Plant layout model establishment method, acquisition method, system and storage medium

Abstract

This invention discloses a method for establishing, acquiring, system, and storage medium for a plant layout model, relating to the field of natural language processing for plant layout. The method comprises: obtaining environmental requirement data for plant cultivation and a data source containing professional plant cultivation knowledge; performing entity labeling on the data source using a BiLSTM-CRF model and an attention mechanism to obtain a standardized first labeled sequence; performing entity disambiguation on the first labeled sequence to obtain a second labeled sequence; performing knowledge reasoning on the second labeled sequence to obtain a knowledge graph; and obtaining a plant layout algorithm model based on the environmental requirement data and the knowledge graph. The resulting plant layout algorithm model improves the relevance of entities within their context, thereby increasing the efficiency and accuracy of constructing plant layout images.

Description

Technical Field

[0001] The present invention relates to the field of plant layout natural language processing, and in particular to a plant layout model establishment method, acquisition method, system and storage medium. Background Art

[0002] Balcony farming is a new type of agricultural cultivation that has developed in recent years and holds great promise. As small plots of land commonly found in cities, balconies can be used to grow plants and reduce air pollution. The widespread application of modern intelligent control technologies has made automated management of balcony farming a reality.

[0003] A related art system provides a plant layout system that allows users to input data about desired plants through a human-computer interaction interface to generate a plant layout image. However, the connectivity between entities in this system is poor, resulting in weak contextual connections between entities and reducing the accuracy of the plant layout image generated by the system. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention provides a method for establishing, acquiring, system, and storage medium for a plant layout model, thereby improving the relevance of entities within their context and thereby enhancing the efficiency and accuracy of constructing plant layout images.

[0005] In a first aspect, an embodiment of the present invention provides a method for establishing a plant layout model, comprising:

[0006] Obtain data on the environmental requirements of plant cultivation and data sources with professional plant cultivation knowledge;

[0007] Performing entity labeling on the data source through a BiLSTM-CRF model and an attention mechanism to obtain a normalized first labeled sequence;

[0008] Performing entity disambiguation on the first labeled sequence to obtain a second labeled sequence;

[0009] Performing knowledge reasoning on the second labeled sequence to obtain a knowledge graph;

[0010] A plant layout algorithm model is obtained based on the environmental demand data and the knowledge graph.

[0011] One or more technical solutions provided in the embodiments of the present invention have at least the following beneficial effects: obtaining environmental requirement data for plant cultivation and a data source containing professional plant cultivation knowledge; performing entity extraction and entity labeling on the data source using a BiLSTM-CRF model and an attention mechanism to obtain a standardized first labeled sequence; then performing entity disambiguation on the labeled sequence to obtain a second labeled sequence with new entity relationships; performing knowledge reasoning on the second labeled sequence to obtain a knowledge graph; and obtaining a plant layout algorithm model based on the environmental requirement data and the knowledge graph. This configuration improves the relevance of entities within the context of the plant layout, thereby improving the efficiency and accuracy of constructing plant layout images.

[0012] According to some embodiments of the first aspect of the present invention, the BiLSTM-CRF model includes a BiLSTM model and a conditional random field model, and the entity labeling processing of the data source using the BiLSTM-CRF model and the attention mechanism to obtain a normalized first labeled sequence includes:

[0013] According to the BiLSTM model and attention mechanism, a joint feature vector sequence is obtained;

[0014] The joint feature vector sequence is used as the input of the conditional random field model, so that the conditional random field model performs entity labeling processing on the data source to obtain a normalized first labeled sequence.

[0015] According to some embodiments of the first aspect of the present invention, obtaining a joint feature vector sequence according to the BiLSTM model and the attention mechanism includes:

[0016] Obtaining a global feature vector corresponding to the data source according to the BiLSTM model;

[0017] Using the global feature vector as the input of the attention mechanism to obtain a local feature vector;

[0018] The global feature vector is combined with the local feature vector to obtain a joint feature vector sequence.

[0019] According to some embodiments of the first aspect of the present invention, performing entity disambiguation on the first labeled sequence to obtain a second labeled sequence includes:

[0020] determining at least one ambiguous entity set from the first labeled sequence;

[0021] Clustering the ambiguous entity set using an aggregated hierarchical clustering algorithm;

[0022] A second labeled sequence is obtained according to the clustering result and the first labeled sequence.

[0023] According to some embodiments of the first aspect of the present invention, performing knowledge reasoning on the second labeled sequence to obtain a knowledge graph includes:

[0024] The second labeled sequence is subjected to knowledge reasoning through a neural tensor network model to obtain a knowledge graph.

[0025] According to some embodiments of the first aspect of the present invention, the environmental demand data includes illuminance data, temperature data, and humidity data.

[0026] In a second aspect, an embodiment of the present invention further provides a method for obtaining plant layout, comprising:

[0027] Obtain light distribution images and plant datasets corresponding to the user's plant planting requirements;

[0028] According to the illumination distribution image, the plant data set is input into the plant layout algorithm model to obtain a plant layout image.

[0029] According to some embodiments of the second aspect of the present invention, light data collected by multiple sensors and an environmental distribution structure corresponding to plants planted by a user are obtained;

[0030] The illumination distribution image is obtained according to the environmental distribution structure and the illumination data.

[0031] According to a third aspect of the present invention, a plant layout system includes 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 following is achieved:

[0032] The method for establishing the plant layout model as described in the first aspect above;

[0033] or,

[0034] The method for obtaining plant layout as described in the second aspect above.

[0035] According to a computer-readable storage medium of an embodiment of the fourth aspect of the present invention, the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to:

[0036] Executing the method for establishing the plant layout model described in the first aspect above;

[0037] or,

[0038] Execute the method for acquiring plant layout described in the second aspect above.

[0039] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings are used to provide a further understanding of the technical solution of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the technical solution of the present invention and do not constitute a limitation to the technical solution of the invention.

[0041] Figure 1 It is a flowchart of a method for establishing a plant layout model provided by one embodiment of the present invention;

[0042] Figure 2 is a flow chart of a method for establishing a plant layout model provided by another embodiment of the present invention;

[0043] Figure 3 Schematic diagram of the structure of the BILSTM-CRF model provided by an embodiment of the present invention;

[0044] Figure 4 is a schematic diagram of a process for obtaining a first marked sequence according to an embodiment of the present invention;

[0045] Figure 5 Schematic diagram of the structure of the LSTM model provided by an embodiment of the present invention;

[0046] Figure 6 This is a schematic diagram of the structure of the Attention-BiLSTM model provided by an embodiment of the present invention;

[0047] Figure 7 1 is a schematic diagram of a process for obtaining a joint feature vector sequence according to an embodiment of the present invention;

[0048] Figure 8 is a schematic diagram of a process for obtaining a second labeled sequence according to an embodiment of the present invention;

[0049] Figure 9 This is a schematic diagram of the process of obtaining a knowledge graph provided by an embodiment of the present invention;

[0050] Figure 10 is a schematic diagram of a process for obtaining a plant layout image provided by an embodiment of the present invention;

[0051] Figure 11 It is a schematic diagram of the process of obtaining a light distribution image provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0053] It should be noted that although the device schematics illustrate functional module divisions and the flowcharts illustrate logical sequences, in certain circumstances, the steps shown or described may be performed in a sequence that differs from the module divisions in the device or the sequence in the flowcharts. The terms "first," "second," and so on, in the specification, claims, and drawings, are used to distinguish similar items and are not necessarily used to describe a specific sequence or precedence.

[0054] Knowledge graph, known as knowledge domain visualization or knowledge domain mapping map in the library and information industry, is a series of various graphics that show the knowledge development process and structural relationship. It uses visualization technology to describe knowledge resources and their carriers, and to mine, analyze, construct, draw and display knowledge and their mutual connections. Knowledge graph can be used to effectively infer the results required by user input values.

[0055] Balcony farming is a new type of agricultural cultivation that has developed in recent years and holds great promise. As small plots of land commonly found in cities, balconies can be used to grow plants and reduce air pollution. The widespread application of modern intelligent control technologies has made automated management of balcony farming a reality.

[0056] A related art system provides a plant layout system that allows users to input data about desired plants through a human-computer interaction interface to generate a plant layout image. However, the connectivity between entities in this system is poor, resulting in weak contextual connections between entities and reducing the accuracy of the plant layout image generated by the system.

[0057] Based on the above situation, the embodiments of the present invention provide a method for establishing, obtaining, system and storage medium for a plant layout model, which improves the relevance of the context of entities and thereby improves the efficiency and accuracy of constructing plant layout images.

[0058] The embodiments of the present invention are further described below with reference to the accompanying drawings.

[0059] The first embodiment of the present invention specifically provides a method for establishing a plant layout model, such as Figure 1 The method for establishing the plant layout model includes but is not limited to the following steps:

[0060] Step S100, obtaining environmental requirement data for plant cultivation and a data source with professional plant cultivation knowledge;

[0061] It should be noted that professional plant cultivation knowledge is entity professional knowledge, and the environmental requirement data for plant cultivation is multi-source heterogeneous data of various factors that affect plant growth.

[0062] Step S200: performing entity labeling on the data source using a BiLSTM-CRF model and an attention mechanism to obtain a normalized first labeled sequence;

[0063] It should be noted that BiLSTM is the abbreviation of Bi-directional Long Short-Term Memory, which is translated as bidirectional long short-term memory network; CRF is the abbreviation of Conditional Random Field, which is translated as conditional random field; LSTM is the abbreviation of Long Short Term Memory, which is translated as long short-term memory network; and the attention mechanism is also called Attention mechanism.

[0064] It's important to note that entity labeling requires entity extraction and entity relationship extraction from the data source. This process is then used to label the professional plant cultivation knowledge text. The entity extraction process first segmentes the text into words, trains it using Word2Vec, and then generates word vectors using a BiLSTM layer. Finally, a CRF layer categorizes and labels the output vectors.

[0065] Specifically, refer to Figure 3 The BiLSTM-CRF model has a word embedding layer, namely the Word embeddings layer, the BILSTM layer, and the CRF layer. The word embedding layer uses the Google open source tool Word2vec model to train the input text, mapping each word in the text to a low-dimensional dense word vector as the input of the BiLSTM layer. The BILSTM layer uses two parallel layers in opposite directions based on the LSTM: the forward layer and the backward layer, which run from the front and back of the sequence respectively, to generate the text sequence x1, x2, x3…, x1 after the Word Embeddings layer. n As the input of the BiLSTM layer, the forward sequence of LSTM output and reverse sequence Perform splicing, mark the context, and obtain global features as the output of the BILSTM network. For example, the output splicing sequence is: Among them, h tThe CRF layer predicts labels for each word in the concatenated sequence output by the bidirectional LSTM, providing label constraints for the BiLSTM model and reducing the chances of incorrectly labeling entities.

[0066] S=(x1,x2,x3…,x n ),

[0067] Among them, x1, x2, x3…, x n Represents the text sequence after the Word Embeddings layer, S represents the text, and the expression of the CRF layer predicting the label sequence is:

[0068] y=(y1,y2,y3…y n ),

[0069] Among them, y and y1, y2, y3…y n Represents the predicted label sequence, and defines the calculation formula for the value of the CRF layer classifying the text S into the label y as follows:

[0070]

[0071] Where T represents the transfer score matrix, T i，j represents the transfer score from label i to label j, y0 and y n+1 are the newly added sentence start and end tags, and the dimension of T is (k+2)*(k+2); E is the output score matrix of the BiLSTM layer, and its dimension is n*k, where k represents the number of different tags, and E i,j Represents the score of the jth label of the i-th word in the sentence. When predicting all possible label sequences of the sentence, the softmax is used to obtain the probability of sequence y. The probability of sequence y is calculated as follows:

[0072]

[0073] During the training process, the logarithmic probability formula for maximizing the correct label sequence is as follows:

[0074]

[0075] Among them, Y S Represents all possible label sequences for sentence S. From the logarithmic probability formula, it can be seen that the model generates the label sequence with the highest probability. During decoding, the label sequence with the highest score is obtained as the final predicted output sequence. The calculation formula for the final predicted output sequence is as follows:

[0076]

[0077] in, argmax() represents a function that selects the maximum value.

[0078] It should be noted that the BiLSTM-CRF model is combined with the attention mechanism to extract entity relationships from the data source, which improves the relevance of the entity context and the effect of the BiLSTM-CRF model, thereby improving the construction efficiency and accuracy of plant layout images.

[0079] Step S300, performing entity disambiguation on the first labeled sequence to obtain a second labeled sequence;

[0080] It should be noted that after obtaining the first labeled sequence, i.e., labeled entities, through the BiLSTM-CRF model and the attention mechanism, the relationships between entities in the first labeled sequence are often ambiguous. Ambiguity means that one entity reference often corresponds to multiple real entities. Therefore, it is necessary to perform entity disambiguation on these entity references to obtain a second labeled sequence with a new entity relationship, i.e., the first entity relationship.

[0081] Step S400: performing knowledge reasoning on the second labeled sequence to obtain a knowledge graph;

[0082] It should be noted that during the construction of the knowledge graph, knowledge reasoning on the second labeled sequence can improve the knowledge graph, add semantic relationships that may have existed in the previous entities but have not been trained, and combine the existing knowledge graph to infer new knowledge to expand the knowledge graph, which can effectively perform deep reasoning and knowledge integration on knowledge, accelerate the construction process of the knowledge graph, and improve data accuracy.

[0083] Step S500: Obtain a plant layout algorithm model based on environmental demand data and knowledge graph.

[0084] It should be noted that incorporating environmental demand data for plant cultivation into the knowledge graph can enhance the organization and correlation of data, facilitate the storage and management of massive data, and form a plant layout algorithm model based on this.

[0085] This embodiment of the present invention includes: obtaining environmental requirement data for plant cultivation and a data source containing professional plant cultivation knowledge; performing entity extraction and entity labeling on the data source using a BiLSTM-CRF model and an attention mechanism to obtain a normalized first labeled sequence; then performing entity disambiguation on the labeled sequence to obtain a second labeled sequence with new entity relationships; performing knowledge reasoning on the second labeled sequence to obtain a knowledge graph; and obtaining a plant layout algorithm model based on the environmental requirement data and the knowledge graph. This setup, combining the BiLSTM-CRF model with an attention mechanism and then extracting entities and entity relationships from the data source containing professional plant cultivation knowledge, results in a plant layout algorithm model that improves the relevance of entities within their context, thereby increasing the efficiency and accuracy of constructing plant layout images.

[0086] Reference Figures 4 to 6 It is understood that the BiLSTM-CRF model includes a BiLSTM model and a conditional random field model. Step S200 includes but is not limited to the following steps:

[0087] Step S210: obtaining a joint feature vector sequence based on the BiLSTM model and the attention mechanism;

[0088] In step S220 , the joint feature vector sequence is used as an input of a conditional random field model, so that the conditional random field model performs entity labeling processing on the data source to obtain a normalized first labeled sequence.

[0089] It should be noted that the LSTM unit is mainly divided into three layers: input gate, forget gate and output gate. The input gate controls the proportion of new input information saved to the cell state, the forget gate controls the historical information retained by the cell state, and the output gate determines the final output information. The implementation process is as follows: Figure 5 As shown. Among them, the input gate is used to input information x t and the output signal h t-1 The information in is selectively saved, and the calculation formula is: t =σ(W i ·[h t-1 , x t ]+b i ), The forget gate is used to determine the cell state C t-1 The information that needs to be forgotten is calculated as follows: t =σ(W f ·[h t-1 , x t ]+b f ), combining the output information of the forget gate at time t-1 and the input information of the input gate at time t as the new cell state, the calculation formula is: The output gate is used to pass the cell state C after the forget gate and input gate at time t-1. t-1 , and the output signal h at time t-1 t-1 and the input signal x at time t t Integrate them together as the output signal at the current moment, and the calculation formula is: t =σ(W o ·[h t-1 , x t ]+b o ), h t =o t *tanh(C t ). For the above calculation formula, where i t 、f t 、o t They represent the input gate, forget gate, and output gate at time t, σ represents the Sigmoid activation function, W and b represent the corresponding weight matrix and bias vector, and h t represents the hidden state at time t, C t represents the cell state at time t, C t-1 represents the cell state at time t-1, h t-1 represents the hidden state at time t-1, Represents the candidate state obtained after the input gate passes through a tanh layer. The BiLSTM model is composed of a forward LSTM and a backward LSTM.

[0090] Reference Figures 6 and 7 It is understood that step S210 includes but is not limited to the following steps:

[0091] Step S211, obtaining a global feature vector corresponding to the data source according to the BiLSTM model;

[0092] Step S212, using the global feature vector as the input of the attention mechanism to obtain the local feature vector;

[0093] Step S213: combining the global feature vector and the local feature vector to obtain a joint feature vector sequence.

[0094] Specifically, the output vector of the BiLSTM layer in the BiLSTM-CRF model, that is, the global feature vector, is used as the input of the attention mechanism layer. The attention mechanism layer calculates the weight vector of each time series, and multiplies the output sequence of the BiLSTM layer by the weight vector so that the vocabulary-level features after each iteration are merged into sentence-level features to obtain local feature vectors. The local feature vectors are combined with the global features output by the BiLSTM layer to generate a joint feature vector sequence. The joint feature vector sequence is used as the input of the CRF layer so that the CRF layer performs entity labeling on the data source to obtain a normalized first labeled sequence.

[0095] It should be noted that, referring to Figure 6 , define x1, x2, x3…, x T The word vector and character representation joint vector sequence input to the BiLSTM layer, T represents the sequence length; a tj It represents the attention weight composition vector of the two. The calculation formula of the attention weight composition vector is as follows:

[0096]

[0097] Among them, e tj It represents the attention correlation coefficient, and the calculation formula of the attention correlation coefficient is as follows:

[0098]

[0099] Among them, v a Represents the weight of the global vector, h j Represents the output vector in the BiLSTM network, u a Represents the weight of the vector at the previous moment, c t-1 Represents the vector of the previous moment of the attention mechanism, w a Represents the weight of the attention mechanism at the previous moment. The weighted sum of the attention weight vector and the output of the BiLSTM model is used as the output of the attention mechanism. The calculation formula of the weighted sum is as follows:

[0100]

[0101] Among them, c t represents the output of the attention mechanism, a tj Represents the attention weights of the two composed of a vector, h j Denotes the output vector in the BiLSTM network, and T denotes the sequence length. The local feature vector obtained by the attention mechanism is combined with the global feature vector obtained by the BiLSTM to form a joint feature vector sequence, which serves as the input of the CRF layer.

[0102] Reference Figure 8It is understood that step S300 includes but is not limited to the following steps:

[0103] Step S310, determining at least one ambiguous entity set from the first labeled sequence;

[0104] Step S320, clustering the ambiguous entity set using an aggregated hierarchical clustering algorithm;

[0105] Step S330: Obtain a second labeled sequence according to the clustering result and the first labeled sequence.

[0106] It should be noted that the aggregation hierarchical clustering algorithm can disambiguate the entity referents in the first labeled sequence, cluster the referents pointing to the same target entity into the same category, and then map them to a known unambiguous structured knowledge graph. Specifically, first, each labeled entity in the first labeled sequence is treated as a separate class. In each iteration, the two classes closest to the aggregation center are found and merged to obtain a new class. The aggregation hierarchical clustering algorithm is used to sort the features according to their importance and select the feature subset with the best clustering quality. The entropy calculation formula is used for each feature, which is as follows:

[0107]

[0108] Among them, S ij Represents the similarity between two points i and j, which is based on the distance dist between the two points after the features are deleted i,j The calculation formula is as follows:

[0109]

[0110] set up Where dist represents the average distance of all points after removing feature t. After calculating the entropy of all features, the optimal feature subset is determined by calculating the clustering quality. Based on the hash matrix in the multi-analysis, the scattering criterion is used to measure the clustering quality. The calculation formulas for intra-class scatter and inter-class scatter are as follows:

[0111]

[0112]

[0113] Among them, P W represents intra-class scattering, p B represents inter-class scattering, m represents the total mean vector, and m j Represents the average vector of cluster j. Finally, the above steps are repeated in a round-robin manner, layer by layer, until there is only one cluster containing all entities, that is, the second labeled sequence with new entity relationships is formed.

[0114] Reference Figure 9 It is understood that step S400 includes but is not limited to the following steps:

[0115] Step S410: Perform knowledge reasoning on the second labeled sequence through a neural tensor network model to obtain a knowledge graph.

[0116] It should be noted that because the original knowledge graph is incomplete, many facts are not stored in the knowledge graph. During the knowledge graph construction process, the neural tensor network model is used to perform knowledge reasoning on the second labeled sequence. This can discover the implicit relationships between different entities in the second labeled sequence, thus improving the knowledge graph and adding semantic relationships that may have existed but were not trained. The neural tensor network model is also called the NTN model, which is the abbreviation of Neural Tensor Network.

[0117] It should be noted that the knowledge reasoning of the second labeled sequence through the neural tensor network model in step S400 includes: representing each entity in the second labeled sequence as a vector through the neural tensor network model to obtain the facts of the entity and the probability of specific relationships between different entities; defining each entity relationship through the parameters of the neural tensor network, representing the entity relationship as a third-order tensor and associating the two entities.

[0118] Specifically, a neural tensor network model is used to represent entities as vectors to obtain entity information, and relationships are represented as parameters, which are then represented as third-order tensors. Two entities are associated across multiple dimensions. This is used to infer new knowledge that is as close to the facts as possible based on the existing knowledge base to expand the knowledge base and obtain a knowledge graph. The calculation formula for knowledge inference using the neural tensor network model is as follows:

[0119]

[0120] Among them, f = tanh represents a standard nonlinear function, represents a tensor, represents a bilinear tensor product, the result of which is h∈Rk, and each h i Represents a tensor slice The other parameters of the relationship R refer to the standard neural network: V R ∈R k×2d 、U∈R k 、b R ∈R k The higher the score of function g(e1, R, e2), the higher the possibility that there is a relationship R between entities e1 and e2.

[0121] It is understandable that the environmental demand data includes illumination data, temperature data and humidity data.

[0122] It should be noted that incorporating the environmental requirement data of plant cultivation, namely illumination data, temperature data, and humidity data, into the knowledge graph can enhance the organization and correlation of data, facilitate the storage and management of massive data, and form a plant layout algorithm model based on this, thereby improving the accuracy of constructing plant layout images.

[0123] It should be noted that, referring to Figure 2 The illumination data is multi-directional illumination data. After the illumination distribution is rendered, the illumination data is combined with the temperature data and humidity data and input into the knowledge graph to obtain a plant layout algorithm model based on the knowledge graph.

[0124] In a second aspect, the embodiment of the present invention further provides a method for obtaining plant layout, referring to Figure 10 , the method includes but is not limited to the following steps:

[0125] Step S600, obtaining a light distribution image and a plant data set corresponding to the user's plant planting requirements;

[0126] Step S700: inputting the plant data set into the plant layout algorithm model according to the illumination distribution image to obtain a plant layout image.

[0127] It should be noted that the plant dataset may include the plant name, quantity, and species, etc., and may also include the characteristics of the plant.

[0128] Specifically, refer to Figure 2 The system then goes through entity labeling based on the BiLSTM-CRF model and attention mechanism, entity disambiguation based on the aggregated hierarchical clustering algorithm, and knowledge reasoning based on the neural tensor network model. The environmental requirements data for plant cultivation is then input into the knowledge graph to create a plant layout algorithm model based on the knowledge graph. Users enter the plant dataset and environmental distribution structure corresponding to their planting requirements into the human-computer interaction interface. Based on the environmental distribution structure and illumination data collected by multiple sensors, a light distribution image is generated. Based on this light distribution image, the plant planting requirements are then input into the plant layout algorithm model to create the optimal plant layout image for each plant.

[0129] It should be noted that in this embodiment, professional plant planting knowledge is used as the data source input, and the BiLSTM-CRF model combined with the attention mechanism outputs the normalized labeled entities, that is, the first labeled sequence. The labeled entities are classified by category using the aggregation hierarchical clustering algorithm, and entities belonging to the same category are clustered together to form a second labeled sequence with a first entity relationship. The neural tensor network model is used to perform knowledge reasoning on the existing knowledge base to obtain new plant planting knowledge and expand the original knowledge graph to form a new knowledge graph. Finally, the knowledge graph is stored in the cloud, and the environmental requirement data of the planted plants is input into the new knowledge graph to obtain a plant layout algorithm model.

[0130] Reference Figure 11 It is understood that obtaining the illumination distribution image in step S600 includes but is not limited to the following steps:

[0131] Step S610, obtaining illumination data collected by multiple sensors and an environment distribution structure corresponding to the plants planted by the user;

[0132] Step S620: Obtain an illumination distribution image according to the environment distribution structure and illumination data.

[0133] In some embodiments, if a user wants to plant plants on a balcony, they need to input a plant dataset and the balcony's environmental distribution structure. Based on the balcony's environmental distribution structure and the illumination data collected by multiple sensors, a light distribution image is obtained. The plant layout algorithm model performs corresponding arrangement calculations based on the environmental demand data for the planted plants, the plant's characteristic weights, and the illumination distribution image, thereby obtaining a layout of the most suitable locations for each plant to be planted. In other embodiments, plants can also be planted in other environments. The plant dataset and the balcony's environmental distribution structure in that environment are both input through a human-computer interaction interface, and a plant layout image is obtained. This is not limited in the embodiments of the present invention.

[0134] In addition, the third embodiment of the present invention further provides a plant layout system, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor.

[0135] The processor and the memory may be connected via a bus or other means.

[0136] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0137] The non-transient software program and instructions required to implement the method for establishing the plant layout model of the first embodiment are stored in the memory. When executed by the processor, the method for establishing the plant layout model of the embodiment is executed, for example, the method described above is executed. Figure 1 Method steps S100 to S500, Figure 4 Steps S210 to S220 of the method, Figure 7 Steps S211 to S213 of the method, Figure 8 Method steps S310 to 330, Figure 9 Method step S410 in .

[0138] The non-transient software program and instructions required to implement the plant layout acquisition method of the second embodiment are stored in the memory. When executed by the processor, the plant layout acquisition method of the embodiment is executed, for example, the above described Figure 10 Steps S600 to S700 of the method, Figure 11 Method steps S610 to S620.

[0139] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0140] In addition, an embodiment of the present invention further provides a computer-readable storage medium, which stores computer-executable instructions. The computer-executable instructions are executed by a processor or a controller, for example, by a processor in the above-mentioned device embodiment, so that the above-mentioned processor can execute the method for establishing the plant layout model or the method for obtaining the plant layout in the above-mentioned embodiment, for example, executing the above-mentioned Figure 1 Method steps S100 to S500, Figure 4 Steps S210 to S220 of the method, Figure 7Steps S211 to S213 of the method, Figure 8 Method steps S310 to 330, Figure 9 Method step S410, Figure 10 Steps S600 to S700 of the method, Figure 11 Method steps S610 to S620.

[0141] Those skilled in the art will appreciate that all or some of the steps and systems in the method disclosed above can be implemented as software, firmware, hardware, and appropriate combinations thereof. Some physical components or all physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, and the computer-readable medium can include computer storage media (or non-transitory media) and communication media (or temporary media). As known to those skilled in the art, the term computer storage media is included in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data) and is volatile and non-volatile, removable, and non-removable. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory, or other memory technology, CD-ROM, digital versatile disks (DVD), or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage, or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, as is well known to those skilled in the art, communication media typically embodies computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0142] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the above implementation. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of the present invention.

Claims

1. A method for establishing a plant layout model, characterized in that: include: Obtain data on the environmental requirements of plant cultivation and data sources with professional plant cultivation knowledge; Performing entity labeling on the data source through a BiLSTM-CRF model and an attention mechanism to obtain a normalized first labeled sequence; Performing entity disambiguation on the first labeled sequence to obtain a second labeled sequence; Performing knowledge reasoning on the second labeled sequence to obtain a knowledge graph; Obtaining a plant layout algorithm model according to the environmental demand data and the knowledge graph; Wherein, the BiLSTM-CRF model includes a BiLSTM model and a conditional random field model; The entity labeling process of the data source is performed using the BiLSTM-CRF model and the attention mechanism to obtain a normalized first labeled sequence, including: Obtaining a global feature vector corresponding to the data source according to the BiLSTM model; Using the global feature vector as the input of the attention mechanism to obtain a local feature vector; Combining the global feature vector with the local feature vector to obtain a joint feature vector sequence; The joint feature vector sequence is used as the input of the conditional random field model, so that the conditional random field model performs entity labeling processing on the data source to obtain a normalized first labeled sequence.

2. The method for establishing a plant layout model according to claim 1, wherein: The performing entity disambiguation on the first labeled sequence to obtain a second labeled sequence includes: determining at least one ambiguous entity set from the first labeled sequence; Clustering the ambiguous entity set using an aggregated hierarchical clustering algorithm; A second labeled sequence is obtained according to the clustering result and the first labeled sequence.

3. The method for establishing a plant layout model according to claim 1, wherein: The performing knowledge reasoning on the second labeled sequence to obtain a knowledge graph includes: The second labeled sequence is subjected to knowledge reasoning through a neural tensor network model to obtain a knowledge graph.

4. The method for establishing a plant layout model according to claim 1, wherein: The environmental demand data includes illumination data, temperature data and humidity data.

5. A method for obtaining plant layout, characterized in that: include: Obtain light distribution images and plant datasets corresponding to the user's plant planting requirements; According to the illumination distribution image, the plant data set is input into a plant layout algorithm model to obtain a plant layout image; wherein the plant layout algorithm model is established by the plant layout model establishment method described in any one of claims 1 to 4.

6. The method for establishing a plant layout model according to claim 5, characterized in that: The obtaining of the illumination distribution image comprises: Obtaining the illumination data collected by multiple sensors and the environmental distribution structure corresponding to the plants planted by the user; The illumination distribution image is obtained according to the environmental distribution structure and the illumination data.

7. A plant layout system, characterized in that: include: 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 establishing a plant layout model according to any one of claims 1 to 4; or, The method for obtaining plant layout according to any one of claims 5 to 6.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions for: Executing the method for establishing a plant layout model according to any one of claims 1 to 4; or, Execute the method for acquiring plant layout according to any one of claims 5 to 6.

Images