Method and device for predicting variety adaptability evaluation result

Abstract

The invention provides a variety adaptability evaluation result prediction method and device, and is applied to the technical field of agricultural information processing. The method comprises the steps that a first knowledge graph of varieties and environments is acquired, nodes of the first knowledge graph are varieties or planting areas, and the edges of the first knowledge graph comprise the edges between two planting areas with the environment similarity higher than a first threshold value and the edges between the varieties and the planting areas with the planting relation; based on the first knowledge graph, a knowledge graph is reconstructed with nodes of a planting promotion area of the variety as center nodes, a second knowledge graph is obtained, and the number of the nodes of the planting area of the second knowledge graph is larger than the number of the nodes of the planting area of the first knowledge graph; and based on the second knowledge graph, training the conditional variation auto-encoder model, sampling a different implicit vector generated by the conditional variation auto-encoder model in each training iteration as additional input, and obtaining a variety adaptability evaluation result prediction model.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural information processing, and in particular to a method and device for predicting variety adaptability evaluation results. Background Art

[0002] In agricultural production, choosing suitable varieties is crucial to improving yield and quality. With the reform of my country's crop variety approval system and the introduction of independent testing channels such as green channels and consortia, the approved crop varieties have shown a "blowout" growth. Many new varieties, coupled with complex and diverse planting environments and climate changes, have brought great difficulties to the precise positioning and planting of crop varieties. How to scientifically and rationally determine the best suitable planting area for crop varieties has become an urgent problem to be solved.

[0003] In the prior art, most commonly used methods for locating suitable planting areas for varieties are based on statistical models or traditional machine learning models. For example, statistical models can perform simple statistical analysis based on a large amount of past planting data, and traditional machine learning models can use some conventional classification and regression algorithms to determine the areas where varieties are suitable for planting.

[0004] However, these traditional models have obvious shortcomings and their learning ability is relatively limited. There are actually very complex interactions between varieties and the environment, and these complex interactions are multidimensional and nonlinear. Statistical models and traditional machine learning models are difficult to fully capture, understand and utilize these complex relationships, resulting in inaccuracy in determining the suitable planting areas for varieties, and the inability to fully consider the impact of various complex factors, which in turn affects the scientificity and rationality of planting decisions. Summary of the invention

[0005] The present invention provides a method and device for predicting variety adaptability evaluation results, which are used to solve the problem that the prior art is not accurate enough in determining the suitable planting area of ​​the variety.

[0006] The present invention provides a method for predicting variety adaptability evaluation results, comprising: obtaining a first knowledge graph of varieties and environments, wherein nodes of the first knowledge graph are varieties or planting areas, and edges of the first knowledge graph include edges between two planting areas with environmental similarity higher than a first threshold, and edges between varieties and planting areas with a planting relationship; based on the first knowledge graph, reconstructing the knowledge graph with nodes of promoted planting areas of the varieties as central nodes to obtain a second knowledge graph, wherein the number of nodes of planting areas of the second knowledge graph is greater than the number of nodes of planting areas of the first knowledge graph; based on the second knowledge graph, training a conditional variational autoencoder model, sampling a different latent vector generated by the conditional variational autoencoder model as an additional input in each training iteration, to obtain a variety adaptability evaluation result prediction model; wherein the variety adaptability evaluation result prediction model is used to predict the adaptability of varieties under multiple environments.

[0007] According to a method for predicting variety adaptability evaluation results provided by the present invention, the first knowledge graph of obtaining varieties and environments includes: obtaining planting data of target varieties under multiple environments, the planting data including whether the target variety is planted in a planting area, meteorological data of the planting area, and soil data of the planting area; cleaning and standardizing the planting data; determining varieties and planting areas with planting relationships based on the processed planting data, and connecting the varieties and planting areas with planting relationships to form edges; determining environmental characteristics of the planting area based on the meteorological data and soil data of the processed planting data, and determining the environmental similarity between two planting areas based on the environmental characteristics, and connecting two planting areas with environmental similarity higher than the first threshold to form an edge.

[0008] According to a method for predicting variety adaptability evaluation results provided by the present invention, the conditional variational autoencoder model is trained based on the second knowledge graph, and a different latent vector generated by the conditional variational autoencoder model is sampled as an additional input in each training iteration to obtain a variety adaptability evaluation result prediction model, including: determining input data from the second knowledge graph, the input data including central node features and neighbor node features; inputting the input data into the encoder of the conditional variational autoencoder model to obtain a latent vector; inputting the latent vector and the central node features into the decoder of the conditional variational autoencoder model to obtain a target feature vector associated with the central node; wherein the target feature vector is used to indicate the adaptability evaluation index of the variety in different planting areas.

[0009] According to a method for predicting variety adaptability evaluation results provided by the present invention, the input data is input into the encoder of the conditional variational autoencoder model to obtain a latent vector, including: mapping the input data to a latent variable space through the encoder of the conditional variational autoencoder model, and sampling from the latent variable space to obtain the latent vector.

[0010] The present invention also provides a device for predicting variety adaptability evaluation results, comprising the following modules: an acquisition module and a processing module; the acquisition module is used to acquire a first knowledge graph of varieties and environments, the nodes of the first knowledge graph are varieties or planting areas, and the edges of the first knowledge graph include edges between two planting areas with environmental similarity higher than a first threshold, and edges between varieties and planting areas with a planting relationship; the processing module is used to reconstruct the knowledge graph based on the first knowledge graph, with the node of the promoted planting area of ​​the variety as the central node, to obtain a second knowledge graph, the number of nodes of the planting area of ​​the second knowledge graph is greater than the number of nodes of the planting area of ​​the first knowledge graph; based on the second knowledge graph, a conditional variational autoencoder model is trained, and in each training iteration, a different latent vector generated by the conditional variational autoencoder model is sampled as an additional input to obtain a variety adaptability evaluation result prediction model; wherein the variety adaptability evaluation result prediction model is used to predict the adaptability of varieties under multiple environments.

[0011] According to a device for predicting the results of variety adaptability evaluation provided by the present invention, the acquisition module is used to acquire the planting data of the target variety under multiple environments, the planting data including whether the target variety is planted in the planting area, the meteorological data of the planting area and the soil data of the planting area; the processing module is used to clean and standardize the planting data; determine the varieties and planting areas with a planting relationship based on the processed planting data, and connect the varieties and planting areas with a planting relationship to form an edge; determine the environmental characteristics of the planting area based on the meteorological data and soil data of the processed planting data, determine the environmental similarity between two planting areas based on the environmental characteristics, and connect two planting areas with an environmental similarity higher than the first threshold to form an edge.

[0012] According to a device for predicting variety adaptability evaluation results provided by the present invention, the processing module is used to determine input data from the second knowledge graph, and the input data includes central node features and neighbor node features; the input data is input into the encoder of the conditional variational autoencoder model to obtain a latent vector; the latent vector and the central node features are input into the decoder of the conditional variational autoencoder model to obtain a target feature vector associated with the central node; wherein the target feature vector is used to indicate the adaptability evaluation index of the variety in different planting areas.

[0013] According to a variety adaptability evaluation result prediction device provided by the present invention, the processing module is used to map the input data to a latent variable space through the encoder of the conditional variational autoencoder model, and obtain the latent vector by sampling from the latent variable space.

[0014] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for predicting the results of variety adaptability evaluation as described in any one of the above-mentioned methods is implemented.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described methods for predicting variety adaptability evaluation results.

[0016] The present invention also provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-mentioned methods for predicting variety adaptability evaluation results.

[0017] The method and device for predicting the results of variety adaptability evaluation provided by the present invention can construct a second knowledge graph based on the first knowledge graph of variety and environment with the node of the promoted planting area of ​​the variety as the central node, and train the conditional variational autoencoder model based on the second knowledge graph. By constructing a heterogeneous knowledge graph of variety and environment, the complex interaction relationship between variety and environment can be deeply explored. The knowledge graph is reconstructed with the node of the promoted planting area of ​​the variety as the central node to obtain a second knowledge graph with a larger number of nodes than the first knowledge graph, so that local enhancement of the knowledge graph can be achieved. By adopting the local enhancement technology, the neighborhood features can be generated by the conditional variational autoencoder model based on the local structure and node features, thereby overcoming the problem that limited neighbor nodes are not conducive to learning node representation, and thus significantly improving the prediction accuracy of the conditional variational autoencoder model. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 It is a flow chart of the method for predicting variety adaptability evaluation results provided by the present invention; Figure 2 It is a structural schematic diagram of a variety adaptability evaluation result prediction device provided by the present invention; Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0021] It should be noted that, in the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0022] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0023] In order to facilitate a clear description of the technical solutions of the embodiments of the present application, in the embodiments of the present application, words such as "first" and "second" are used to distinguish between identical or similar items with substantially the same functions and effects. Those skilled in the art can understand that words such as "first" and "second" do not limit the quantity and order of execution.

[0024] The embodiments of the present application describe some exemplary embodiments for the purpose of explanation. It should be understood that the present application can be implemented in other ways that are not specifically shown in the drawings.

[0025] like Figure 1 As shown, the embodiment of the present application provides a method for predicting the results of a variety adaptability evaluation, which can be applied to a device for predicting the results of a variety adaptability evaluation. The method for predicting the results of a variety adaptability evaluation can include S101-S103: S101. A device for predicting variety adaptability evaluation results obtains a first knowledge graph of variety and environment.

[0026] Among them, the nodes of the first knowledge graph are varieties or planting areas, and the edges of the first knowledge graph include edges between two planting areas whose environmental similarity is higher than a first threshold, and edges between varieties and planting areas with planting relationships.

[0027] Optionally, the first knowledge graph for obtaining varieties and environments includes: obtaining planting data of target varieties under multiple environments, the planting data including whether the target variety is planted in a planting area, meteorological data of the planting area, and soil data of the planting area; cleaning and standardizing the planting data; determining varieties and planting areas with planting relationships based on the processed planting data, and connecting the varieties and planting areas with planting relationships to form edges; determining environmental characteristics of the planting area based on the meteorological data and soil data of the processed planting data, and determining the environmental similarity between two planting areas based on the environmental characteristics, and connecting two planting areas with environmental similarity higher than the first threshold to form an edge.

[0028] Specifically, the variety adaptability evaluation result prediction device must first collect the planting data of the target variety in multiple environments. The planting data covers the relevant information when the target variety is planted in different environments, such as growth period, yield, disease resistance and other trait indicators. These data can reflect the various performance characteristics of the variety in actual planting. At the same time, meteorological data and soil data of the planting area and meteorological data and soil data of the unplanted area must also be collected. Among them, meteorological data includes temperature, precipitation, humidity, sunshine hours, etc., and soil data involves soil pH, organic matter, total nitrogen, available phosphorus and other indicators. By obtaining these data, the environmental conditions of different planting areas can be fully grasped, whether it is a place where planting activities are actually carried out, or a place that has not been planted but may have potential planting value.

[0029] Afterwards, the variety adaptability evaluation result prediction device needs to clean and standardize the planting data. The data cleaning process includes: 1. In the process of collecting data, duplicate records may appear due to various reasons (such as multiple records, data import errors, etc.). These duplicate data will interfere with subsequent analysis, so they must be removed to ensure the uniqueness of each data. 2. Some data may be missing due to measurement errors, record omissions, etc., and need to be processed by appropriate methods, such as mean filling, adjacent value filling or more complex interpolation algorithms, etc., to make the data complete and avoid analysis bias due to missing data. 3. If there is data that is obviously inconsistent with common sense or contradictory to other relevant data (for example, the temperature records in a certain area are obviously beyond the normal temperature range of the area, etc.), it is necessary to verify and correct it to ensure the accuracy and rationality of the data.

[0030] Since the data collected come from different sources, their magnitude and dimensions may vary. In order to make these data play a fair role in subsequent analysis and model building, they need to be normalized and standardized. For example, the yield data may be a large value of several hundred kilograms, while the soil pH value is a small value between 0-14. If it is not processed, the impact of the yield data in subsequent calculations may far exceed the pH data.

[0031] Optionally, the normalization method includes but is not limited to Z-Score normalization, Max-Min normalization, standard deviation normalization, etc. Take Z-Score normalization as an example. The variety adaptability evaluation result prediction device can be based on the formula Normalize the data, where It is one of the indicators of variety phenotypic data, meteorological and soil environmental data of planting sites and unplanted sites. is the average value, is the standard deviation.

[0032] Finally, the variety adaptability evaluation result prediction device can extract key features from the collected and processed planting data, and organize these features into feature vectors to comprehensively represent the environmental characteristics of each location. For example, for a certain planting area, the processed key features such as temperature, humidity, soil pH, etc. are combined to form a vector, and this vector represents the environmental conditions of the planting area. The cosine similarity calculation formula is then used to quantify the similarity between the environmental characteristics of different planting areas. Information such as varieties and planting areas are used as nodes. For example, a specific crop variety can be a node, and a planting area or unplanted area can also be used as a node. In this way, the basic elements of the graph are constructed. If a variety is planted in a certain planting area, then the variety is connected to the corresponding planting area to indicate that there is an actual planting association between them. And according to the environmental feature similarity results calculated previously, nodes with similar environmental features (such as different planting areas in similar environments) are connected to form edges, and the relationship between different nodes is reflected through the edges. Finally, the first knowledge graph of heterogeneous interactions between varieties and environments is constructed, and it is used To indicate that Represents a node set, including all varieties and planting areas, etc. Represents an edge set, which reflects the association between nodes based on planting relationships and environmental similarities.

[0033] S102. The variety adaptability evaluation result prediction device reconstructs the knowledge graph based on the first knowledge graph, taking the node of the variety's promotion and planting area as the central node to obtain a second knowledge graph.

[0034] Among them, the number of nodes of the planting area of ​​the second knowledge graph is greater than the number of nodes of the planting area of ​​the first knowledge graph.

[0035] Specifically, the variety adaptability evaluation result prediction device can construct a graph data set specifically used for training the model based on the first knowledge graph to obtain the second knowledge graph. .

[0036] First, determine the core nodes. The variety adaptability evaluation result prediction device can use the nodes of the actual promotion areas of the variety as the central nodes. For example, if a certain crop variety is promoted and planted in multiple regions, then the nodes corresponding to these promotion areas are the core of attention when constructing the training data set. They are selected as central nodes because the model is likely to predict and analyze the relevant conditions of the variety in different promotion areas in the future.

[0037] Then, determine the neighbor nodes. Second Knowledge Graph The scope includes the neighboring nodes of the central node, which means that we not only focus on the node of the variety promotion area itself, but also include the nodes directly related to it. For example, other location nodes around the promotion area node that are connected to it (may be surrounding areas with similar environmental characteristics), or other related nodes that have some connection with the promoted variety (such as planting association under similar conditions, etc.) should be included. Doing so can make full use of the correlation between nodes in the graph structure, making the training data more relevant and holistic, thereby better reflecting the various complex relationships between the variety and its promotion environment.

[0038] After that, expand the data range. The data set consisting only of the variety promotion region node and its neighboring nodes may have limited data volume, which is not enough for the model to fully learn the rules and relationships therein. In order to meet the model's demand for data volume, The construction requires the graph dataset Appropriately add some nodes that are indirectly related to the central node but are also valuable for analyzing the promotion of varieties, as well as the edges between them, so as to increase the amount of training data and make the subsequently trained model more generalizable and better cope with various actual prediction scenarios.

[0039] S103. The device for predicting the results of variety adaptability evaluation trains the conditional variational autoencoder model based on the second knowledge graph, samples a different latent vector generated by the conditional variational autoencoder model as an additional input in each training iteration, and obtains a model for predicting the results of variety adaptability evaluation.

[0040] Among them, the variety adaptability evaluation result prediction model is used to predict the adaptability of the variety in multiple environments.

[0041] Optionally, the variety adaptability evaluation result prediction device can determine input data from the second knowledge graph, and the input data includes central node features and neighbor node features; input the input data into the encoder of the conditional variational autoencoder model to obtain a latent vector; input the latent vector and the central node features into the decoder of the conditional variational autoencoder model to obtain a target feature vector associated with the central node; wherein the target feature vector is used to indicate the adaptability evaluation index of the variety in different planting areas.

[0042] Optionally, the input data is input into the encoder of the conditional variational autoencoder model to obtain a latent vector, including: mapping the input data to a latent variable space through the encoder of the conditional variational autoencoder model, and sampling from the latent variable space to obtain the latent vector.

[0043] Specifically, the training process includes: 1. Determine the input data: As a condition, here It is the characteristic data of the central node (that is, the node of the variety promotion area), and the adjacent paired nodes As input, It is the characteristic data of each neighbor node (node ​​directly associated with the central node).

[0044] 2. Mapping to latent variable space and sampling: Mapping these input data to a latent variable space through the encoder network , sampling in this space to obtain a hidden vector The encoder network is used to extract features from the input data and transform it into a latent space, that is, to transform high-dimensional and complex node feature data into a more abstract latent vector representation that is more conducive to subsequent generation.

[0045] 3. The decoder generates a new feature vector: the latent vector obtained by sampling and central node The eigenvector of Together as the input of the decoder network, after being processed by the decoder, the target feature vector associated with the central node is finally obtained The target feature vector is the new feature representation related to the central node learned by the conditional variational autoencoder model based on the input data. It integrates the information of the latent space and the conditional information of the central node itself, and can mine more in-depth relationships and features.

[0046] For example, taking into account a variety of agronomic traits, the entropy weight method is used to calculate the adaptability evaluation index of corn varieties based on lodging rate, stem rot, ear rot, curvularia leaf spot, leaf spot, smut, southern rust, growth period and per mu yield. The calculation process of the adaptability evaluation index includes: (1) Assume that the corn variety evaluation index system includes Varieties and Indicators, the original matrix is: ; (2) In order to avoid inconsistency in the units of different indicator characteristics, standardization measures are taken for each indicator.

[0047] Normalization of positive indicators (e.g. production): ; here is the data in the original matrix, It is the minimum value of this indicator among all varieties. It is the maximum value. The standardized value range is between 0 and 1. The larger the value, the better the performance of the variety on this positive indicator.

[0048] Standardization of negative indicators (such as disease indicators): ; Similarly, after standardization, It is also between 0 and 1, but the larger the value is, the lighter the disease situation of the variety on this negative indicator, that is, the better the performance.

[0049] (3) Standardize the original matrix , expressed as ,in Represents the weight, and its formula is: ; Through this formula, the weight of each variety on each indicator is calculated, and the size of the weight reflects the relative importance of the variety on this indicator compared with other varieties.

[0050] (4) Define indicators The entropy value of is: ; It should be noted that entropy is used to measure the uncertainty or information content of an indicator. The larger the entropy value, the smaller the difference between different varieties of the indicator, and the less effective information provided; the smaller the entropy value, the greater the difference between different varieties of the indicator, and the more effective information provided.

[0051] (5) Calculate the Entropy weight of an indicator , the formula is: ; The entropy weight reflects the weight determined according to the amount of information of the indicator. The larger the entropy weight, the more important the indicator is in the comprehensive evaluation, because it can provide more effective information to distinguish the adaptability of different varieties.

[0052] (6) Calculate the adaptability evaluation index. The formula for the adaptability evaluation index is: ; This index takes into account the weight of each indicator (determined by the entropy weight method) and the standardized indicator value, and can comprehensively evaluate the adaptability of each corn variety. The larger the value, the higher the comprehensive evaluation of the adaptability of the variety. This index can provide a quantitative indicator for the selection and evaluation of corn varieties.

[0053] Optionally, the conditional variational autoencoder model can be constructed by a multi-layer perceptron MLP neural network, and both the encoder and the decoder are two fully connected layers.

[0054] Optionally, after completing the training of the conditional variational autoencoder model, the trained conditional variational autoencoder model based on the local enhanced graph network can be used to predict the data to be tested, obtain the prediction results, and build a graphical interface to display the variety adaptability prediction results.

[0055] In an embodiment of the present application, a second knowledge graph can be constructed based on a first knowledge graph of varieties and environments, with the node of the promoted planting area of ​​the variety as the central node, and based on the second knowledge graph, the conditional variational autoencoder model can be trained. By constructing a heterogeneous knowledge graph of varieties and environments, the complex interaction relationship between varieties and environments can be deeply explored. The knowledge graph is reconstructed with the node of the promoted planting area of ​​the variety as the central node, and a second knowledge graph with a larger number of nodes than the first knowledge graph is obtained, so that local enhancement of the knowledge graph can be achieved. By using local enhancement technology, neighborhood features can be generated through a conditional variational autoencoder model based on local structure and node features, thereby overcoming the problem that limited neighbor nodes are not conducive to learning node representation, and thus significantly improving the prediction accuracy of the conditional variational autoencoder model.

[0056] The above mainly introduces the solution provided by the embodiment of the present application from the perspective of the method. In order to realize the above functions, it includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiment disclosed herein, the embodiment of the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0057] It should be noted that the device in the embodiment of the present application includes a virtual device and a physical device. The virtual device can be a variety adaptability evaluation result prediction device, and the physical device can include an electronic device, a computer storage medium, and a computer program product.

[0058] The variety adaptability evaluation result prediction method provided in the embodiment of the present application can be performed by a variety adaptability evaluation result prediction device, or a control module for variety adaptability evaluation result prediction in the variety adaptability evaluation result prediction device. In the embodiment of the present application, the variety adaptability evaluation result prediction method is performed by a variety adaptability evaluation result prediction device as an example to illustrate the variety adaptability evaluation result prediction device provided in the embodiment of the present application.

[0059] It should be noted that the embodiment of the present application can divide the functional modules of the variety adaptability evaluation result prediction device according to the above method example. For example, each functional module can be divided corresponding to each function, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware or in the form of software functional modules. Optionally, the division of modules in the embodiment of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation.

[0060] like Figure 2 As shown, the embodiment of the present application provides a device 200 for predicting the results of variety adaptability evaluation. The device 200 for predicting the results of variety adaptability evaluation includes: an acquisition module 201 and a processing module 202. The acquisition module 201 can be used to acquire a first knowledge graph of varieties and environments, wherein the nodes of the first knowledge graph are varieties or planting areas, and the edges of the first knowledge graph include edges between two planting areas with environmental similarity higher than a first threshold, and edges between varieties and planting areas with planting relationships; the processing module 202 is used to reconstruct the knowledge graph based on the first knowledge graph with the node of the promoted planting area of ​​the variety as the central node, to obtain a second knowledge graph, wherein the number of nodes of the planting area of ​​the second knowledge graph is greater than the number of nodes of the planting area of ​​the first knowledge graph; based on the second knowledge graph, the conditional variational autoencoder model is trained, and a different latent vector generated by the conditional variational autoencoder model is sampled as an additional input in each training iteration to obtain a variety adaptability evaluation result prediction model; wherein the variety adaptability evaluation result prediction model is used to predict the adaptability of varieties under multiple environments.

[0061] Optionally, the acquisition module 201 is used to acquire planting data of the target variety under multiple environments, the planting data including whether the target variety is planted in the planting area, the meteorological data of the planting area and the soil data of the planting area; the processing module 202 is used to clean and standardize the planting data; determine the varieties and planting areas with planting relationships based on the processed planting data, and connect the varieties and planting areas with planting relationships to form edges; determine the environmental characteristics of the planting area based on the meteorological data and soil data of the processed planting data, determine the environmental similarity between two planting areas based on the environmental characteristics, and connect two planting areas with environmental similarity higher than the first threshold to form an edge.

[0062] Optionally, the processing module 202 is used to determine input data from the second knowledge graph, the input data including central node features and neighbor node features; input the input data into the encoder of the conditional variational autoencoder model to obtain a latent vector; input the latent vector and the central node features into the decoder of the conditional variational autoencoder model to obtain a target feature vector associated with the central node; wherein the target feature vector is used to indicate the adaptability evaluation index of the variety in different planting areas.

[0063] Optionally, the processing module 202 is used to map the input data to a latent variable space through an encoder of the conditional variational autoencoder model, and obtain the latent vector by sampling from the latent variable space.

[0064] In an embodiment of the present application, a second knowledge graph can be constructed based on a first knowledge graph of varieties and environments, with the node of the promoted planting area of ​​the variety as the central node, and based on the second knowledge graph, the conditional variational autoencoder model can be trained. By constructing a heterogeneous knowledge graph of varieties and environments, the complex interaction relationship between varieties and environments can be deeply explored. The knowledge graph is reconstructed with the node of the promoted planting area of ​​the variety as the central node, and a second knowledge graph with a larger number of nodes than the first knowledge graph is obtained, so that local enhancement of the knowledge graph can be achieved. By using local enhancement technology, neighborhood features can be generated through a conditional variational autoencoder model based on local structure and node features, thereby overcoming the problem that limited neighbor nodes are not conducive to learning node representation, and thus significantly improving the prediction accuracy of the conditional variational autoencoder model.

[0065] Figure 3 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 3As shown, the electronic device may include: a processor 310 , a communications interface 320 , a memory 330 and a communication bus 340 , wherein the processor 310 , the communications interface 320 , and the memory 330 communicate with each other via the communication bus 340 . The processor 310 can call the logic instructions in the memory 330 to execute a method for predicting variety adaptability evaluation results, which includes: obtaining a first knowledge graph of varieties and environments, wherein the nodes of the first knowledge graph are varieties or planting areas, and the edges of the first knowledge graph include edges between two planting areas with environmental similarity higher than a first threshold, and edges between varieties and planting areas with a planting relationship; based on the first knowledge graph, reconstructing the knowledge graph with the node of the promoted planting area of ​​the variety as the central node to obtain a second knowledge graph, wherein the number of nodes of the planting area of ​​the second knowledge graph is greater than the number of nodes of the planting area of ​​the first knowledge graph; based on the second knowledge graph, training a conditional variational autoencoder model, sampling a different latent vector generated by the conditional variational autoencoder model as an additional input in each training iteration, and obtaining a variety adaptability evaluation result prediction model; wherein the variety adaptability evaluation result prediction model is used to predict the adaptability of varieties under multiple environments.

[0066] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0067] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the variety adaptability evaluation result prediction method provided by the above-mentioned methods, the method including: obtaining a first knowledge graph of varieties and environments, the nodes of the first knowledge graph are varieties or planting areas, and the edges of the first knowledge graph include edges between two planting areas with an environmental similarity higher than a first threshold, and edges between varieties and planting areas with a planting relationship; based on the first knowledge graph, the knowledge graph is reconstructed with the node of the promoted planting area of ​​the variety as the central node to obtain a second knowledge graph, and the number of nodes of the planting area of ​​the second knowledge graph is greater than the number of nodes of the planting area of ​​the first knowledge graph; based on the second knowledge graph, a conditional variational autoencoder model is trained, and in each training iteration, a different latent vector generated by the conditional variational autoencoder model is sampled as an additional input to obtain a variety adaptability evaluation result prediction model; wherein the variety adaptability evaluation result prediction model is used to predict the adaptability of varieties under multiple environments.

[0068] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the variety adaptability evaluation result prediction method provided by the above-mentioned methods, the method comprising: obtaining a first knowledge graph of varieties and environments, the nodes of the first knowledge graph being varieties or planting areas, the edges of the first knowledge graph comprising edges between two planting areas having an environmental similarity higher than a first threshold, and edges between varieties having a planting relationship and planting areas; based on the first knowledge graph, reconstructing the knowledge graph with the node of the promoted planting area of ​​the variety as the central node to obtain a second knowledge graph, the number of nodes of the planting area of ​​the second knowledge graph being greater than the number of nodes of the planting area of ​​the first knowledge graph; based on the second knowledge graph, training a conditional variational autoencoder model, sampling a different latent vector generated by the conditional variational autoencoder model as an additional input in each training iteration to obtain a variety adaptability evaluation result prediction model; wherein the variety adaptability evaluation result prediction model is used to predict the adaptability of varieties under multiple environments.

[0069] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0070] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting variety adaptability evaluation results, characterized in that: include: Obtaining a first knowledge graph of varieties and environments, wherein the nodes of the first knowledge graph are varieties or planting areas, and the edges of the first knowledge graph include an edge between two planting areas whose environmental similarity is higher than a first threshold, and an edge between varieties and planting areas having a planting relationship; Based on the first knowledge graph, the knowledge graph is reconstructed with the node of the promoted planting area of ​​the variety as the central node to obtain a second knowledge graph, wherein the number of nodes of the planting area of ​​the second knowledge graph is greater than the number of nodes of the planting area of ​​the first knowledge graph; Based on the second knowledge graph, a conditional variational autoencoder model is trained, and a different latent vector generated by the conditional variational autoencoder model is sampled as an additional input in each training iteration to obtain a variety adaptability evaluation result prediction model; Among them, the variety adaptability evaluation result prediction model is used to predict the adaptability of the variety in multiple environments.

2. The method for predicting variety adaptability evaluation results according to claim 1, characterized in that: The step of obtaining the first knowledge graph of varieties and environments includes: Acquire planting data of a target variety under multiple environments, the planting data including whether the target variety is planted in a planting area, meteorological data of the planting area, and soil data of the planting area; Cleaning and standardizing the implantation data; Determine the varieties and planting areas with planting relationships based on the processed planting data, and connect the varieties and planting areas with planting relationships to form edges; The environmental characteristics of the planting area are determined based on the meteorological data and soil data of the processed planting data, and the environmental similarity between two planting areas is determined based on the environmental characteristics, and the two planting areas whose environmental similarity is higher than the first threshold are connected to form an edge.

3. The method for predicting variety adaptability evaluation results according to claim 1, characterized in that: The conditional variational autoencoder model is trained based on the second knowledge graph, and a different latent vector generated by the conditional variational autoencoder model is sampled as an additional input in each training iteration to obtain a variety adaptability evaluation result prediction model, including: Determining input data from the second knowledge graph, wherein the input data includes central node features and neighbor node features; Inputting the input data into the encoder of the conditional variational autoencoder model to obtain a latent vector; Inputting the latent vector and the central node feature into the decoder of the conditional variational autoencoder model to obtain a target feature vector associated with the central node; The target feature vector is used to indicate the adaptability evaluation index of the variety in different planting areas.

4. The method for predicting variety adaptability evaluation results according to claim 3, characterized in that: The step of inputting the input data into the encoder of the conditional variational autoencoder model to obtain a latent vector comprises: The input data is mapped to a latent variable space through the encoder of the conditional variational autoencoder model, and the latent vector is obtained by sampling from the latent variable space.

5. A device for predicting variety adaptability evaluation results, characterized in that: include: Acquisition module and processing module; The acquisition module is used to acquire a first knowledge graph of varieties and environments, wherein the nodes of the first knowledge graph are varieties or planting areas, and the edges of the first knowledge graph include an edge between two planting areas whose environmental similarity is higher than a first threshold, and an edge between a variety and a planting area having a planting relationship; The processing module is used to reconstruct the knowledge graph based on the first knowledge graph, taking the node of the promoted planting area of ​​the variety as the central node, to obtain a second knowledge graph, wherein the number of nodes of the planting area of ​​the second knowledge graph is greater than the number of nodes of the planting area of ​​the first knowledge graph; Based on the second knowledge graph, the conditional variational autoencoder model is trained, and a different latent vector generated by the conditional variational autoencoder model is sampled as an additional input in each training iteration to obtain a variety adaptability evaluation result prediction model; wherein the variety adaptability evaluation result prediction model is used to predict the adaptability of the variety in multiple environments.

6. The device for predicting variety adaptability evaluation results according to claim 5, characterized in that: The acquisition module is used to acquire planting data of the target variety under multiple environments, wherein the planting data includes whether the target variety is planted in a planting area, meteorological data of the planting area, and soil data of the planting area; The processing module is used to clean and standardize the planting data; Based on the processed planting data, varieties and planting areas with planting relationships are determined, and varieties and planting areas with planting relationships are connected to form edges; based on the meteorological data and soil data of the processed planting data, environmental characteristics of the planting areas are determined, and based on the environmental characteristics, the environmental similarity between two planting areas is determined, and two planting areas with environmental similarity higher than the first threshold are connected to form an edge.

7. The device for predicting variety adaptability evaluation results according to claim 5, characterized in that: The processing module is used to determine input data from the second knowledge graph, wherein the input data includes central node features and neighbor node features; input the input data into the encoder of the conditional variational autoencoder model to obtain a latent vector; input the latent vector and the central node features into the decoder of the conditional variational autoencoder model to obtain a target feature vector associated with the central node; wherein the target feature vector is used to indicate an adaptability evaluation index of a variety in different planting areas.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for predicting variety adaptability evaluation results as described in any one of claims 1 to 4 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for predicting variety adaptability evaluation results as described in any one of claims 1 to 4 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for predicting variety adaptability evaluation results as described in any one of claims 1 to 4 is implemented.

Images