and the program code is used to execute a tree growth prediction method described in the first aspect.

[0037] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0038] In this application, a tree growth prediction method is provided, including: obtaining tree growth-related information, where the growth-related information includes basic information, growth environment information, and growth status information; classifying trees and constructing a knowledge graph based on the growth-related information respectively, and generating an initial tree information library; using the initial tree information library as an initial training set, and combining multiple real-time growth information of the tree in different growth cycles to perform temporal tree growth prediction training on the initial MHSA-LSTM model to obtain an optimized MHSA-LSTM model; using the optimized MHSA-LSTM model to predict the tree growth in a future time period to obtain a target tree growth prediction result.

[0039] The tree growth prediction method provided by this application integrates and correlates various growth-related information at different levels through the method of constructing a knowledge graph, and constructs an initial data information library; this process can not only fully consider the influence of different influencing factors on the tree growth, but also consider the relevant influence between different influencing factors, which is more in line with the actual situation, so it can ensure the accuracy of the growth prediction based on this; in addition, combining the initial tree information library and the real-time growth information in different growth cycles can perform temporal tree growth prediction training on the constructed MHSA-LSTM model, which can meet the prediction requirements of the tree in different growth cycles, that is, this process can ensure the generalization ability of the model and the reliability of the prediction result. Therefore, this application can solve the technical problems that the prior art does not consider the interaction relationship between various different influencing factors, and the generalization ability of the model is limited, and it cannot adapt to various prediction requirements, resulting in the lack of accuracy and reliability of the prediction result. Description of the Drawings

[0040] Figure 1 It is a flowchart of a tree growth prediction method provided by an embodiment of this application;

[0041] Figure 2 It is a structural diagram of a tree growth prediction device provided by an embodiment of this application;

[0042] Figure 3 It is a schematic diagram of the knowledge graph construction process provided by an embodiment of this application;

[0043] Figure 4 Schematic diagram of the MHSA-LSTM model structure provided by the embodiments of the present application;

[0044] Figure 5 Schematic diagram of the tree growth prediction process based on the MHSA-LSTM model and knowledge graph provided by the embodiments of the present application. Detailed implementation manners

[0045] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0046] For ease of understanding, please refer to Figure 1 , an embodiment of a tree growth prediction method provided by the present application, includes:

[0047] Step 101: Obtain information related to the growth of trees. The growth-related information includes basic information, growth environment information, and growth status information.

[0048] It should be noted that the basic principle of tree growth prediction is to integrate and analyze a large amount of tree growth status information, and use knowledge graph technology to deeply extract various influencing factors and their complex interaction relationships involved in the tree growth process. Subsequently, time series prediction algorithms in the field of deep learning are used to refine the processing and modeling of this information, so as to obtain accurate and reliable tree growth prediction results.

[0049] The data related to the growth of trees includes the basic information of trees, including but not limited to species, age, tree height, and diameter at breast height; growth environment factors, including but not limited to light intensity, light duration, precipitation, soil humidity, soil fertility, soil pH value, average temperature; growth status data, including but not limited to tree height growth amount, tree diameter growth amount, number and area of leaves, etc.

[0050] In this embodiment, by accessing professional tree information websites, a number of information closely related to the growth trend of trees is collected. The collected information includes the growth status of trees, specific tree species types, key growth climate factors, and geographical conditions. By deeply visiting professional websites in the field of tree information, detailed information on the growth trends of various trees is systematically collected. These information comprehensively cover the species classification of trees themselves, accurate growth status parameters, and external environmental factors that have a significant impact on their growth, specifically including key climate conditions and geographical features.

[0051] It can be understood that in order to ensure the accuracy and reliability of subsequent tree growth prediction, preprocessing operations can be performed on the obtained different types of influencing factors, that is, tree growth-related information. This can also be regarded as information collation, making these tree growth information show stronger regularity and practicability, and can also improve the quality of data information to ensure the robustness of subsequent data analysis. The specific preprocessing operation process is not limited in this embodiment and can be designed or selected according to the actual situation.

[0052] Step 102: Classify trees and construct a knowledge graph based on the growth-related information, and generate an initial tree information library.

[0053] Further, step 102 includes:

[0054] Classify trees according to the basic information in the growth-related information to obtain tree categories;

[0055] Construct entity relationships based on tree individuals, environmental factors, time factors, and expert knowledge according to the growth-related information to generate a knowledge graph;

[0056] Associate and integrate the tree categories and the knowledge graph to generate an initial tree information library.

[0057] It should be noted that the tree classification in this embodiment refers to classifying all trees according to the obtained basic information of the trees. For example, if there is a species category in the basic information, for the classification of trees in a certain area, the attributes of each variety of trees can be integrated into the tree categories in this area according to the classification of genus and species, which is also convenient for subsequent knowledge graph construction.

[0058] Information integration can be carried out according to different types of growth-related information to define entities closely related to tree growth and their relationships with each other. These entities specifically cover specific individual tree instances, diverse environmental factors, accurate time annotations, and detailed expert knowledge. Then, a set of mapping rules from the relational database to the tree ontology is formulated. According to the rules, the entities and relationships are stored in the form of a graphical structure to form a complete knowledge graph.

[0059] Please refer to Figure 3 , in this embodiment, data extraction is performed on the tree growth information, and then it is distinguished according to the degree of data structuring, and information integration is carried out according to different structured forms of data; then the data is converted into the form of triples in the preliminary knowledge expression; knowledge fusion is performed on the triple information. The main purpose of this process is entity disambiguation and co-reference disambiguation, so that a standard knowledge expression can be formed; after quality evaluation and knowledge discovery and reasoning, a complete tree growth knowledge graph can be constructed.

[0060] Integrating the classified information related to the growth of trees and the complex and refined associations between entities accurately extracted from the knowledge graph can form a comprehensive information dataset, namely the initial tree information database; this information database will be used in the training process of the time series prediction model; it is called the initial tree information database because new real-time growth-related information will be added to the information database according to different growth cycles in the subsequent prediction stage to update the information database.

[0061] Step 103: Use the initial tree information database as the initial training set, and combine the multiple real-time growth information of the trees in different growth cycles to perform time-series tree growth prediction training on the initial MHSA-LSTM model to obtain an optimized MHSA-LSTM model.

[0062] Further, step 103 includes:

[0063] Use the initial tree information database as the initial training set and input it into the initial MHSA-LSTM model for growth prediction training in the initial growth stage to obtain initial prediction data;

[0064] Combine the real-time growth information of the next growth cycle of the tree and the initial prediction data to perform time-series tree growth prediction training on the initial MHSA-LSTM model of the next growth cycle to obtain an optimized MHSA-LSTM model.

[0065] Further, before step 103, it also includes:

[0066] Construct an MHSA-LSTM model by combining the multi-head self-attention mechanism and the LSTM network to obtain the initial MHSA-LSTM model.

[0067] Further, step 103 also includes:

[0068] Add the real-time growth information to the knowledge graph for information update and generate an updated tree information database.

[0069] Step 104: Use the optimized MHSA-LSTM model to predict the growth of the tree in the future time period to obtain the target tree growth prediction result.

[0070] It should be noted that the MHSA-LSTM model in this embodiment is a fusion prediction model constructed based on the multi-head attention mechanism MHSA (Multi-Head Self-Attention) and the long short-term memory artificial neural network; MHSA can capture information at different levels and aspects in the input sequence, enhance the expressive ability of the model, and achieve multi-angle learning by parallel processing multiple self-attention modules, enabling the model to understand data more comprehensively and improving accuracy and robustness. Since the growth of trees is affected by various complex factors such as climate, soil quality, humidity, and competition from surrounding plants, and there may be complex interactions and dependencies among these factors, the multi-head self-attention mechanism that can capture these complex relationships and understand their impact on tree growth from multiple angles and levels has become the key to improving prediction accuracy and reliability. LSTM can effectively handle the long-term dependence problem in long sequences and capture context information, effectively solving the problem that the growth of trees is affected by multiple environmental factors, resulting in the inability to effectively capture time series information.

[0071] Please refer to Figure 4 , the input data set of the MHSA-LSTM model in this embodiment includes various influencing factors related to tree growth, such as climate data, soil parameters, and historical growth records, etc.; the tree initial information library contains all the basic information, growth status, environmental parameters, and expert knowledge of trees in the initial prediction stage; as different growth data and prediction data at different growth stages are input into the model, time series prediction analysis based on LSTM can be realized. It can be understood that the data input into the MHSA-LSTM model is in time series format.

[0072] Among them, there are three gate structures in the LSTM unit, the forget gate , the input gate , and the output gate . Among them, and respectively represent time and time of the hidden state, and respectively represent time and time of the memory cell state, represents time of the input sequence; the forget gate determines how much information of the cell state at time should be forgotten; the input gate determines the input at time for time memory cell state The degree of update, where the tanh activation function is calculated to obtain The candidate value of the memory cell state at a moment , so as to control How much information should be saved; the output gate Determine The memory cell state at a moment How much information needs to be output to the hidden state . The internal calculation formula of the LSTM layer is as follows:

[0073]

[0074]

[0075]

[0076]

[0077]

[0078]

[0079] Among them, Is the activation function, , Are different weights in each network structure respectively, Is the different bias value in each network structure, Is the Hadamard product.

[0080] The output of the hidden layer of the LSTM is divided into h subspaces, and h heads respectively focus on the hidden information of different scale data in each subspace. The multi-head self-attention mechanism MHSA receives the output from the hidden layer of the LSTM network and can be expressed as:

[0081]

[0082] Among them, Is the model time step.

[0083] In the calculation process of MHSA, the input needs to be converted into a query matrix , key matrix And value matrix ; the query matrix Is responsible for calculating the attention weights, the key matrix Calculates the key information of the input sequence through linear transformation, and the value matrix Provides position information for each input position. Use Multiple parallel self-attention heads process different parts of the input sequence simultaneously. The self-attention branch is calculated through scaled dot product, which can be specifically expressed as:

[0084]

[0085]

[0086]

[0087]

[0088] Among them, is the LSTM output vector, , , are the parameter matrices corresponding to the query matrix , the key matrix and the value matrix respectively. The softmax function can map the outputs of multiple neurons to . is the read size of the query matrix and the key matrix . represents a scaling factor used to balance the size of the dot product result, avoiding the dot product result being too large due to too high vector dimensions, which in turn makes the gradient of the softmax function very small.

[0089] The attention output of each head can be expressed as:

[0090]

[0091] The formula for multi-head attention calculation is expressed as:

[0092]

[0093] Among them, , , are all the parameter matrices of the i-th head, is the attention output of the i-th head, is the learnable linear transformation matrix, is the concatenation operation, is the output result of the multi-head self-attention layer. Multiply the extracted feature vector by its corresponding feature weight to obtain the weighted result and input it into the fully connected layer to achieve the prediction of key quality variables; The above is the specific structure of the MHSA-LSTM model in this application and the calculation process in the network.

[0094] Since the MHSA-LSTM model needs to be used in this embodiment to accurately predict the growth trends of trees in different growth cycles, for the prediction of the growth trends of trees in each growth cycle, it is necessary not only to obtain the data of the current growth environment and the growth characteristic parameters of the trees, but also to conduct a combined analysis of the past historical growth states. The so-called historical growth state refers to the prediction results of the growth trends of trees obtained from the prediction of historical growth stages. In this way, the prediction results of each growth stage can be more accurate and reliable. Therefore, the initial tree information database, as the initial training set, is used to train the initial MHSA-LSTM model for predicting growth trends, and the model can be trained to generate prediction data of the growth trends of trees. In the prediction task of the next growth stage, the model will combine the new growth state and environmental parameters, as well as the prediction data generated in the previous growth stage, to predict the growth trends of trees and obtain the prediction results of the growth trends of trees in this growth stage. That is to say, in each growth stage, the inputs of the model are different, which are reflected in the obtained growth state and environmental parameters, as well as the prediction data of the previous stage input. In this way, it can be ensured that more practical prediction results can be obtained in each growth stage, and the model is more in line with the actual application requirements.

[0095] For the process of predicting the growth trends of trees by combining the knowledge graph and the network model in this embodiment, please refer to Figure 5 . From the analysis of the growth cycle of trees, the model prediction processes at all growth time points of the target tree can be linked to form a complete growth trend prediction process of the trees. Among them, the input of the tree growth trend prediction model consists of the current environmental parameters and the output of the superior MHSA-LSTM model. represents the input environmental parameters at the nth level. The left number of the subscript comma is the level number of the MHSA-LSTM model in different growth stages, and the right side of the comma represents the type and serial number of the input environmental parameters. represents the output of the nth level MHSA-LSTM model, that is, the prediction data of the tree growth.

[0096] In addition, it should be noted that the knowledge graph of the initial tree information database in this example is only the initial knowledge graph. As the growth stage progresses, more real-time growth information such as the growth state and environmental parameters of the trees in the current growth stage is obtained and participated in the model training. These data and information will also be synchronously updated to the initial knowledge graph, and then the updated tree information database is generated; the information related to the growth trends of the trees in the information database can be enriched. It can be understood that the so-called real-time growth information only refers to a certain recording time point in a certain growth cycle and does not have the real-time expressiveness of the actual scenario. In addition, the growth cycle of trees can be divided into growth stages such as the seedling stage, the growth stage, the mature stage, and the senescence stage. The growth laws of trees themselves and the external environmental requirements are different in different stages. Therefore, when predicting the growth trends in each stage, it is necessary to obtain the current growth state and environmental parameters so as to obtain accurate and reliable prediction results.

[0097] The prediction model in this embodiment combines a knowledge graph and a neural network architecture, which can comprehensively consider the impacts of various environmental factors on the growth of trees and reduce the prediction errors caused by single factors or simple linear relationships. By continuously optimizing the model parameters and training strategies, this embodiment can further improve the accuracy and stability of the prediction results, providing more reliable decision-making support for fields such as forestry management and ecological protection.

[0098] The method for predicting the growth of trees provided by the embodiment of the present application integrates and correlates various growth-related information at different levels by constructing a knowledge graph, and constructs an initial database of data information; this process can not only fully consider the impacts of different influencing factors on the growth of trees, but also take into account the mutual influences between different influencing factors, which is more in line with the actual situation, so it can ensure the accuracy of the growth prediction based on this; in addition, combining the initial information database of trees and the real-time growth information in different growth cycles can perform time-series prediction training on the constructed MHSA-LSTM model for the growth of trees, which can meet the prediction requirements of trees in different growth cycles, that is, this process can ensure the generalization ability of the model and the reliability of the prediction results. Therefore, the embodiment of the present application can solve the technical problems that the prior art does not consider the interaction relationship between various different influencing factors, and the generalization ability of the model is limited and cannot adapt to various prediction requirements, resulting in the lack of accuracy and reliability of the prediction results.

[0099] For ease of understanding, please refer to Figure 2 , an embodiment of a device for predicting the growth of trees provided by the present application includes:

[0100] An information acquisition unit 201, configured to acquire information related to the growth of trees, and the information related to the growth includes basic information, growth environment information, and growth status information;

[0101] An information analysis unit 202, configured to classify trees and construct a knowledge graph respectively according to the information related to the growth, and generate an initial information database of trees;

[0102] A model training unit 203, configured to use the initial information database of trees as an initial training set, and perform time-series prediction training on the initial MHSA-LSTM model in combination with multiple real-time growth information of trees in different growth cycles to obtain an optimized MHSA-LSTM model;

[0103] A growth prediction unit 204, configured to use the optimized MHSA-LSTM model to predict the growth of trees in a future time period to obtain a target tree growth prediction result.

[0104] Further, the information analysis unit 202 is specifically configured to:

[0105] Classify trees according to the basic information in the growth-related information to obtain tree categories;

[0106] Construct an entity relationship based on tree individuals, environmental factors, time factors, and expert knowledge according to the growth-related information to generate a knowledge graph;

[0107] Associate and integrate the tree categories and the knowledge graph to generate an initial tree information database.

[0108] Furthermore, the model training unit 203 is specifically used for:

[0109] Use the initial tree information database as an initial training set to input into the initial MHSA-LSTM model for growth trend prediction training in the initial growth stage to obtain initial prediction data;

[0110] Combine the real-time growth trend information of the next growth cycle of the tree and the initial prediction data to perform temporal tree growth trend prediction training on the initial MHSA-LSTM model of the next growth cycle to obtain an optimized MHSA-LSTM model.

[0111] The third aspect of the present application provides a tree growth trend prediction device, which includes a processor and a memory;

[0112] The memory is used to store program codes and transmit the program codes to the processor;

[0113] The processor is used to execute the tree growth trend prediction method of the first aspect according to the instructions in the program codes.

[0114] The fourth aspect of the present application is a computer-readable storage medium, which is used to store program codes, and the program codes are used to execute a tree growth trend prediction method of the first aspect.

[0115] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point, the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the devices or units can be in an electrical, mechanical or other form.

[0116] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0117] In addition, each functional unit in various embodiments of the present application may be integrated in a processing unit, may exist independently physically for each unit, or two or more units may be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0118] If the above integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to execute all or part of the steps of the methods described in various embodiments of the present application by a computer device (which can be a personal computer, a server, or a network device, etc.). The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (English full name: Read-Only Memory, English abbreviation: ROM), random access memories (English full name: Random Access Memory, English abbreviation: RAM), magnetic disks, or optical discs and other various media that can store program codes.

[0119] As mentioned above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of various embodiments of the present application.

Claims

1. A tree growth prediction method, characterized in that: include: Acquire tree growth related information, wherein the tree growth related information includes basic information, growth environment information and growth status information; Classifying trees and constructing knowledge graphs according to the growth-related information, and generating an initial tree information database; The tree initial information library is used as an initial training set, and the initial MHSA-LSTM model is trained for time-series tree growth prediction in combination with a plurality of real-time growth information of trees in different growth cycles to obtain an optimized MHSA-LSTM model; The optimized MHSA-LSTM model is used to predict the growth of trees in the future time period to obtain the growth prediction results of the target trees.

2. The tree growth prediction method according to claim 1, characterized in that: The tree classification and knowledge graph construction are performed according to the growth-related information, and an initial tree information base is generated, including: Classify trees according to the basic information in the growth related information to obtain tree categories; According to the growth-related information, entity relationships based on individual trees, environmental factors, time factors and expert knowledge are constructed to generate a knowledge graph; The tree categories and the knowledge graph are associated and integrated to generate an initial tree information base.

3. The tree growth prediction method according to claim 1, characterized in that: The tree initial information library is used as an initial training set, and a plurality of real-time growth information of trees in different growth cycles are combined to perform time-series tree growth prediction training on the initial MHSA-LSTM model to obtain an optimized MHSA-LSTM model, including: Inputting the tree initial information database as an initial training set into the initial MHSA-LSTM model to perform growth prediction training in the initial growth stage to obtain initial prediction data; The initial MHSA-LSTM model of the next growth cycle is trained for time-series tree growth prediction by combining the real-time growth information of the next growth cycle of the tree with the initial prediction data to obtain an optimized MHSA-LSTM model.

4. The tree growth prediction method according to claim 1, characterized in that: The method uses the tree initial information library as an initial training set, combines a plurality of real-time growth information of trees in different growth cycles to perform time-series tree growth prediction training on the initial MHSA-LSTM model, and obtains an optimized MHSA-LSTM model, which also includes: The MHSA-LSTM model is constructed by combining the multi-head self-attention mechanism and the LSTM network to obtain the initial MHSA-LSTM model.

5. The tree growth prediction method according to claim 1, characterized in that: The method uses the tree initial information library as an initial training set, combines a plurality of real-time growth information of trees in different growth cycles to perform time-series tree growth prediction training on the initial MHSA-LSTM model, and obtains an optimized MHSA-LSTM model, and further includes: The real-time growth information is added to the knowledge graph to update the information and generate a tree update information library.

6. A tree growth prediction device, characterized in that: include: An information acquisition unit, used to acquire tree growth related information, wherein the growth related information includes basic information, growth environment information and growth status information; An information analysis unit, used to classify trees and construct knowledge graphs according to the growth-related information, and generate an initial tree information base; A model training unit is used to use the tree initial information library as an initial training set, and combine multiple real-time growth information of trees in different growth cycles to perform time-series tree growth prediction training on the initial MHSA-LSTM model to obtain an optimized MHSA-LSTM model; The growth prediction unit is used to predict the growth of trees in the future time period by using the optimized MHSA-LSTM model to obtain the growth prediction result of the target tree.

7. The tree growth prediction device according to claim 6, characterized in that: The information analysis unit is specifically used for: Classify trees according to the basic information in the growth related information to obtain tree categories; According to the growth-related information, entity relationships based on individual trees, environmental factors, time factors and expert knowledge are constructed to generate a knowledge graph; The tree categories and the knowledge graph are associated and integrated to generate an initial tree information base.

8. The tree growth prediction device according to claim 6, characterized in that: The model training unit is specifically used for: Inputting the tree initial information database as an initial training set into the initial MHSA-LSTM model to perform growth prediction training in the initial growth stage to obtain initial prediction data; The initial MHSA-LSTM model of the next growth cycle is trained for time-series tree growth prediction by combining the real-time growth information of the next growth cycle of the tree with the initial prediction data to obtain an optimized MHSA-LSTM model.

9. A tree growth prediction device, characterized in that: The device comprises a processor and a memory; The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the tree growth prediction method according to any one of claims 1-5 according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store program codes, and the program codes are used to execute the tree growth prediction method according to any one of claims 1-5.

Images