Cabbage growth environment relationship reasoning method and device based on graph convolutional neural network

Abstract

The present invention provides a method and device for reasoning about the relationship between the growth environment of cabbage based on a graph convolutional neural network, and relates to the technical field of agricultural artificial intelligence. The method processes the growth index data of cabbage and the environmental factor data of cabbage to obtain the predicted yield of cabbage, the predicted quality of cabbage and the weighted environmental characteristics, and obtains the relationship between the predicted yield of cabbage and the environmental factors and the relationship between the predicted quality of cabbage and the environmental factors through a graph convolutional neural network. The method for reasoning about the relationship between the growth state of cabbage and the environmental factors based on a graph convolutional neural network provided by the present invention realizes the accurate and efficient reasoning about the relationship between the growth state of cabbage and the environmental factors under the condition of comprehensively considering the growth index data and the environmental factor data of cabbage, so as to provide suggestions for the planting management of cabbage based on this, optimize the growth environment of cabbage, and improve the yield and quality of cabbage.

Description

Technical Field

[0001] The present invention relates to the field of agricultural artificial intelligence technology, and in particular to a method and device for reasoning about cabbage growth environment relationships based on a graph convolutional neural network. Background Art

[0002] Cabbage is a biennial herbaceous plant of the genus Brassica in the family Cruciferae. It has a short growth cycle of about 80 to 120 days. It is not only a major edible vegetable and feed source, but also an important Chinese medicinal material. Cabbage leaves are used as medicine, which has anti-cancer effects and improves gastric ulcers, hyperglycemia and hyperlipidemia. Therefore, understanding and predicting how environmental factors affect cabbage growth and focusing on improving cabbage yield and quality are of great significance to the agricultural field and even the medical field.

[0003] In the analysis of the relationship between cabbage growth and environmental factors, existing technologies either rely on experience and manual judgment, lack of systematicity and scientificity, and have problems such as low efficiency, waste of resources and unstable yield; or analyze the interaction between cabbage growth indicators and a certain environmental factor based on deep learning neural network models (such as graph convolution network (GCN) model, etc.). However, due to the short growth cycle of cabbage, its various growth stages change rapidly and are sensitive to changes in multiple environmental factors such as light intensity, ambient temperature, soil moisture and precipitation. Therefore, the existing deep learning neural network model cannot accurately infer the relationship between the growth state of cabbage and environmental factors, so as to optimize its growth environment and improve the yield and quality of cabbage. Summary of the invention

[0004] The present invention provides a method and device for reasoning about the relationship between cabbage growth environment based on a graph convolutional neural network, so as to solve the above-mentioned problems existing in the prior art.

[0005] The present invention provides a cabbage growth environment relationship reasoning method based on a graph convolutional neural network, comprising the following steps: Collecting growth index data of cabbage and environmental factor data of cabbage; Inputting the growth index data of the cabbage into a long short-term memory network model to obtain the predicted yield of the cabbage and the predicted quality of the cabbage output by the long short-term memory network model, and inputting the environmental factor data of the cabbage into a convolutional neural network model to obtain the weighted environmental characteristics output by the convolutional neural network model; Inputting the predicted yield of the cabbage, the predicted quality of the cabbage, and the weighted environmental features into a graph convolutional neural network, and obtaining the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network; Among them, the graph convolutional neural network includes a graph convolutional network, a gated recurrent unit and a fully connected layer; the graph convolutional network is used to extract the spatial characteristics of cabbage growth; the gated recurrent unit is used to extract the temporal characteristics of cabbage growth; the fully connected layer is used to output the relationship between the predicted yield of the cabbage and environmental factors and the relationship between the predicted quality of the cabbage and environmental factors.

[0006] According to a method for reasoning about the relationship between the growth environment of cabbage based on a graph convolutional neural network provided by the present invention, the predicted yield of the cabbage, the predicted quality of the cabbage and the weighted environmental features are input into the graph convolutional neural network to obtain the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network, including: Inputting the growth index data of the cabbage and the environmental factor data of the cabbage into the graph convolutional network to extract the spatial characteristics of cabbage growth; Inputting the spatial characteristics of the cabbage growth, the growth index data of the cabbage and the environmental factor data of the cabbage into the gated recursive unit to extract the temporal characteristics of the cabbage growth; Performing feature splicing on the spatial features of the cabbage growth and the temporal features of the cabbage growth to obtain the global features of the cabbage growth; Based on the predicted yield of the cabbage, the predicted quality of the cabbage, the weighted environmental characteristics and the global characteristics of the cabbage growth, the fully connected layer is input to obtain the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the fully connected layer.

[0007] According to a method for reasoning about the relationship between cabbage growth and the environmental factors based on a graph convolutional neural network provided by the present invention, the graph convolutional network is constructed with growth indicators and environmental factors of cabbage growth as nodes of the graph, and with the correlation between the growth indicators and the environmental factors as edges of the graph.

[0008] According to a method for reasoning about the relationship between the growth environment of cabbage based on a graph convolutional neural network provided by the present invention, the environmental factor data of the cabbage is input into a convolutional neural network model to obtain weighted environmental features output by the convolutional neural network model, including: Inputting the environmental factor data of the cabbage into the convolutional layer of the convolutional neural network model to extract environmental features; Inputting the environmental features into the attention layer of the convolutional neural network model to obtain the attention weights of the environmental features; Based on the attention weights of the environmental features, the weighted environmental features are obtained through the fusion layer of the convolutional neural network model.

[0009] According to a cabbage growth environment relationship reasoning method based on a graph convolutional neural network provided by the present invention, the growth indicators include at least two of the growth stage, growth rate, plant height and leaf area index.

[0010] According to a cabbage growth environment relationship reasoning method based on a graph convolutional neural network provided by the present invention, the environmental factors include at least two of soil moisture, ambient temperature, precipitation, light intensity and pest and disease information.

[0011] The present invention also provides a cabbage growth environment relationship reasoning device based on a graph convolutional neural network, comprising the following modules: A collection module is used to collect growth index data of cabbage and environmental factor data of cabbage; A processing module, for inputting the growth index data of the cabbage into a long short-term memory network model to obtain the predicted yield of the cabbage and the predicted quality of the cabbage output by the long short-term memory network model, and for inputting the environmental factor data of the cabbage into a convolutional neural network model to obtain the weighted environmental characteristics output by the convolutional neural network model; A determination module, used for inputting the predicted yield of the cabbage, the predicted quality of the cabbage and the weighted environmental features into a graph convolutional neural network, and obtaining the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network; Among them, the graph convolutional neural network includes a graph convolutional network, a gated recurrent unit and a fully connected layer; the graph convolutional network is used to extract the spatial characteristics of cabbage growth; the gated recurrent unit is used to extract the temporal characteristics of cabbage growth; the fully connected layer is used to output the relationship between the predicted yield of the cabbage and environmental factors and the relationship between the predicted quality of the cabbage and environmental factors.

[0012] 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 computer program, the method for reasoning about the relationship between cabbage growth environment and based on a graph convolutional neural network as described above is implemented.

[0013] 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 reasoning about the relationship between cabbage growth environment and graph convolutional neural networks.

[0014] 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 cabbage growth environment relationship reasoning methods based on graph convolutional neural networks.

[0015] The present invention provides a method for reasoning about the relationship between the growth environment of cabbage based on a graph convolutional neural network. The method first collects growth index data of cabbage and environmental factor data of cabbage, then obtains the predicted yield of cabbage and the predicted quality of cabbage based on the growth index data of cabbage, and obtains weighted environmental characteristics based on the environmental factor data of cabbage. Finally, based on the predicted yield of cabbage, the predicted quality of cabbage and the weighted environmental characteristics, the relationship between the predicted yield of cabbage and environmental factors and the relationship between the predicted quality of cabbage and environmental factors are obtained through a graph convolutional neural network. This method achieves accurate and efficient reasoning about the relationship between the growth state of cabbage and environmental factors under comprehensive consideration of various growth data of cabbage, so as to provide suggestions for cabbage planting management based on this, optimize the growth environment of cabbage, and improve the yield and quality of cabbage. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] 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.

[0017] Figure 1 It is a flow chart of a method for reasoning about the relationship between cabbage growth environment based on a graph convolutional neural network provided by the present invention.

[0018] Figure 2 It is a schematic diagram of the reasoning framework of a cabbage growth and environmental factor relationship map in a cabbage growth environment relationship reasoning method based on a graph convolutional neural network provided by the present invention.

[0019] Figure 3 It is a schematic diagram of the graph convolutional neural network structure in a cabbage growth environment relationship reasoning method based on a graph convolutional neural network provided by the present invention.

[0020] Figure 4 It is a structural schematic diagram of a cabbage growth environment relationship reasoning device based on a graph convolutional neural network provided by the present invention.

[0021] Figure 5 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0022] In agricultural production, the growth process of cabbage is affected by environmental factors such as soil moisture, water and fertilizer supply, temperature, rainfall and light intensity. There is a complex correlation between these factors, which directly affects the growth, yield and quality of cabbage.

[0023] Therefore, understanding and predicting how environmental factors affect cabbage growth is of great significance for improving cabbage yield and quality. For example, in terms of growth cycle, cabbage usually has a short growth cycle of about 80 to 120 days. This characteristic can be used to analyze the impact of environmental factors at different growth stages on yield and quality; in terms of environmental adaptability, cabbage has strong adaptability to temperature, humidity and soil conditions, but performs best within a specific range. Based on these conditions, the best planting environment can be identified and the growth performance of cabbage under different environments can be predicted; in terms of disease and pest resistance, cabbage is susceptible to specific diseases and pests during growth (such as cabbage moth, black rot, etc.). Based on historical data, the frequency and impact of these diseases and pests can be analyzed, and the yield and quality of cabbage under different prevention and control measures can be predicted.

[0024] Traditional agricultural production management usually relies on experience and manual judgment, lacks systematicity and scientificity, and has problems such as low efficiency, waste of resources and unstable yield. With the development of smart agriculture, precision agricultural management systems are gaining more and more attention. Traditional statistical analysis methods cannot fully capture the complex environmental interaction effects, while graph neural networks, as an emerging deep learning technology, have unique advantages in processing graph structured data, providing new ideas and methods for solving the reasoning problem of the relationship between cabbage growth and environmental factors.

[0025] GCN is a type of deep learning model that specializes in processing graph-structured data. It can effectively capture the topological relationship between nodes and learn the representation of nodes. In the relationship analysis between cabbage growth and environmental factors, GCN can comprehensively consider the interaction between cabbage growth indicators and environmental factors, and represent this complex dependency through a graph structure. Existing research and applications based on GCN are mainly concentrated in the fields of social networks and recommendation systems, and their applications in the agricultural field are relatively small. In addition, the joint analysis of cabbage growth and environmental factors in the existing technology often ignores the importance of time series data, and lacks an effective mechanism to integrate spatiotemporal data, expert knowledge, and large-scale unlabeled data, resulting in poor interpretation of the model.

[0026] Therefore, a reasoning method and device for the cabbage growth-environmental factor relationship map based on a graph convolutional neural network was developed, which can combine spatiotemporal data, expert knowledge and large-scale unlabeled data to accurately predict and infer the growth status of cabbage. This has important theoretical significance and practical application value for improving cabbage production efficiency and quality and realizing intelligent and precise cabbage planting management.

[0027] In modern agricultural production, accurate prediction and management of cabbage growth status is the key to improving yield and quality. However, existing technologies have the following defects: 1. Lack of comprehensive consideration of environmental factors during cabbage growth: Existing technologies usually ignore the heterogeneity of environmental factors in spatial distribution, resulting in the inability of the model to accurately describe the impact of environmental factors on cabbage at different growth stages.

[0028] 2. Insufficient time dynamic processing: Cabbage growth is a process that evolves over time, but existing technologies often do not fully consider the important role of time series data in analysis and fail to use historical data to improve prediction accuracy.

[0029] 3. Insufficient integration of expert knowledge: Although some experts in the agricultural field have rich experience and knowledge, this knowledge is often not effectively integrated into the reasoning model, resulting in the model lacking guidance and support for agricultural production practices.

[0030] In view of the above defects, the purpose of the present invention is to provide a method and device for reasoning about the relationship between cabbage growth and environmental factors based on graph convolutional neural network, aiming to overcome the problems of insufficient data, complexity of environmental factors, limited model reasoning ability, etc. in traditional agricultural production management, and realize accurate reasoning and intelligent prediction of the complex relationship between cabbage growth and environmental factors. Specific measures include the following two aspects: (1) In-depth exploration of the impact of environmental factors on cabbage growth: The graph convolutional neural network-based method can better capture the complex correlation between environmental factors, achieve comprehensive analysis and in-depth exploration of cabbage growth, and provide a scientific basis for agricultural production management.

[0031] (2) Improve model reasoning ability and accuracy: Utilize the powerful reasoning ability of graph neural networks, combine environmental factor coupling models and expert knowledge, and achieve accurate reasoning and intelligent prediction of the relationship between cabbage growth and environmental factors, providing accurate and reliable support for agricultural production decision-making.

[0032] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] Combine the following Figures 1 to 5 The present invention describes a method and device for reasoning about the relationship between cabbage growth environment based on a graph convolutional neural network.

[0034] Figure 1 is a flow chart of a cabbage growth environment relationship reasoning method based on a graph convolutional neural network provided by the present invention, such as Figure 1 As shown, the method includes the following: Step 101: Collect growth index data of cabbage and environmental factor data of cabbage.

[0035] Obtain data on various growth indicators of cabbage and various environmental factors involved in the growth process of cabbage, such as soil moisture, temperature, rainfall, light intensity, etc.

[0036] Among them, growth indicator data refers to the growth indicator numbers with a timestamp sequence (the growth indicator data hereinafter referred to have the same meaning), and environmental factor data refers to the environmental factor data with a timestamp sequence (the environmental factor data hereinafter referred to have the same meaning).

[0037] By collecting growth index data and environmental factor data, the embodiment of the present invention can analyze the characteristics of various growth indicators changing over time during the growth of cabbage, as well as the effects of various environmental factors on cabbage growth at different times, thereby providing a scientific basis for subsequent prediction of cabbage yield and quality.

[0038] Optionally, the growth indicators include at least two of growth stage, growth rate, plant height and leaf area index.

[0039] Specifically, the growth index includes at least two of the growth stage, growth rate, plant height and leaf area index, wherein the growth stage includes germination stage, vegetative growth stage, and flowering stage.

[0040] For example, we can analyze the characteristics of cabbage at different growth stages, such as the germination period, the vegetative growth period, and the flowering period, and predict the yield and quality of cabbage through time series data (i.e., data with timestamps) at different growth stages.

[0041] For another example, feature extraction and data analysis can be performed based on time-stamped data such as growth rate, plant height, and leaf area index to predict the yield and quality of cabbage.

[0042] The embodiment of the present invention predicts the yield and quality of cabbage by analyzing at least two growth indicator data, and comprehensively analyzes multiple data to improve the accuracy of the prediction result.

[0043] Optionally, the environmental factors include at least two of soil moisture, ambient temperature, precipitation, light intensity and pest and disease information.

[0044] Specifically, the environmental factors include at least two of soil moisture, ambient temperature, precipitation, light intensity, and pest and disease information.

[0045] There are many environmental factors that affect the growth of cabbage, among which the more typical data include soil moisture, ambient temperature, precipitation, light intensity, and pest and disease information.

[0046] In the embodiment of the present invention, the cabbage growth index data and the cabbage environmental factor data are collected and preprocessed in the following manner.

[0047] First, determine the frequency and scope of data collection to ensure that all key environmental factors during the cabbage growth cycle are covered. Then, select appropriate sensors and data acquisition devices to collect data, such as soil moisture sensors, thermometers, light sensors, etc. Finally, use modern agricultural IoT technology to achieve real-time monitoring and remote access to data.

[0048] (1) Growth indicator data collection: Determine the frequency and scope of data collection to ensure that all key environmental factors during the cabbage growth cycle are covered. For example, through agricultural production records and existing databases, key indicators of cabbage growth, including growth stage, growth rate, plant height, and leaf area index, are collected to improve the accuracy of subsequent verification and inference results.

[0049] (2) Environmental factor data collection: Consider the impact of seasonal changes and extreme weather events on cabbage growth. For example, deploy weather stations and soil monitoring devices to collect data on environmental factors related to cabbage growth, including ambient air temperature, soil moisture, precipitation, light intensity, and information on pests and diseases.

[0050] (3) Data preprocessing: Clean and normalize the collected data to ensure data quality and consistency. Cleaning data can eliminate outliers and erroneous data, while normalization can unify data of different scales and ranges into the same standard range.

[0051] The embodiment of the present invention collects and preprocesses the data as described above, and according to the input requirements of the graph convolutional neural network, converts the processed data into graph structure data, thereby providing a basis for subsequent model training and reasoning analysis.

[0052] Step 102: input the growth index data of the cabbage into the long short-term memory network model to obtain the predicted yield of the cabbage and the predicted quality of the cabbage output by the long short-term memory network model, and input the environmental factor data of the cabbage into the convolutional neural network model to obtain the weighted environmental characteristics output by the convolutional neural network model.

[0053] Based on the above embodiment, after completing data collection and pre-processing, the growth index data of cabbage is input into the long short-term memory network model to obtain the predicted yield of cabbage and the predicted quality of cabbage output by the long short-term memory network model, and the environmental factor data of cabbage is input into the convolutional neural network model to obtain the weighted environmental characteristics output by the convolutional neural network model. Figure 2: is a schematic diagram of the reasoning framework of the cabbage growth and environmental factor relationship map in the cabbage growth environment relationship reasoning method based on graph convolutional neural network provided by the present invention, such as Figure 2 As shown, the crops in the picture refer to cabbage.

[0054] In the embodiment of the present invention, the long short-term memory network model is a crop growth model, which is trained by a semi-supervised learning method based on growth index data. The input of the crop growth model is the growth index data of cabbage, and by extracting the characteristics of the cabbage growth index changing over time, the growth state prediction of cabbage is output, that is, the predicted yield of cabbage and the predicted quality of cabbage.

[0055] The convolutional neural network model is an environmental factor coupling model, which is trained based on environmental factor data through semi-supervised learning. The input of the environmental factor coupling model is environmental factor data. By extracting environmental factor characteristics and according to the different effects of different environmental factors on cabbage growth, the potential impact score of each environmental factor on cabbage growth is obtained, so as to dynamically adjust its weight and output weighted environmental characteristics.

[0056] The embodiment of the present invention extracts the characteristics of cabbage growth indicators changing over time through a crop growth model to obtain the predicted yield of cabbage and the predicted quality of cabbage, extracts the characteristics of environmental factors through an environmental factor coupling model, and dynamically adjusts the weights of different environmental factors according to the different effects of different environmental factors on cabbage growth to obtain weighted environmental characteristics, thereby improving the accuracy of subsequent reasoning on the relationship between the predicted yield of cabbage and environmental factors and the relationship between the predicted quality of cabbage and environmental factors based on the predicted yield of cabbage and the predicted quality of cabbage and the weighted environmental characteristics.

[0057] Optionally, the step of inputting the environmental factor data of the cabbage into a convolutional neural network model to obtain weighted environmental features output by the convolutional neural network model comprises: Inputting the environmental factor data of the cabbage into the convolutional layer of the convolutional neural network model to extract environmental features; Inputting the environmental features into the attention layer of the convolutional neural network model to obtain the attention weights of the environmental features; Based on the attention weights of the environmental features, the weighted environmental features are obtained through the fusion layer of the convolutional neural network model.

[0058] Specifically, in the embodiment of the present invention, the convolutional neural network model is an environmental factor coupling model, which mainly includes a convolution layer, an attention layer, and a fusion layer. The weighted environmental features are obtained by the following steps: First, the environmental factor data of cabbage was input into the convolutional layer of the environmental factor coupling model to extract environmental features; Then, the environmental features are input into the attention layer of the environmental element coupling model to obtain the attention weights of the environmental features; Finally, based on the attention weights of environmental features, weighted environmental features are obtained through the fusion layer of the environmental element coupling model.

[0059] In the embodiment of the present invention, the attention layer (i.e. Figure 2 The attention module in is built based on the attention mechanism, which is used to identify the environmental features that have a greater impact on the growth of cabbage, take it as the key factor, and specifically improve the attention of the environmental factor coupling model to the key factor, that is, increase the attention weight of the key factor. Through the attention mechanism, the growth state of cabbage can be predicted more accurately, and the accuracy and efficiency of reasoning can be improved. The expression of the attention mechanism is as follows: In the formula, Representation Node i In the l +1 layer feature representation. Through the attention mechanism, the feature representation of the current node is fused with the feature representation of the neighboring nodes to update the node i δ(⋅) represents the activation function, which is used to increase nonlinearity and make the model more expressive. Attention weight, representing the neighborhood nodes j For Node i The importance of the node determines j The features of the update node i The contribution size of the feature representation. The weight matrix is ​​used to l Node characteristics of the layer Linearly transform to a new feature space to enhance the expressive power of the model. node j In the l The feature representation of the layer. A learnable vector that transforms nodes i and nodes j The feature combination vector is mapped to a scalar, which determines the attention weight The calculation method of . Indicates that the node i and nodes j In the l Layers pass through the weight matrix The transformed feature concatenation. LeakyReLU is an activation function used to convert the concatenated feature vector into a scalar for further calculation of the attention weight. Normalization term, for each node in the neighborhoodk The attention weights are softmax normalized so that the sum of the attention weights in the neighborhood is 1.

[0060] The attention layer in the environmental factor coupling model is constructed based on the attention mechanism, which enhances the sensitivity of the environmental factor coupling model to the environmental features that have a greater impact on cabbage growth, so that the attention weights of the environmental features can be determined and the weighted environmental features can be obtained.

[0061] The embodiment of the present invention constructs an environmental factor coupling model by introducing an attention mechanism to obtain weighted environmental features, so that the model can focus on key environmental factors that have a greater impact on cabbage growth, thereby improving the subsequent reasoning accuracy and efficiency of the relationship between the predicted yield of cabbage and environmental factors and the relationship between the predicted quality of cabbage and environmental factors.

[0062] Step 103: input the predicted yield of the cabbage, the predicted quality of the cabbage, and the weighted environmental characteristics into a graph convolutional neural network to obtain the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network.

[0063] Among them, the graph convolutional neural network includes a graph convolutional network, a gated recurrent unit and a fully connected layer; the graph convolutional network is used to extract the spatial characteristics of cabbage growth; the gated recurrent unit is used to extract the temporal characteristics of cabbage growth; the fully connected layer is used to output the relationship between the predicted yield of the cabbage and environmental factors and the relationship between the predicted quality of the cabbage and environmental factors.

[0064] After obtaining the predicted yield and quality of cabbage through the crop growth model and obtaining weighted environmental characteristics through the environmental factor coupling model, these are used as the input of the graph convolutional neural network to obtain the relationship between the predicted yield of cabbage and environmental factors, as well as the relationship between the predicted quality of cabbage and environmental factors output by the graph convolutional neural network.

[0065] In an embodiment of the present invention, a semi-supervised learning method is used to train a graph convolutional neural network to adapt to different application scenarios.

[0066] First, some labeled growth indicator data, environmental factor data and expert knowledge, as well as a large amount of unlabeled growth indicator data, environmental factor data and expert knowledge are used to train the model, so as to improve the generalization ability and adaptability of the model.

[0067] Secondly, strategies such as early stopping and regularization are applied in graph convolutional neural networks to prevent overfitting.

[0068] Finally, the network hierarchy of the graph convolutional neural network is designed, including the number of convolutional layers, the number of hidden units, and the activation function LeakyReLU. The cross entropy loss function is used for classification problems, and the Adam optimization algorithm is selected for gradient update to improve the prediction ability of the graph convolutional neural network.

[0069] In an embodiment of the present invention, not only the growth index data and environmental factor data of cabbage are used as training data of the graph convolutional neural network, but also the expert knowledge is integrated into the reasoning process of the graph convolutional neural network through the rule engine, thereby improving the interpretability, reliability and robustness of the graph convolutional neural network. Among them, the rule engine can embed expert knowledge into the reasoning process in the form of rules. For example, based on the empirical rules of agricultural experts, the degree of influence of specific environmental factors on the growth of cabbage is determined. These rules can help the model better understand the growth environment of cabbage, and adjust and correct key factors during the reasoning process, thereby improving the credibility and accuracy of the reasoning results.

[0070] The present invention provides a method for reasoning about the relationship between the growth environment of cabbage based on a graph convolutional neural network. The method first collects growth index data of cabbage and environmental factor data of cabbage, then obtains the predicted yield of cabbage and the predicted quality of cabbage based on the growth index data of cabbage, and obtains weighted environmental characteristics based on the environmental factor data of cabbage. Finally, based on the predicted yield of cabbage, the predicted quality of cabbage and the weighted environmental characteristics, the relationship between the predicted yield of cabbage and environmental factors and the relationship between the predicted quality of cabbage and environmental factors are obtained through a graph convolutional neural network. The method achieves accurate and efficient reasoning about the relationship between the growth state of cabbage and environmental factors under comprehensive consideration of the growth index data and environmental factor data of cabbage, so that suggestions can be provided for the planting management of cabbage based on this, for example, adjusting the irrigation amount, fertilization strategy and pest and disease control measures, so as to optimize the growth environment of cabbage and improve the yield and quality of cabbage.

[0071] Optionally, the predicted yield of the cabbage, the predicted quality of the cabbage, and the weighted environmental features are input into a graph convolutional neural network to obtain the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network, including: Inputting the growth index data of the cabbage and the environmental factor data of the cabbage into the graph convolutional network to extract the spatial characteristics of cabbage growth; Inputting the spatial characteristics of the cabbage growth, the growth index data of the cabbage and the environmental factor data of the cabbage into the gated recursive unit to extract the temporal characteristics of the cabbage growth; Performing feature splicing on the spatial features of the cabbage growth and the temporal features of the cabbage growth to obtain the global features of the cabbage growth; Based on the predicted yield of the cabbage, the predicted quality of the cabbage, the weighted environmental characteristics and the global characteristics of the cabbage growth, the fully connected layer is input to obtain the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the fully connected layer.

[0072] Specifically, the graph convolutional neural network includes a graph convolutional network, a gated recurrent unit and a fully connected layer; the graph convolutional network is used to extract the spatial characteristics of cabbage growth; the gated recurrent unit is used to extract the temporal characteristics of cabbage growth; the fully connected layer is used to output the relationship between the predicted yield of cabbage and environmental factors, as well as the relationship between the predicted quality of cabbage and environmental factors.

[0073] Figure 3 is a schematic diagram of the structure of a graph convolutional neural network in a method for reasoning about the relationship between the growth environment of cabbage based on a graph convolutional neural network provided by an embodiment of the present invention, such as Figure 3 The crop growth data in the figure are the growth index data of cabbage, and the environmental coupling data are the environmental factor data of cabbage.

[0074] First, using historical time series data (i.e., cabbage growth index data and cabbage environmental factor data) as input, the topological structure of the graph is obtained through GCN, and then the spatial characteristics of cabbage growth are obtained. In GCN, the graph structure is represented as G = (V, E), where V represents the set of nodes and E represents the connection relationship between nodes. The expression of GCN is: In the formula, Indicates The node representation matrix of the layer Indicates The nodes of a layer represent matrices, is the adjacency matrix With the identity matrix The sum of represents the connection relationship between nodes. Represents the degree matrix of the graph, which represents the degree of each node (that is, the number of adjacent nodes of each node). Indicates The weight matrix of the layer, is the activation function. Through this process, a new feature matrix with spatial information (i.e., the spatial features of cabbage growth) is obtained for subsequent reasoning tasks.

[0075] Secondly, the obtained new feature matrix of spatial information, the growth index data of cabbage and the environmental factor data of the cabbage are input into the gated recursive unit, and dynamic changes are obtained through information transmission between units, so as to obtain the time characteristics of cabbage growth.

[0076] Thirdly, the spatial characteristics of cabbage growth and the temporal characteristics of cabbage growth are spliced ​​to obtain the global characteristics of cabbage growth.

[0077] Among them, feature splicing is done through the graph reasoning module in the graph convolutional neural network (see Figure 2 ). The graph reasoning module can use the characteristics of the graph structure to reason and analyze the relationship between cabbage growth and environmental factors from a global perspective. Through information transmission and aggregation, each node can obtain relevant information from surrounding nodes. This process can be achieved through multiple rounds of graph convolution operations. The specific formula is as follows: In the formula, Indicates l The node feature matrix of a layer contains the feature representation of each node in the layer. Represents the adjacency matrix, which represents the connection relationship between nodes in the graph. is the update function of the graph convolution operation, indicating the l Node characteristics of the layer In the adjacency matrix The process of aggregation and transformation is carried out under the guidance of.

[0078] The relationship between cabbage growth and environmental factors is modeled through the aggregated features. This process can be achieved by splicing the spatial features of cabbage growth extracted by GCN and the temporal features of cabbage growth extracted by gated recurrent units. The feature representation of each node in the graph is spliced ​​or weighted summed to obtain the global feature: In the formula, Represents global features, Indicates the last layer (i.e. L The feature matrix of the nodes in the layer, Represents the splicing operation. Through this process, the global characteristics of cabbage growth can be obtained, that is, a global understanding of the cabbage growth status (including yield and quality) and environmental factors.

[0079] Finally, based on the predicted yield of cabbage, the predicted quality of cabbage, the weighted environmental characteristics and the global characteristics of cabbage growth, the fully connected layer in the graph convolutional neural network is input to obtain the relationship between the predicted yield of cabbage and environmental factors and the relationship between the predicted quality of cabbage and environmental factors output by the fully connected layer. Based on this, the planting management of cabbage is guided. For example, based on a certain yield target or quality target, the growth environment of cabbage is adjusted and controlled based on the relationship between the predicted yield of cabbage and environmental factors and the relationship between the predicted quality of cabbage and environmental factors, including adjusting the irrigation amount, fertilization strategy, and pest and disease control measures, so as to achieve the yield target or quality target of cabbage.

[0080] In the embodiment of the present invention, the prediction result is normalized using the softmax function to obtain the probability distribution of each category. The expression of the softmax function is: In the formula, Represents the prediction result, Represents global features, represents the weight of the fully connected layer, Represents the bias parameters of the fully connected layer.

[0081] The embodiment of the present invention is based on a graph convolutional neural network, which can accurately analyze the influence of various environmental factors on the growth of cabbage, improve the accuracy of yield prediction and quality prediction of cabbage, and then optimize planting strategies and irrigation plans and other management programs through the relationship between the predicted yield of cabbage and environmental factors and the relationship between the predicted quality of cabbage and environmental factors, thereby improving the yield and quality of cabbage, reducing production costs, and promoting the modernization and intelligent development of cabbage planting.

[0082] Optionally, the graph convolutional network is constructed using growth indicators and environmental factors of cabbage growth as nodes of the graph, and using the correlation between the growth indicators and the environmental factors as edges of the graph.

[0083] Specifically, according to various factors involved in the growth process of cabbage, growth indicators and environmental factors of cabbage growth are used as nodes of the graph. Growth indicators include at least two of growth stage, growth rate, plant height and leaf area index. Growth stage includes germination period, vegetative growth period, and flowering period. Environmental factors include at least two of soil moisture, ambient temperature, precipitation, light intensity and pest information.

[0084] The edges represent the associations between these nodes, that is, the associations between growth indicators and environmental factors, for example, the relationship that the growth rate of cabbage is affected by environmental factors such as soil moisture or light intensity.

[0085] The embodiment of the present invention constructs the graph structure of the graph convolutional network based on the complex correlation between growth indicators and environmental factors of cabbage growth, and makes full use of various spatiotemporal data of cabbage growth, such as growth indicator data and environmental factor data of cabbage growth, thereby improving the reasoning, explanation, reliability and accuracy of the graph convolutional neural network for the relationship between the predicted yield of cabbage and environmental factors and the relationship between the predicted quality of cabbage and environmental factors.

[0086] The cabbage growth environment relationship reasoning device based on graph convolutional neural network provided by the present invention is described below. The cabbage growth environment relationship reasoning device based on graph convolutional neural network described below and the cabbage growth environment relationship reasoning method based on graph convolutional neural network described above can be referenced to each other.

[0087] Based on any of the above embodiments, Figure 4 This is a schematic diagram of the structure of a cabbage growth environment relationship reasoning device based on a graph convolutional neural network provided in this paper. Figure 4 The embodiment of the present invention provides a cabbage growth environment relationship reasoning device based on a graph convolutional neural network, comprising a collection module 401, a processing module 402 and a determination module 403, wherein: The collection module 401 is used to collect the growth index data of cabbage and the environmental factor data of cabbage; The processing module 402 is used to input the growth index data of the cabbage into the long short-term memory network model to obtain the predicted yield of the cabbage and the predicted quality of the cabbage output by the long short-term memory network model, and input the environmental factor data of the cabbage into the convolutional neural network model to obtain the weighted environmental characteristics output by the convolutional neural network model; The determination module 403 is used to input the predicted yield of the cabbage, the predicted quality of the cabbage and the weighted environmental features into a graph convolutional neural network to obtain the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network; Among them, the graph convolutional neural network includes a graph convolutional network, a gated recurrent unit and a fully connected layer; the graph convolutional network is used to extract the spatial characteristics of cabbage growth; the gated recurrent unit is used to extract the temporal characteristics of cabbage growth; the fully connected layer is used to output the relationship between the predicted yield of the cabbage and environmental factors and the relationship between the predicted quality of the cabbage and environmental factors.

[0088] The present invention provides a cabbage growth environment relationship reasoning device based on a graph convolutional neural network. The device first collects cabbage growth index data and cabbage environmental factor data, then obtains the predicted yield and predicted quality of cabbage based on the growth index data of cabbage, and obtains weighted environmental characteristics based on the environmental factor data of cabbage. Finally, based on the predicted yield of cabbage, the predicted quality of cabbage and the weighted environmental characteristics, the relationship between the predicted yield of cabbage and environmental factors and the relationship between the predicted quality of cabbage and environmental factors are obtained through a graph convolutional neural network. The device achieves accurate and efficient reasoning of the relationship between the growth state of cabbage and environmental factors under comprehensive consideration of the growth index data and environmental factor data of cabbage, so that suggestions can be provided for cabbage planting management based on this, for example, adjusting the irrigation amount, fertilization strategy and pest and disease control measures, so as to optimize the growth environment of cabbage and improve the yield and quality of cabbage.

[0089] Figure 5 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530 and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 540. The processor 510 may call the logic instructions in the memory 530 to execute the cabbage growth environment relationship reasoning method based on the graph convolutional neural network, and the method includes: Collecting growth index data of cabbage and environmental factor data of cabbage; Inputting the growth index data of the cabbage into a long short-term memory network model to obtain the predicted yield of the cabbage and the predicted quality of the cabbage output by the long short-term memory network model, and inputting the environmental factor data of the cabbage into a convolutional neural network model to obtain the weighted environmental characteristics output by the convolutional neural network model; Inputting the predicted yield of the cabbage, the predicted quality of the cabbage, and the weighted environmental features into a graph convolutional neural network, and obtaining the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network; Among them, the graph convolutional neural network includes a graph convolutional network, a gated recurrent unit and a fully connected layer; the graph convolutional network is used to extract the spatial characteristics of cabbage growth; the gated recurrent unit is used to extract the temporal characteristics of cabbage growth; the fully connected layer is used to output the relationship between the predicted yield of the cabbage and environmental factors and the relationship between the predicted quality of the cabbage and environmental factors.

[0090] In addition, the logic instructions in the above-mentioned memory 530 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 this 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 to enable 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.

[0091] On the other hand, the present invention also provides a computer program product, the computer program product includes a computer program, the computer program can be stored on a non-transitory computer-readable storage medium, and when the computer program is executed by a processor, the computer can execute the cabbage growth environment relationship reasoning method based on a graph convolutional neural network provided by the above methods, the method comprising: Collecting growth index data of cabbage and environmental factor data of cabbage; Inputting the growth index data of the cabbage into a long short-term memory network model to obtain the predicted yield of the cabbage and the predicted quality of the cabbage output by the long short-term memory network model, and inputting the environmental factor data of the cabbage into a convolutional neural network model to obtain the weighted environmental characteristics output by the convolutional neural network model; Inputting the predicted yield of the cabbage, the predicted quality of the cabbage, and the weighted environmental features into a graph convolutional neural network, and obtaining the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network; Among them, the graph convolutional neural network includes a graph convolutional network, a gated recurrent unit and a fully connected layer; the graph convolutional network is used to extract the spatial characteristics of cabbage growth; the gated recurrent unit is used to extract the temporal characteristics of cabbage growth; the fully connected layer is used to output the relationship between the predicted yield of the cabbage and environmental factors and the relationship between the predicted quality of the cabbage and environmental factors.

[0092] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to execute the cabbage growth environment relationship reasoning method based on a graph convolutional neural network provided by the above methods, the method comprising: Collecting growth index data of cabbage and environmental factor data of cabbage; Inputting the growth index data of the cabbage into a long short-term memory network model to obtain the predicted yield of the cabbage and the predicted quality of the cabbage output by the long short-term memory network model, and inputting the environmental factor data of the cabbage into a convolutional neural network model to obtain the weighted environmental characteristics output by the convolutional neural network model; Inputting the predicted yield of the cabbage, the predicted quality of the cabbage, and the weighted environmental features into a graph convolutional neural network, and obtaining the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network; Among them, the graph convolutional neural network includes a graph convolutional network, a gated recurrent unit and a fully connected layer; the graph convolutional network is used to extract the spatial characteristics of cabbage growth; the gated recurrent unit is used to extract the temporal characteristics of cabbage growth; the fully connected layer is used to output the relationship between the predicted yield of the cabbage and environmental factors and the relationship between the predicted quality of the cabbage and environmental factors.

[0093] 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.

[0094] 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.

[0095] It should be noted that, in this article, the term "comprises", "includes" or any other variant thereof is 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 "including 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 method and device in the embodiment 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.

[0096] "Determine B based on A" in the embodiments of the present application means that the factor A should be considered when determining B. It is not limited to "B can be determined based on A alone", but should also include: "Determine B based on A and C", "Determine B based on A, C and E", "Determine C based on A, and further determine B based on C", etc. In addition, it can also include taking A as a condition for determining B, for example, "When A meets the first condition, use the first method to determine B"; for another example, "When A meets the second condition, determine B", etc.; for another example, "When A meets the third condition, determine B based on the first parameter", etc. Of course, it can also be a condition that takes A as a factor for determining B, for example, "When A meets the first condition, use the first method to determine C, and further determine B based on C", etc.

[0097] In the present invention, the term "plurality" refers to two or more than two, and other quantifiers are similar to them.

[0098] 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 cabbage growth environment relationship reasoning method based on graph convolutional neural network, characterized in that: include: Collecting growth index data and environmental factor data of cabbage; Inputting the growth index data of the cabbage into a long short-term memory network model to obtain the predicted yield of the cabbage and the predicted quality of the cabbage output by the long short-term memory network model, and inputting the environmental factor data of the cabbage into a convolutional neural network model to obtain the weighted environmental characteristics output by the convolutional neural network model; Inputting the predicted yield of the cabbage, the predicted quality of the cabbage, and the weighted environmental features into a graph convolutional neural network, and obtaining the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network; Among them, the graph convolutional neural network includes a graph convolutional network, a gated recurrent unit and a fully connected layer; the graph convolutional network is used to extract the spatial characteristics of cabbage growth; the gated recurrent unit is used to extract the temporal characteristics of cabbage growth; the fully connected layer is used to output the relationship between the predicted yield of the cabbage and environmental factors and the relationship between the predicted quality of the cabbage and environmental factors.

2. The cabbage growth environment relationship reasoning method based on graph convolutional neural network according to claim 1, characterized in that: The predicted yield of the cabbage, the predicted quality of the cabbage, and the weighted environmental features are input into a graph convolutional neural network to obtain the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network, including: Inputting the growth index data of the cabbage and the environmental factor data of the cabbage into the graph convolutional network to extract the spatial characteristics of cabbage growth; Inputting the spatial characteristics of the cabbage growth, the growth index data of the cabbage and the environmental factor data of the cabbage into the gated recursive unit to extract the temporal characteristics of the cabbage growth; Performing feature splicing on the spatial features of the cabbage growth and the temporal features of the cabbage growth to obtain the global features of the cabbage growth; Based on the predicted yield of the cabbage, the predicted quality of the cabbage, the weighted environmental characteristics and the global characteristics of the cabbage growth, the fully connected layer is input to obtain the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the fully connected layer.

3. The cabbage growth environment relationship reasoning method based on graph convolutional neural network according to claim 2 is characterized in that: The graph convolutional network is constructed by using growth indicators and environmental factors of cabbage growth as nodes of the graph, and using the correlation between the growth indicators and the environmental factors as edges of the graph.

4. The cabbage growth environment relationship reasoning method based on graph convolutional neural network according to claim 1, characterized in that: The step of inputting the environmental factor data of the cabbage into a convolutional neural network model to obtain weighted environmental features output by the convolutional neural network model comprises: Inputting the environmental factor data of the cabbage into the convolutional layer of the convolutional neural network model to extract environmental features; Inputting the environmental features into the attention layer of the convolutional neural network model to obtain the attention weights of the environmental features; Based on the attention weights of the environmental features, the weighted environmental features are obtained through the fusion layer of the convolutional neural network model.

5. The cabbage growth environment relationship reasoning method based on graph convolutional neural network according to any one of claims 1 or 3, characterized in that: The growth indicators include at least two of growth stage, growth rate, plant height and leaf area index.

6. The cabbage growth environment relationship reasoning method based on graph convolutional neural network according to any one of claims 1 or 3, characterized in that: The environmental factors include at least two of soil moisture, ambient temperature, precipitation, light intensity, and pest and disease information.

7. A cabbage growth environment relationship reasoning device based on graph convolutional neural network, characterized in that: include: A collection module is used to collect growth index data of cabbage and environmental factor data of cabbage; A processing module, for inputting the growth index data of the cabbage into a long short-term memory network model to obtain the predicted yield of the cabbage and the predicted quality of the cabbage output by the long short-term memory network model, and for inputting the environmental factor data of the cabbage into a convolutional neural network model to obtain the weighted environmental characteristics output by the convolutional neural network model; A determination module, used for inputting the predicted yield of the cabbage, the predicted quality of the cabbage and the weighted environmental features into a graph convolutional neural network, and obtaining the relationship between the predicted yield of the cabbage and the environmental factors and the relationship between the predicted quality of the cabbage and the environmental factors output by the graph convolutional neural network; Among them, the graph convolutional neural network includes a graph convolutional network, a gated recurrent unit and a fully connected layer; the graph convolutional network is used to extract the spatial characteristics of cabbage growth; the gated recurrent unit is used to extract the temporal characteristics of cabbage growth; the fully connected layer is used to output the relationship between the predicted yield of the cabbage and environmental factors and the relationship between the predicted quality of the cabbage and environmental factors.

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 cabbage growth environment relationship reasoning method based on graph convolutional neural network as described in any one of claims 1 to 6 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 reasoning about the relationship between the growth environment of cabbage based on a graph convolutional neural network as described in any one of claims 1 to 6 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 reasoning about the relationship between the growth environment of cabbage based on a graph convolutional neural network as described in any one of claims 1 to 6 is implemented.

Images