Intelligent sugarcane yield prediction method and system

Abstract

The invention discloses an intelligent sugarcane yield prediction method and system, and relates to the technical field of sugarcane planting, and the method comprises the steps: constructing a sample multi-modal environment perception matrix corresponding to each sample sugarcane planting region; respectively analyzing each sample multi-modal environment perception matrix to obtain a plurality of sample space-time fusion feature codes associated with the sugarcane yield; based on all the sample space-time fusion feature codes, taking the space-time fusion feature codes as supervision signals of a lightweight student model, and guiding the lightweight student model to perform feature learning on a sample multi-modal environment perception matrix through a feature mapping loss function to obtain a trained sugarcane yield prediction model; and predicting the sugarcane yield of the target sugarcane planting area by using the trained sugarcane yield prediction model. According to the method, the prediction precision of the sugarcane yield can be improved, and meanwhile low calculation complexity and high calculation efficiency are kept.

Description

Background Art

[0002] Currently, the methods for predicting sugarcane yield include: field investigation method, remote sensing technology method, and statistical model analysis method. Specifically:

[0003] 1) The field investigation method requires a large amount of manpower, material resources and time, with low efficiency. Moreover, the sample size of field investigation is limited, making it difficult to comprehensively cover the sugarcane planting area, which may lead to insufficient representativeness and accuracy of data. In addition, during the field investigation process, the subjective judgment and experience of investigators will also affect the accuracy of data.

[0004] 2) The remote sensing technology method has the advantages of high efficiency and wide coverage, but it is sensitive to weather conditions. For example, cloudy, rainy and other weather will affect the acquisition quality of satellite images, resulting in missing or unclear remote sensing images during the critical growth period.

[0005] 3) The statistical model analysis method depends on the quality and integrity of historical data. If the data is missing, has errors or is inconsistent, it will affect the accuracy and reliability of the model. At the same time, the parameter setting of the statistical model requires professional knowledge and experience. Different parameter settings may lead to different prediction results, increasing the difficulty and uncertainty of model construction. In addition, traditional statistical models may have limitations in dealing with complex non-linear relationships and cannot fully capture various complex factors and their interactions during the sugarcane growth process. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a sugarcane yield intelligent prediction method and system in view of the deficiencies of the prior art, specifically as follows:

[0007] 1) In the first aspect, the present invention provides a sugarcane yield intelligent prediction method, and the specific technical solution is as follows:

[0008] Construct a sample multi-modal environment perception matrix corresponding to each sample sugarcane planting area;

[0009] Analyze each sample multi-modal environment perception matrix respectively to obtain a plurality of sample spatio-temporal fusion feature encodings associated with sugarcane yield;

[0010] Based on all the sample spatio-temporal fusion feature encodings, use the spatio-temporal fusion feature encoding as the supervision signal of the lightweight student model, and guide the lightweight student model to perform feature learning on the sample multi-modal environment perception matrix through the feature mapping loss function to obtain a trained sugarcane yield prediction model;

[0011] Use the trained sugarcane yield prediction model to predict the sugarcane yield in the target sugarcane planting area.

[0012] The beneficial effects of an intelligent sugarcane yield prediction method provided by the present invention are as follows:

[0013] On the one hand, through the sample multi-modal environment perception matrix, the growth status and influencing factors of sugarcane can be more comprehensively characterized, providing richer information for the model, thereby improving the accuracy and reliability of prediction. On the other hand, using the spatio-temporal fusion feature encoding as the supervision signal for the lightweight student model can provide guidance and supervision for the lightweight student model, helping the lightweight student model better learn and understand the complex patterns and rules in the sample multi-modal environment perception matrix, so that the trained sugarcane yield prediction model can more accurately capture the key factors in the sugarcane growth process and their impact on yield, thereby improving the prediction accuracy of sugarcane yield while maintaining a low computational complexity and high computational efficiency.

[0014] Based on the above solution, an intelligent sugarcane yield prediction method of the present invention can also be improved as follows.

[0015] Further, the process of constructing the sample multi-modal environment perception matrix includes:

[0016] Performing spatio-temporal alignment on the vegetation index, hyperspectral data, soil data, and meteorological data of any sample sugarcane planting area to obtain the corresponding multi-modal environment perception matrix of the sample sugarcane planting area. The multi-modal environment perception matrix has three dimensions, namely, the spatial dimension, the time dimension, and the modal dimension. The modal dimension includes the vegetation index, hyperspectral data, soil data, and meteorological data.

[0017] Further, parsing each sample multi-modal environment perception matrix respectively includes:

[0018] Parsing each sample multi-modal environment perception matrix respectively through the trained dual attention spatio-temporal graph convolutional network.

[0019] The vegetation index of any sample sugarcane planting area includes: the normalized difference vegetation index, enhanced vegetation index, and land surface water index of the sample sugarcane planting area at different times. The hyperspectral data of any sample sugarcane planting area includes: the spectral reflectance curve of sugarcane, the texture parameters of sugarcane leaves, and the temporal change curve of the canopy greenness and water content of sugarcane in the sample sugarcane planting area at different times. The meteorological data of any sample sugarcane planting area includes: the temperature, humidity, light intensity, wind speed, and wind direction in the sample sugarcane planting area at different times.

[0020] 2) Second, the present invention also provides an intelligent sugarcane yield prediction system, and the specific technical solution is as follows:

[0021] Including a matrix construction module, an analysis module, a training module, and a sugarcane yield prediction module;

[0022] The matrix construction module is used to: construct a sample multi-modal environmental perception matrix corresponding to each sample sugarcane planting area;

[0023] The parsing module is used to: parse each sample multi-modal environmental perception matrix respectively to obtain multiple sample spatio-temporal fusion feature encodings associated with sugarcane yield;

[0024] The training module is used to: based on all the sample spatio-temporal fusion feature encodings, use the spatio-temporal fusion feature encodings as the supervision signal of the lightweight student model, and through the feature mapping loss function, guide the lightweight student model to perform feature learning on the sample multi-modal environmental perception matrix to obtain a trained sugarcane yield prediction model;

[0025] The sugarcane yield prediction module is used to: use the trained sugarcane yield prediction model to predict the sugarcane yield of the target sugarcane planting area.

[0026] Based on the above solution, an intelligent sugarcane yield prediction system of the present invention can also be improved as follows.

[0027] Further, the matrix construction module is specifically used to:

[0028] Perform spatio-temporal alignment on the vegetation index, hyperspectral data, soil data, and meteorological data of any sample sugarcane planting area to obtain the multi-modal environmental perception matrix corresponding to the sample sugarcane planting area. The multi-modal environmental perception matrix has three dimensions, namely the spatial dimension, the time dimension, and the modal dimension. The modal dimension includes the vegetation index, hyperspectral data, soil data, and meteorological data.

[0029] Further, the parsing module is specifically used to: parse each sample multi-modal environmental perception matrix respectively through the trained dual attention spatio-temporal graph convolutional network.

[0030] Further, the vegetation index of any sample sugarcane planting area includes: the normalized difference vegetation index, enhanced vegetation index, and land surface water index of the sample sugarcane planting area at different times. The hyperspectral data of any sample sugarcane planting area includes: the spectral reflectance curve of sugarcane, the texture parameters of sugarcane leaves, and the temporal variation curve of the canopy greenness and water content of sugarcane in the sample sugarcane planting area at different times. The meteorological data of any sample sugarcane planting area includes: the temperature, humidity, light intensity, wind speed, and wind direction in the sample sugarcane planting area at different times.

[0031] 3) Thirdly, the present invention also provides an electronic device, which includes a processor coupled to a memory. At least one computer program is stored in the memory and is loaded and executed by the processor to enable the electronic device to implement any of the above sugarcane yield intelligent prediction methods.

[0032] 4) Fourthly, the present invention also provides a computer-readable storage medium with a computer program stored thereon. When the computer program is executed by a processor, it implements any of the above sugarcane yield intelligent prediction methods.

[0033] It should be noted that for the beneficial effects obtained by the technical solutions of the second to fourth aspects of the present invention and their corresponding possible implementation manners, reference may be made to the above technical effects of the first aspect and its corresponding possible implementation manners, which will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for description in the embodiments of the present invention:

[0035] Figure 1 It is a schematic flowchart of a sugarcane yield intelligent prediction method according to an embodiment of the present invention;

[0036] Figure 2 It is a schematic structural diagram of a sugarcane yield intelligent prediction system according to an embodiment of the present invention;

[0037] Figure 3 It is a schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0038] The following describes the principles and features of the present invention. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0039] The following uses specific embodiments to detail the technical solutions of the present invention and how the technical solutions of the present invention solve the above technical problems. These several specific embodiments can be combined with each other. For the same or similar concepts or processes, they may not be repeated in some embodiments. The following will describe the embodiments of the present invention in conjunction with the drawings.

[0040] As Figure 1 shown, a sugarcane yield intelligent prediction method according to an embodiment of the present invention includes the following steps:

[0041] S1. Construct a sample multi-modal environmental perception matrix corresponding to each sample sugarcane planting area;

[0042] Among them, the process of constructing the sample multi-modal environmental perception matrix includes:

[0043] Spatially and temporally align the vegetation index, hyperspectral data, soil data, and meteorological data of any sample sugarcane planting area to obtain the corresponding multimodal environmental perception matrix for the sample sugarcane planting area. The multimodal environmental perception matrix has three dimensions, namely the spatial dimension, the temporal dimension, and the modal dimension. The modal dimension includes the vegetation index, hyperspectral data, soil data, and meteorological data.

[0044] Among them, the vegetation index of any sample sugarcane planting area includes: the normalized difference vegetation index, enhanced vegetation index, and land surface water index of the sample sugarcane planting area at different times. The hyperspectral data of any sample sugarcane planting area includes: the spectral reflectance curve of sugarcane, the texture parameters of sugarcane leaves, and the temporal variation curve of the canopy greenness and water content of sugarcane in the sample sugarcane planting area at different times. The meteorological data of any sample sugarcane planting area includes: the air temperature, humidity, light intensity, wind speed, and wind direction of the sample sugarcane planting area at different times.

[0045] Among them, the acquisition process of the vegetation index is as follows:

[0046] Collect the infrared spectral data of the sample sugarcane planting area through a multispectral and thermal infrared satellite or a thermal infrared camera carried by an unmanned aerial vehicle (UAV), and calculate the normalized difference vegetation index, enhanced vegetation index, and land surface water index at different times based on the multispectral data.

[0047] Among them, the acquisition process of the hyperspectral data is as follows:

[0048] Collect the hyperspectral image data of each sample sugarcane planting area through a hyperspectral imager carried by an unmanned aerial vehicle (UAV). Through radiometric calibration and reflectance inversion, extract the spectral characteristics of each pixel point in the hyperspectral image data, calculate the spectral reflectance curve of sugarcane at different times, and calculate the texture parameters of sugarcane leaves based on the hyperspectral image data and using the gray-level co-occurrence matrix (GLCM). Align the hyperspectral image data at different growth stages, and generate the temporal variation curve of the canopy greenness and water content of sugarcane.

[0049] Among them, the acquisition process of the meteorological data is as follows:

[0050] Collect the air temperature, humidity, light intensity, wind speed, and wind direction of each sample sugarcane planting area at different times through an air temperature sensor, humidity sensor, light intensity sensor, and wind speed and direction integrated sensor.

[0051] Among them, the soil data includes soil conductivity, pH value, water content, and nitrogen, phosphorus, and potassium ion concentrations, which are specifically detected by corresponding sensors.

[0052] Spatially and temporally align the vegetation index, hyperspectral data, soil data, and meteorological data of the sample sugarcane planting areas.

[0053] Construct a multimodal environmental perception matrix with three-layer tensor dimensions. The three layers of tensors correspond to the spatial dimension, the temporal dimension, and the modal dimension respectively. In the spatial dimension, the sample sugarcane planting areas are divided into 1m×1m plots in a grid manner. In the temporal dimension, it is divided by "day". In the modal dimension, it includes the vegetation index, hyperspectral data, soil data, and meteorological data.

[0054] S2. Analyze each sample multimodal environmental perception matrix respectively to obtain multiple sample spatio-temporal fusion feature encodings associated with sugarcane yield. Specifically:

[0055] Analyze each sample multimodal environmental perception matrix respectively through a trained dual-attention spatio-temporal graph convolutional network to obtain multiple sample spatio-temporal fusion feature encodings associated with sugarcane yield.

[0056] The dual-attention spatio-temporal graph convolutional network includes a spatial topology module, a temporal causal convolution module, and a spatio-temporal graph convolutional layer. Specifically:

[0057] 1) The spatial topology module is used to: extract the modal features (i.e., the data in the modal dimension) within a day from the sample multimodal environmental perception matrix through a spatial attention layer. After inputting the modal features into a fully connected layer, a query matrix and a key matrix are obtained. The query matrix and the key matrix are input into the Softmax function to obtain a normalized weight matrix. The normalized weight matrix is input into a static topology encoding layer to calculate an adjacency weight matrix. The adjacency weight matrix can represent the dynamic environmental correlation between different plots. For example, the similarity of the vegetation index, hyperspectral data, soil data, and meteorological data between different plots. The adjacency weight matrix can strengthen the environmental coupling effect between different plots.

[0058] 2) The temporal causal convolution module is used to: extract the vegetation index, hyperspectral data, soil data, and meteorological data that change with time for each plot from the sample multimodal environmental perception matrix through a general time attention layer, and after inputting them into the Sigmoid function, obtain the time attention weight of each plot. This process can enhance the temporal dependence of the key growth stages of sugarcane (such as the filling stage) and suppress the noise in the non-key growth stages. Input the time attention weight of each plot into a one-way masked causal convolution layer to obtain a temporal evolution feature. The temporal evolution feature is: a feature used to represent the temporal dependence across growth stages in the entire life cycle of sugarcane, including periodic, trend, and mutation environmental evolution laws. Specifically:

[0059] The specific characteristics of temporal evolution include periodic characteristics, trend characteristics, and mutation characteristics. The periodic characteristics include: the daily peak and nightly trough of photosynthetically active radiation (PAR), the periodic changes in soil water content at the critical growth stages of sugarcane, etc. The trend characteristics include: the characteristics corresponding to the decreasing trend of nitrogen, phosphorus, and potassium ion concentrations caused by continuous planting and the characteristics corresponding to the increasing trend of soil electrical conductivity (EC value) caused by improper irrigation, etc. The mutation characteristics include: the characteristics corresponding to the sudden increase in soil water content caused by heavy rain and the characteristics corresponding to the canopy water stress caused by drought, etc.

[0060] 3) The spatio-temporal graph convolutional layer is used to: receive the adjacency weight matrix and the temporal evolution characteristics through the spatial graph convolutional layer, and use the convolutional layer to splice the adjacency weight matrix and the temporal evolution characteristics along the channel dimension to obtain the spatio-temporal fusion feature encoding.

[0061] By constructing the corresponding dataset, a trained dual-attention spatio-temporal graph convolutional network can be obtained.

[0062] S3. Based on all the sample spatio-temporal fusion feature encodings, use the spatio-temporal fusion feature encoding as the supervision signal for the lightweight student model, and through the feature mapping loss function, guide the lightweight student model to perform feature learning on the sample multi-modal environmental perception matrix to obtain a trained sugarcane yield prediction model;

[0063] Use the trained dual-attention spatio-temporal graph convolutional network as the teacher model, use the spatio-temporal fusion feature encoding as the supervision signal for the lightweight student model, and through the feature mapping loss function, guide the lightweight student model to perform feature learning on the sample multi-modal environmental perception matrix to obtain a trained lightweight student model. The specific training process is as follows:

[0064] S30. Input the sample multi-modal environmental perception matrix into the teacher model and output the spatio-temporal fusion feature encoding;

[0065] S31. Use the principal component analysis method to perform principal component dimensionality reduction on the sample multi-modal environmental perception matrix to generate the dimensionality-reduced sample multi-modal environmental perception matrix.

[0066] S32. Construct a lightweight student model, input the dimensionality-reduced sample multi-modal environmental perception matrix into the lightweight student model, and output the calibrated spatio-temporal feature encoding and the sugarcane yield prediction value;

[0067] The lightweight student model includes: a static spatial graph convolutional layer, a depthwise separable causal convolutional layer, a channel attention module, a first average pooling layer, a second average pooling layer, and a fully connected layer. Specifically:

[0068] 1) The static spatial graph convolutional layer is used to: process the sample multi-modal environmental perception matrix after dimensionality reduction using the adjacency weight matrix to aggregate the environmental features of adjacent fields and obtain spatially enhanced features;

[0069] 2) The depthwise separable causal convolutional layer is used to: perform depth convolution operations and point convolution operations on the spatially enhanced features to obtain temporal evolution features;

[0070] 3) The channel attention module is used to: compress the temporal evolution features into channel description vectors through the first average pooling layer. After the channel description vectors are processed through two fully connected layers in sequence, channel weights are generated. The channel weights are multiplied element-wise with the temporal evolution features to obtain the calibrated spatio-temporal feature encoding;

[0071] 4) The calibrated spatio-temporal feature encoding is average pooled through the second average pooling layer to obtain the field-level feature vector. The field-level feature vector is input into the fully connected layer to obtain the predicted sugarcane yield value.

[0072] S33. Construct a feature mapping loss function and calculate the total loss through the feature mapping loss function.

[0073] Among them, the purpose of constructing the feature mapping loss function is to make the calibrated spatio-temporal feature encoding output by the lightweight student model approximate the spatio-temporal fusion feature encoding of the teacher model, while ensuring the prediction accuracy of the sugarcane yield. The feature mapping loss function can specifically be:

[0074] L = βL 1 +(1 - β)L 2

[0075] Among them, L represents the total loss, β is the balance coefficient, and the value range of β is: 0 to 1. L 1 represents: the KL divergence used to characterize the feature distribution alignment loss between the calibrated spatio-temporal feature encoding output by the lightweight student model and the spatio-temporal fusion feature encoding of the teacher model. L 2 represents: the mean square error between the calibrated spatio-temporal feature encoding output by the lightweight student model and the spatio-temporal fusion feature encoding of the teacher model.

[0076] S34. Based on the total loss, dynamically adjust the lightweight student model through the gradient backpropagation algorithm, and then use the adjusted lightweight student model as the lightweight student model constructed in S32, and return to execute S32 until the trained lightweight student model is obtained. The trained lightweight student model is the trained sugarcane yield prediction model.

[0077] S4. Use the trained sugarcane yield prediction model to predict the sugarcane yield in the target sugarcane planting area.

[0078] The beneficial effects of a method for intelligent prediction of sugarcane yield provided by the present invention are as follows:

[0079] 1) Through the sample multi-modal environment perception matrix, the growth conditions and influencing factors of sugarcane can be more comprehensively characterized, providing richer information for the model, thereby improving the accuracy and reliability of prediction.

[0080] 2) Spatiotemporal fusion feature encoding, as the supervision signal of the lightweight student model, can provide guidance and supervision for the lightweight student model, helping the lightweight student model better learn and understand the complex patterns and rules in the sample multi-modal environment perception matrix, so that the trained sugarcane yield prediction model can more accurately capture the key factors in the sugarcane growth process and their impact on yield, thereby improving the prediction accuracy of sugarcane yield, while maintaining a low computational complexity and high computational efficiency.

[0081] 3) The lightweight student model has low computational resource requirements. The amount of data participating in the calculation is about 25%-30% of that of the teacher model, which is suitable for field embedded deployment. Moreover, through the KL divergence, the lightweight student model can be forced to integrate the feature distributions between the spatiotemporal fusion feature encodings of the teacher model, and only a small amount of labeled data is required for fine-tuning. The dual attention spatiotemporal graph convolutional network is advantageous in high-precision spatiotemporal modeling and multi-modal fusion, providing a reliable distillation benchmark, that is, a supervision signal, which helps to improve the sugarcane yield prediction accuracy of the trained lightweight student model.

[0082] In the above embodiments, although the steps are numbered S1, S2, etc., these are only specific embodiments given by the present invention. Those skilled in the art can adjust the execution order of S1, S2, etc. according to the actual situation, which is also within the protection scope of the present invention. It can be understood that in some embodiments, it may include some or all of the above embodiments.

[0083] As Figure 2 shown, an intelligent sugarcane yield prediction system 200 according to an embodiment of the present invention includes a matrix construction module 201, an analysis module 202, a training module 203, and a sugarcane yield prediction module 204;

[0084] The matrix construction module 201 is used to: construct a sample multi-modal environment perception matrix corresponding to each sample sugarcane planting area;

[0085] The analysis module 202 is used to: respectively analyze each sample multi-modal environment perception matrix to obtain a plurality of sample spatiotemporal fusion feature encodings associated with sugarcane yield;

[0086] The training module 203 is configured to: based on all the sample spatio-temporal fusion feature encodings, use the spatio-temporal fusion feature encoding as the supervision signal for the lightweight student model, and guide the lightweight student model to perform feature learning on the sample multi-modal environment perception matrix through the feature mapping loss function, so as to obtain a trained sugarcane yield prediction model;

[0087] The sugarcane yield prediction module 204 is configured to: use the trained sugarcane yield prediction model to predict the sugarcane yield in the target sugarcane planting area.

[0088] Optionally, in the above technical solution, the matrix construction module 201 is specifically configured to:

[0089] Perform spatio-temporal alignment on the vegetation index, hyperspectral data, soil data, and meteorological data of any sample sugarcane planting area to obtain the corresponding multi-modal environment perception matrix for the sample sugarcane planting area. The multi-modal environment perception matrix has three dimensions, namely the spatial dimension, the time dimension, and the modality dimension. The modality dimension includes the vegetation index, hyperspectral data, soil data, and meteorological data.

[0090] Optionally, in the above technical solution, the parsing module 202 is specifically configured to: parse each sample multi-modal environment perception matrix respectively through the trained dual attention spatio-temporal graph convolutional network.

[0091] Optionally, in the above technical solution, the vegetation index of any sample sugarcane planting area includes: the normalized difference vegetation index, enhanced vegetation index, and land surface water index of the sample sugarcane planting area at different times. The hyperspectral data of any sample sugarcane planting area includes: the spectral reflectance curve of sugarcane, the texture parameters of sugarcane leaves, and the temporal variation curve of the canopy greenness and water content of sugarcane in the sample sugarcane planting area at different times. The meteorological data of any sample sugarcane planting area includes: the air temperature, humidity, light intensity, wind speed, and wind direction in the sample sugarcane planting area at different times.

[0092] It should be noted that the beneficial effects of the sugarcane yield intelligent prediction system 200 provided in the above embodiment are the same as those of the above sugarcane yield intelligent prediction method, and will not be elaborated here. In addition, when the system provided in the above embodiment implements its functions, only the above-mentioned division of each functional module is used for illustration. In practical applications, the above functions can be allocated to different functional modules according to needs, that is, the system can be divided into different functional modules according to the actual situation to complete all or part of the functions described above. In addition, the system provided in the above embodiment and the method embodiment belong to the same concept, and the specific implementation process can be seen in the method embodiment, which will not be elaborated here.

[0093] Among them, the intelligent sugarcane yield prediction system of the present invention can be a computer program (including program code) running on a computer device. For example, the intelligent sugarcane yield prediction system of the present invention is an application software and can be used to execute the corresponding steps in the intelligent sugarcane yield prediction method of the present invention.

[0094] In some embodiments, the intelligent sugarcane yield prediction system of the present invention can be implemented in a combination of software and hardware. As an example, the intelligent sugarcane yield prediction system of the present invention can be a processor in the form of a hardware decoding processor, which is programmed to execute the intelligent sugarcane yield prediction method of the present invention. For example, the processor in the form of a hardware decoding processor can employ one or more application specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs) or other electronic components.

[0095] Among them, the modules involved in the embodiments of the present invention can be implemented in software or in hardware. Among them, the name of the module does not constitute a limitation on the module itself in some cases.

[0096] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the above-mentioned any intelligent sugarcane yield prediction method is implemented. That is to say, an electronic device according to an embodiment of the present invention may include, but is not limited to: a processor and a memory; the memory is used to store the computer program; the processor is used to execute the intelligent sugarcane yield prediction method shown in any embodiment of the present invention by calling the computer program.

[0097] In an alternative embodiment, an electronic device is provided, as Figure 3 shown, Figure 3 The electronic device 4000 shown includes: a processor 4001 and a memory 4003. Among them, the processor 4001 and the memory 4003 are connected, such as connected through a bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004, and the transceiver 4004 can be used for data interaction between the electronic device and other electronic devices, such as data sending and / or data receiving, etc. It should be noted that in practical applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0098] The processor 4001 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logic blocks, modules, and circuits described in connection with the disclosure of the present invention. The processor 4001 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0099] The bus 4002 can include a path for transmitting information between the above components. The bus 4002 can be a PCI (Peripheral Component Interconnect) bus, an EISA (Extended Industry Standard Architecture) bus, or the like. The bus 4002 can be divided into an address bus, a data bus, a control bus, etc. For the sake of representation, Figure 3 only a thick line is used to represent the bus 4002 in the figure, but it does not mean that there is only one bus or one type of bus.

[0100] The memory 4003 can be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, or it can also be an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic storage media, or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0101] The memory 4003 is used to store the application program code (computer program) for executing the solution of the present invention, and is controlled by the processor 4001 for execution. The processor 4001 is used to execute the application program code stored in the memory 4003 to implement the content shown in the foregoing method embodiments.

[0102] Among them, the electronic device may also be a terminal device, and the terminal device may be any device on which an application can be installed, including at least one of a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch, a smart TV, and a smart vehicle-mounted device.

[0103] It should be noted that Figure 3 The electronic device shown is only an example and should not impose any limitations on the functions and usage scope of the embodiments of the present invention.

[0104] A computer-readable storage medium according to an embodiment of the present invention has a computer program stored thereon, and when the computer program is executed by a processor, it implements any one of the above-mentioned sugarcane yield intelligent prediction methods.

[0105] Optionally, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, a floppy disk, and an optical data storage device, etc.

[0106] In an exemplary embodiment, there is also provided a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the electronic device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the electronic device executes any one of the above-mentioned sugarcane yield intelligent prediction methods.

[0107] Computer program code for performing the operations of the present invention may be written in one or more programming languages or combinations thereof. The above-mentioned programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0108] It should be understood that the flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, may be implemented by a dedicated hardware-based system for performing the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0109] The computer-readable storage medium provided by the embodiments of the present invention may be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EEPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0110] The above computer-readable storage medium stores one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.

[0111] The above description is only a preferred embodiment of the present invention and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of disclosure involved in the present invention is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosure concept. For example, the technical solutions formed by mutually replacing the above features with the technical features (but not limited to) having similar functions disclosed in the present invention.

[0112] It should be noted that the terms "first", "second", etc. in the specification and claims of this application are used to distinguish similar objects, and represent a limitation on a specific order or sequence. In appropriate cases, the order of use of similar objects can be interchanged so that the embodiments of the present application described here can be implemented in an order other than the illustrated or described order.

[0113] Those skilled in the art know that the present invention can be implemented as a system, a method or a computer program product. Therefore, the present invention can be specifically implemented in the following forms, that is: it can be completely hardware, can also be completely software (including firmware, resident software, microcode, etc.), and can also be a combination of hardware and software, which is generally referred to as "circuit", "module" or "system" in this article. In addition, in some embodiments, the present invention can also be implemented in the form of a computer program product in one or more computer-readable media, and the computer-readable media contains computer-readable program code.

[0114] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A sugarcane yield intelligent prediction method, characterized in that: include: Construct a sample multimodal environmental perception matrix corresponding to each sample sugarcane planting area; The multimodal environmental perception matrix of each sample is analyzed separately to obtain multiple sample spatiotemporal fusion feature codes associated with sugarcane yield; Based on all the sample spatiotemporal fusion feature codes, the spatiotemporal fusion feature codes are used as the supervision signal of the lightweight student model, and the lightweight student model is guided by the feature mapping loss function to perform feature learning on the sample multimodal environment perception matrix to obtain a trained sugarcane yield prediction model; The trained sugarcane yield prediction model is used to predict the sugarcane yield in the target sugarcane planting area.

2. The method for intelligent prediction of sugarcane yield according to claim 1, characterized in that: The process of constructing a sample multimodal environment perception matrix includes: The vegetation index, hyperspectral data, soil data and meteorological data of any sample sugarcane planting area are spatiotemporally aligned to obtain a multimodal environmental perception matrix corresponding to the sample sugarcane planting area. The multimodal environmental perception matrix has three dimensions, namely, spatial dimension, temporal dimension and modal dimension. The modal dimension includes vegetation index, hyperspectral data, soil data and meteorological data.

3. The method for intelligent prediction of sugarcane yield according to claim 1, characterized in that: The multimodal environment perception matrix of each sample is analyzed separately, including: Through the trained dual-attention spatiotemporal graph convolutional network, the multimodal environmental perception matrix of each sample is analyzed separately.

4. The method for intelligent prediction of sugarcane yield according to claim 2, characterized in that: The vegetation index of any sample sugarcane planting area includes: the normalized difference vegetation index, enhanced vegetation index and surface moisture index of the sample sugarcane planting area at different times; the hyperspectral data of any sample sugarcane planting area includes: the spectral reflectance curve of sugarcane in the sample sugarcane planting area at different times, the texture parameters of sugarcane leaves and the time series change curve of sugarcane canopy greenness and moisture content; the meteorological data of any sample sugarcane planting area includes: the temperature, humidity, light intensity, wind speed and wind direction of the sample sugarcane planting area at different times.

5. A sugarcane yield intelligent prediction system, characterized in that: It includes matrix construction module, parsing module, training module and sugarcane yield prediction module; The matrix construction module is used to: construct a sample multimodal environmental perception matrix corresponding to each sample sugarcane planting area; The analysis module is used to: analyze each sample multimodal environment perception matrix respectively to obtain multiple sample spatiotemporal fusion feature codes associated with sugarcane yield; The training module is used to: based on all the sample spatiotemporal fusion feature codes, use the spatiotemporal fusion feature codes as the supervision signal of the lightweight student model, guide the lightweight student model to perform feature learning on the sample multimodal environment perception matrix through the feature mapping loss function, and obtain a trained sugarcane yield prediction model; The sugarcane yield prediction module is used to predict the sugarcane yield of the target sugarcane planting area using the trained sugarcane yield prediction model.

6. The sugarcane yield intelligent prediction system according to claim 5, characterized in that: The matrix building module is specifically used for: The vegetation index, hyperspectral data, soil data and meteorological data of any sample sugarcane planting area are spatiotemporally aligned to obtain a multimodal environmental perception matrix corresponding to the sample sugarcane planting area. The multimodal environmental perception matrix has three dimensions, namely, spatial dimension, temporal dimension and modal dimension. The modal dimension includes vegetation index, hyperspectral data, soil data and meteorological data.

7. The sugarcane yield intelligent prediction system according to claim 5, characterized in that: The parsing module is specifically used to parse each sample multimodal environment perception matrix separately through the trained dual attention spatiotemporal graph convolutional network.

8. The sugarcane yield intelligent prediction system according to claim 6, characterized in that: The vegetation index of any sample sugarcane planting area includes: the normalized difference vegetation index, enhanced vegetation index and surface moisture index of the sample sugarcane planting area at different times; the hyperspectral data of any sample sugarcane planting area includes: the spectral reflectance curve of sugarcane in the sample sugarcane planting area at different times, the texture parameters of sugarcane leaves and the time series change curve of sugarcane canopy greenness and moisture content; the meteorological data of any sample sugarcane planting area includes: the temperature, humidity, light intensity, wind speed and wind direction of the sample sugarcane planting area at different times.

9. An electronic device, characterized in that: The method comprises 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 intelligently predicting sugarcane yield according to any one of claims 1 to 4 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for intelligently predicting sugarcane yield according to any one of claims 1 to 4 is implemented.

Images