Predicting device and method
Patent Information
- Application Number
- PCT/JP2025/014671
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2025-04-14
- Publication Date
- 2026-10-22
Smart Images
Figure JP2025014671_22102026_PF_FP_ABST
Abstract
Description
Prediction Apparatus and Method
[0001] The present disclosure relates to a prediction apparatus and method for predicting the optimal harvest time.
[0002] Patent Document 1 describes performing statistics and analysis on remote sensing data obtained by remotely observing agricultural crops over a farm field, analyzing weather, temperature, humidity and the like from meteorological information data, and calculating the optimal harvest time based on these analysis results.
[0003] Japanese Unexamined Patent Publication No. 2023-83973
[0004] By the way, it is known that grain moisture serves as an index for determining the optimal harvest time for wheat. It is also known that grain moisture has a correlation with the vegetation index NDVI (Normalized Difference Vegetation Index) which is measured by utilizing the reflection characteristics of plants.
[0005] Accordingly, an object of the present disclosure is to provide a prediction apparatus and method for predicting the optimal harvest time using NDVI.
[0006] The prediction apparatus of the present disclosure includes: a reception unit that receives harvest association information including at least information related to a plot; an acquisition unit that acquires transition data indicating an NDVI transition based on the harvest association information; and an output unit that outputs a predicted harvest date of a harvest target in the plot using a harvest date prediction model based on the transition data and future weather prediction data.
[0007] According to the present disclosure, the optimal harvest time can be predicted using NDVI.
[0008] Figure 1 shows the storage of data captured from field N according to the present disclosure. Figure 2 shows the system configuration including the prediction device 100 according to the present disclosure. Figure 3 is a block diagram showing the functional configuration of the prediction device 100 and learning device 150 according to the present disclosure. Figure 4 shows specific examples of various databases. Figure 5 is a flowchart showing the operation of the learning device 150. Figure 6 is a diagram illustrating the transition of NDVI. Figure 7 is a graph showing the transition data of NDVI. Figure 8 is a flowchart showing the harvest date prediction process. Figure 9 shows an example of the hardware configuration of the prediction device 100 and learning device 150 according to one embodiment of the present disclosure.
[0009] Embodiments of this disclosure will be described with reference to the attached drawings. Where possible, the same parts will be denoted by the same reference numerals, and redundant descriptions will be omitted.
[0010] Figure 1 shows the storage of data captured by photographing field N in this disclosure. As shown in the figure, drone D takes aerial photographs of field N using, for example, a multispectral camera. The photographic data obtained from the aerial photography is sensing data that includes the near-infrared band in addition to ordinary RGB images. This photographic data is stored in the sensing data storage device 200. In this disclosure, field data is acquired in order to determine the harvest date (harvest time) of the crops in field N. In this disclosure, the harvest date is predicted, but the harvest time may be given with a certain degree of flexibility. In this disclosure, the photographic data is taken by drone D, but of course it may be obtained by other means. For example, an aircraft, helicopter, or satellite may photograph field N on the ground.
[0011] Figure 2 shows a system configuration including the prediction device 100 of this disclosure. As shown in the figure, the system comprises the prediction device 100, a learning device 150, a sensing data storage device 200, a farmland plot data storage device 210, and a weather data storage device 300.
[0012] The sensing data storage device 200 stores sensing data (image data) acquired based on the drone D, as described above. The farmland plot data storage device 210 and the weather data storage device 300 store farmland plot data and weather data, respectively. The weather data includes past weather data and future weather forecast data.
[0013] The prediction device 100 in this disclosure accesses a sensing data storage device 200 and a weather data storage device 300 to acquire sensing data and weather data, respectively. Based on the sensing data and weather data, the prediction device 100 predicts the harvest date of crops in field N. Although wheat is assumed as the crop in this disclosure, it can of course be applied to other crops as well.
[0014] The learning device 150 is a device that learns the harvest date prediction model used for prediction in the prediction device 100. As described later, the learning device 150 learns the harvest date prediction model based on sensing data (NDVI trend data), future weather forecast data, and harvest date predictions. The prediction device 100 uses the learned harvest date prediction model to predict the harvest date.
[0015] Figure 3 is a block diagram showing the functional configuration of the prediction device 100 and the learning device 150 of this disclosure. As shown in Figure 3(a), the learning device 150 includes a learning unit 101 and a model storage unit 104. Also, as shown in Figure 3(b), the prediction device 100 is configured to include a prediction unit 105. The learning unit 101 consists of a learning data creation unit 102 and a model creation unit 103. The prediction unit 105 consists of a prediction data creation unit 106 and a prediction unit 107.
[0016] The learning data creation unit 102 is the part that creates learning data by referring to the data storage unit 108. The learning data is data that includes NDVI based on the imaging data (sensing data) of field N and the harvest dates of crops in field N. NDVI is calculated by the learning data creation unit 102. The harvest dates, which serve as training data, are calculated based on the changes in NDVI.
[0017] The model creation unit 103 is responsible for creating a harvest date prediction model using the following data as training data: crop / variety data, weather data (future weather forecast data and past weather data), farmland plot data, and NDVI trend data as explanatory variables, and the harvest date as the objective variable. This harvest date prediction model is a machine learning model constructed using known machine learning methods. In this disclosure, at least the NDVI trend data and future weather forecast data are used as explanatory variables, and machine learning is performed with the harvest date as the objective variable. Other elements are not mandatory. The trend data consists of multiple NDVI values and their respective measurement dates.
[0018] The model storage unit 104 is the part that stores the harvest date prediction model learned by the model creation unit 103.
[0019] The forecast data creation unit 106 is the part that creates forecast data by referring to the sensing data storage device 200. The forecast data includes crop and variety data, weather data (future weather forecast data and past weather data), farmland plot data, and NDVI trend data, but as mentioned above, NDVI trend data and future weather forecast data are the minimum required data.
[0020] The prediction unit 107 is the part that inputs the prediction data into the harvest date prediction model and outputs the harvest date that is the result of that model.
[0021] Figure 4 shows specific examples of various databases. These various databases are stored in the sensing data storage device 200, the farmland plot data storage device 210, and the weather data storage device 300.
[0022] Figure 4(a) shows a specific example of a cultivation record database. This cultivation record database is stored in the farmland plot data storage device 210. As shown in the figure, the cultivation record database stores the date and the work performed. The harvest date stored in the cultivation record database may be used as correct answer data for learning in the learning device 150.
[0023] Figure 4(b) shows a specific example of a weather data database. This weather data database is stored in the weather data storage device 300. As shown in the figure, the weather data database stores the date and precipitation. The weather data database may also store the weather conditions for that day (sunny, cloudy, rainy, etc.).
[0024] Figure 4(c) shows a specific example of a weather forecast database. This weather forecast database is stored in the meteorological data storage device 300. As shown in the figure, the weather forecast database stores the date and the probability of precipitation. The weather forecast database may also store the weather and its probability for that day.
[0025] Figure 4(d) shows a specific example of a farmland plot database. This farmland plot database is stored in the farmland plot data storage device 210. As shown in the figure, the farmland plot database stores the field ID and its coordinate set (field boundaries). The field ID is the identifier of the field. The coordinate set is information that indicates the boundaries of the field using coordinates.
[0026] Figure 4(e) shows a specific example of the crop / variety database. The crop / variety database is stored in the farmland plot data storage device 210. As shown in the figure, the crop / variety database is a database that shows the crops planted in each field and stores the field ID, classification, crop, and variety. The classification shows the major category of crops (grains, vegetables, fruits, etc.). The crop shows the type of crop. The variety shows the detailed type (genetic characteristics or cultivation characteristics) of the crop shown in the crop.
[0027] Figure 4(f) shows a specific example of the NDVI transition database. This NDVI transition database is data calculated from NDVI by the learning data creation unit 102 and is temporarily stored in the learning data creation unit 102. As shown in the figure, the NDVI transition database is a database that shows the NDVI transition for each field and stores the field ID, date, and NDVI (average, minimum, maximum).
[0028] Figure 5 is a flowchart illustrating the operation of the learning device 150. The learning data creation unit 102 acquires sensing data from the sensing data storage device 200 (S101). The learning data creation unit 102 also acquires field data (farmland plot data) from the farmland plot data storage device 210 (S102). The learning data creation unit 102 acquires sensing data corresponding to the field data (S103), then calculates NDVI from the sensing data corresponding to the field (S104), and calculates NDVI transition data (S105). Before, during, or in parallel with these processes, the learning data creation unit 102 determines the harvest date based on the NDVI transition (S106) and sets the harvest date (S107).
[0029] Furthermore, the learning data creation unit 102 acquires crop and variety data from the farmland plot data storage device 210 (S109). Also, the learning data creation unit 102 acquires weather data (future weather forecast data and past weather data) from the weather data storage device 300 (S110).
[0030] Then, the model creation unit 103 creates a harvest date prediction model for each field (plot) based on crop / variety data, weather data (future weather forecast data and past weather data), NDVI trend data, and harvest date as training data (S108). This creation process is performed using so-called known machine learning, with crop / variety data, weather data, and NDVI as explanatory variables and harvest date as the dependent variable. Once the model creation unit 103 has created the harvest date prediction model, it stores the harvest date prediction model in the model storage unit 104 (S111).
[0031] In this way, a harvest date prediction model can be created to predict the harvest date based on crop / variety data, weather data, and NDVI. In this disclosure, fields (farm plots) are identified based on field data, and a harvest date prediction model is created for each field. That is, in this disclosure, one model is created for each crop, but the harvest prediction model is created so that different inputs are input for each field N, and the output also changes for each field N.
[0032] Here, we will explain processes S104 to S107 in more detail. In process S104, the NDVI is calculated. NDVI is an index that represents the density of green vegetation by utilizing the reflectivity characteristics of plants.
[0033] Chlorophyll in plants absorbs visible light with wavelengths of 400-700 nm, and the reflectivity of light at these wavelengths decreases. Green wavelengths, which fall exactly in the middle of this range, are not absorbed because they are not required for photosynthesis, so their reflectivity is high, which is why plant leaves appear green. Chlorophyll is sensitive to heat, and it protects itself by reflecting a large amount of wavelengths in the infrared region, resulting in high reflectivity. By utilizing this, the NDVI can be calculated by calculating the reflectivity at each wavelength as follows, and areas with denser greenery can be made more noticeable.
[0034] NDVI is calculated using the near-infrared reflectance (NIR) and the visible red reflectance (R) using the following formula. In this disclosure, the near-infrared reflectance (NIR) and the visible red reflectance (R) are obtained from sensing data. For example, the sensing data is satellite imagery or drone imagery, and this data records the intensity of reflected light in a specific wavelength band (near-infrared and visible red) for each pixel. This reflected light intensity is then calibrated, and the reflectance is calculated. If the image data is obtained from an artificial satellite, atmospheric correction is also performed. NDVI = (NIR - R) / (NIR + R) In this disclosure, the daily trend of the NDVI calculated in this way is determined.
[0035] Figure 6 illustrates the changes in NDVI. Figures 6(a) to 6(d) show the image data (sensing data) captured by drone D. As described above, this image data records the intensity in the near-infrared and visible red regions. In this disclosure, the NDVI is calculated for the entire region indicated by the rectangle. Multiple NDVI values can be obtained within the farm plot of the target field, but they may be treated as statistical values such as the maximum / minimum / average / median at a given point in time. In this disclosure, the maximum / minimum / average is used. Figure 6(a) shows the image data from April 13, Figure 6(b) shows the image data from June 12, Figure 6(c) shows the image data from July 15, and Figure 6(d) shows the image data from August 4. In each case, the region x indicated by the rectangle is the target of NDVI.
[0036] Figure 6(e) shows a graph of the trend with the horizontal axis representing date and the vertical axis representing NDVI. Three graphs are shown in Figure 6(e), each representing the minimum, maximum, and average values of NDVI.
[0037] In this disclosure, the harvest date of July 27th is predicted based on the NDVI on July 15th and the NDVI on August 4th. In other words, in this disclosure, multiple image data from a single field (image data from the same field on different dates) are extracted from the sensing data, and the harvest date is determined by calculating the NDVI and comparing the time series data, thereby determining the correct harvest date.
[0038] For example, the harvest date can be determined based on the following formula. The date d^ at which the slope of the NDVI transition is minimized, as shown in the formula below where d is the date and Nd is the NDVI, is taken as the estimated harvest date and used as the ground truth data. 2Δd is the observation interval for NDVI, and Δd represents half the observation interval. Figure 7 is a graph illustrating the above formula and shows the trend data. As shown in the figure, the date on which the slope is smallest is d + Δd, and this day is determined to be the harvest day.
[0039] Furthermore, the harvest date is not limited to the above, and the following may also be used: If the NDVI value falls below a predetermined standard value, the date at which it falls may be set as the harvest date. Alternatively, the harvest date may be determined based on cultivation record data obtained separately from the farming support application. This cultivation record data should include the harvest date.
[0040] Next, the harvest date prediction process in the prediction device 100 will be described. Figure 8 is a flowchart of the harvest date prediction process. When the prediction data creation unit 106 receives harvest date inquiry information (harvest date prediction instruction) which includes at least information specifying the field, it acquires sensing data from the sensing data storage device 200 (S201). This harvest date inquiry information is received, for example, from an instruction from a user terminal (not shown) or based on an operation instruction from an operator of the prediction device 100 to the operation unit. In this disclosure, the prediction data creation unit 106 calculates the NDVI and acquires crop and variety data from the farmland plot data storage device 210. The prediction data creation unit 106 also acquires weather data from the weather data storage device 300.
[0041] The prediction unit 107 then inputs the crop / variety data, weather data, and NDVI trend data acquired by the prediction data creation unit 106 into the harvest date prediction model stored in the model storage unit 104 (S202), and obtains the harvest date which will be the output value (S203).
[0042] Through this process, the prediction device 100 can predict the harvest date based on sensing data such as NDVI.
[0043] In this disclosure, information such as fertilizer application data from cultivation records may be used for learning and prediction. Specifically, the sensing data storage device 200 stores information on fertilizer applied to crops (date and type of fertilizer), and the learning device 150 may use this as an explanatory variable to train a harvest date prediction model. Furthermore, by inputting fertilizer information along with the cultivation data during prediction, a more accurate harvest date can be output.
[0044] Weather data may handle two types: past performance from a reference date (past weather data) and future forecasts (future weather prediction data). These pieces of information are acquired via an API. However, for past weather data, sensors may be installed to handle pinpoint information in the farm field.
[0045] The harvest date prediction model may be composed of the following two models. For example, it may be a hybrid model that is divided into a model that predicts the harvest date from NDVI and a model that predicts from weather data, and integrates the respective results. In that case, the learning device 150 generates a first model that is machine-learned with NDVI as an explanatory variable and the harvest date as an objective variable. Further, the learning device 150 may generate a second model that is machine-learned with weather data as an explanatory variable and the harvest date as an objective variable.
[0046] At the time of prediction, the prediction device 100 calculates a harvest date based on the respective harvest dates from the first model and the second model. For example, it may be the average value of the respective harvest dates.
[0047] When considering cultivars, a model structure may be used in which parameters are set for each cultivar, the parameters are learned with data for each cultivar, and common parameters not dependent on the cultivar are learned with all data. The model is not limited to any method, such as a hierarchical Bayesian model or a neural network.
[0048] The harvest date prediction model may adopt a structure that internally estimates grain moisture content. In this case, the correlation coefficient between NDVI and grain moisture content may be adopted as the initial value of the parameter.
[0049] When there is a shortage of crop / cultivar data, a model for estimating the crop / cultivar from sensing data of farm field plots may be created using the small amount of crop / cultivar data as correct answer information, and the estimated crop / cultivar data may be treated as an input. That is, a model may be learned that inputs image data or the near-infrared region and visible red region recorded therein, and estimates the crop / cultivar from the transition of NDVI.
[0050] It is uncertain whether the harvest date estimated from sensing data or obtained from cultivation record data was truly the appropriate harvest date. Therefore, the model creation unit 103 may use a model structure that takes into account the reliability of the harvest date. Specifically, the learning device 150 may calculate an index representing reliability from the likelihood of estimation and the frequency of missing data in cultivation record data, and learn by weighting it for each field. Alternatively, the learning device 150 (model creation unit 103) may prepare a parameter representing reliability and learn by multiplying it by the field-specific term of the cost function.
[0051] The detailed process for calculating reliability will be explained. If the reliability of the data for field i is wi, then when training the harvest date prediction model, a loss function such as the following is used. L is the loss function when reliability is not considered, yi is the correct harvest date, and y^i is the predicted harvest date. The Loss in this loss function is information that indicates the evaluation of the harvest date prediction model. Since this Loss and harvest date prediction model exist for each crop, the parameters of the harvest date prediction model are modified based on the Loss. The parameters of the harvest date prediction model are modified by error backpropagation, a common learning method in deep learning, and also by the least squares method in multiple regression models. Of course, these parameters may also be modified using other methods.
[0052] The following patterns are also possible for calculating reliability wi: 1. When the harvest date from cultivation record data is used as the ground truth data, the reliability w_i is set to 1 / ΔT, where ΔT is the average recording interval of the cultivation record. 2. When the estimated harvest date is used as the ground truth data, the reliability wi is set to 1 / Δd, based on Δd used when calculating the estimated harvest date which is considered the ground truth (Δd is the observation interval). 3. An alternative example where reliability wi is applicable in both cases 1 and 2 is estimated as a parameter during learning.
[0053] Next, the effects of the prediction device 100 and learning device 150 of this disclosure will be described. In the prediction device 100 of this disclosure, the prediction data creation unit 106 functions as a receiving unit that receives harvest date inquiry information, which includes at least information about the field (plot). The prediction data creation unit 106 then acquires transition data showing the NDVI transition based on the harvest date inquiry information. That is, when the prediction data creation unit 106 receives the harvest date inquiry information, it accesses the sensing data storage device 200, the farmland plot data storage device 210, and the weather data storage device 300 to acquire sensing data, field data, and weather data, respectively. The prediction unit 107 then functions as an output unit and predicts (outputs) the predicted harvest date of crops in the field using a harvest date prediction model based on the NDVI transition data and future weather forecast data.
[0054] This configuration allows for the prediction of harvest dates based on NDVI and future weather forecast data. Using NDVI makes it easy and accurate to predict harvest dates.
[0055] Furthermore, in this disclosure, the harvest date prediction model is generated by machine learning using a learning device 150. This learning device 150 uses NDVI trend data and weather forecast data obtained from multiple training sensing data of a field (plot) containing agricultural crops (objects to be harvested) as explanatory variables, and the harvest date data of the training agricultural crops as the objective variable, to generate a harvest date prediction model by machine learning.
[0056] For example, this sensing data consists of multiple sensing data points taken at different times and locations, capturing a field containing crops, and the NDVI trend data is derived from these multiple sensing data points.
[0057] Furthermore, in the learning device 150, the learning data creation unit 102 functions as a plot-related data acquisition unit and acquires at least one of the following as field-related data: field-related data (regional characteristics) and crop / variety data. The model creation unit 103 then uses the field-related data to generate a harvest date prediction model. Generally, different regions have different characteristics such as double cropping areas, soil types, and cultivation methods. Regional characteristics refer to information that indicates these characteristics.
[0058] This configuration allows for the generation of harvest date prediction models based on NDVI trend data and weather forecast data, as well as regional characteristics and crop / variety data, enabling more accurate predictions.
[0059] Furthermore, the learning device 150 generates a harvest date prediction model by adding past weather data in addition to future weather forecast data.
[0060] Weather data, such as sunny and rainy days, significantly impacts the harvest date. Therefore, generating a harvest date prediction model that takes this information into account enables highly accurate predictions.
[0061] In this disclosure, harvest date data is identified as the ground truth using training sensing data. This harvest date data is identified by performing a time-series comparison of the sensing data. For example, NDVI is determined for each measurement day, and the harvest date data is identified by the measurement day on which the rate of decrease is minimized, based on the NDVI trend data. Generally, it is believed that the NDVI decreases when crops are harvested.
[0062] Furthermore, in this disclosure, the learning device 150 learns using a loss function during the training of the harvest date prediction model, and evaluates the prediction performance of the harvest date prediction model based on the loss function. Based on this evaluation (value), the predicted harvest date can be weighted to enable adjustment of the harvest date.
[0063] The prediction device and prediction method described herein have the following configurations.
[0064] [1] A prediction device comprising: a receiving unit that receives harvest date inquiry information including at least information about a plot; an acquisition unit that acquires transition data showing the transition of NDVI based on the harvest date inquiry information; and an output unit that outputs a predicted harvest date for the harvest target in the plot using a harvest date prediction model based on the transition data and future weather forecast data.
[0065] [2] The prediction device according to [1], wherein the harvest date prediction model is generated by machine learning using a model generation device, and the model generation device generates the harvest date prediction model by machine learning using NDVI transition data and weather forecast data obtained from multiple training sensing data of a field including the harvest target as explanatory variables, and the harvest date data of the harvest target for training as the objective variable.
[0066] [3] The sensing data is a plurality of sensing data taken at different times and locations of the plot containing the harvested material, and the NDVI transition data is obtained from the plurality of sensing data, as described in [2].
[0067] [4] The prediction device according to any one of [2] to [3], wherein the model generation device further comprises a plot-related data acquisition unit that acquires at least one of plot-related data and crop / variety data as plot-related data, and the model generation unit further generates the harvest date prediction model using the plot-related data.
[0068] [5] The prediction device according to [4], wherein the model generation device generates a harvest date prediction model by adding past weather data in addition to the future weather forecast data.
[0069] [6] The prediction device according to any one of [2] to [4], wherein the harvest date data is identified as the correct value using the sensing data for training.
[0070] [7] The harvest date data is identified by performing a time-series comparison of the sensing data, as described in [6].
[0071] [8] The prediction device according to [7], wherein the NDVI is determined for each measurement day, and the harvest date data is identified by the measurement day on which the rate of decrease is minimized, based on the trend data of the NDVI.
[0072] [9] The prediction device according to any one of [2] to [8], wherein the model generation device learns the harvest date prediction model using a loss function during training and evaluates the prediction performance of the harvest date prediction model based on the loss function.
[0073]
[10] A method for a prediction device comprising: a receiving step of receiving harvest date inquiry information which includes at least information about a plot; an acquisition step of acquiring transition data which shows the transition of NDVI based on the harvest date inquiry information; and an output step of outputting a harvest date prediction for the harvest target in the plot using a harvest date prediction model based on the transition data and future weather forecast data.
[0074] The block diagram used in the description of the above embodiment shows functional units. These functional blocks (components) are realized by any combination of at least one of hardware and software. Furthermore, the method of realizing each functional block is not particularly limited. That is, each functional block may be realized using one device that is physically or logically coupled, or it may be realized using two or more physically or logically separated devices that are directly or indirectly connected (for example, using wired or wireless connections). A functional block may be realized by combining the one or more devices with software.
[0075] Functions include, but are not limited to, judgment, decision, determination, calculation, calculation, processing, derivation, investigation, exploration, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, assumption, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating (mapping), and assigning. For example, a functional block (configuration part) that enables transmission is called a transmitting unit or transmitter. In all cases, as mentioned above, the method of implementation is not particularly limited.
[0076] For example, the prediction device 100 and learning device 150 in one embodiment of the present disclosure may function as a computer that processes the prediction method and learning method of the present disclosure. Figure 9 is a diagram showing an example of the hardware configuration of the prediction device 100 and learning device 150 according to one embodiment of the present disclosure. The prediction device 100 and learning device 150 described above may be physically configured as a computer device including a processor 1001, memory 1002, storage 1003, communication device 1004, input device 1005, output device 1006, bus 1007, etc.
[0077] In the following explanation, the term "device" can be replaced with "circuit," "device," "unit," etc. The hardware configuration of the prediction device 100 and the learning device 150 may include one or more of the devices shown in the figure, or it may be configured to omit some of the devices.
[0078] Each function in the prediction device 100 and the learning device 150 is realized by loading predetermined software (programs) onto hardware such as the processor 1001 and memory 1002, which causes the processor 1001 to perform calculations, control communication by the communication device 1004, and control at least one of the reading and writing of data in the memory 1002 and storage 1003.
[0079] The processor 1001 controls the entire computer, for example, by running the operating system. The processor 1001 may be composed of a central processing unit (CPU) that includes interfaces with peripheral devices, control devices, arithmetic units, registers, etc. For example, the above-mentioned training data creation unit 102, model creation unit 103, prediction data creation unit 106, prediction unit 107, etc., may be implemented by the processor 1001.
[0080] Furthermore, the processor 1001 reads programs (program code), software modules, data, etc., from at least one of the storage 1003 and the communication device 1004 into the memory 1002 and executes various processes accordingly. The program used is one that causes the computer to execute at least a part of the operations described in the above embodiment. For example, the learning data creation unit 102, the model creation unit 103, the prediction data creation unit 106, and the prediction unit 107 may be implemented by a control program stored in the memory 1002 and running on the processor 1001, and other functional blocks may be implemented similarly. The above-described various processes have been explained as being executed by one processor 1001, but they may be executed simultaneously or sequentially by two or more processors 1001. The processor 1001 may be implemented by one or more chips. The program may also be transmitted from a network via a telecommunications line.
[0081] The memory 1002 is a computer-readable recording medium and may consist of at least one of the following: ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), RAM (Random Access Memory), etc. The memory 1002 may also be called a register, cache, main memory, etc. The memory 1002 can store executable programs (program code), software modules, etc., for carrying out the prediction method and learning method according to one embodiment of the present disclosure.
[0082] The storage 1003 is a computer-readable recording medium and may consist of at least one of the following: an optical disc such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (e.g., a compact disc, a digital multipurpose disc, a Blu-ray® disc), a smart card, flash memory (e.g., a card, a stick, a key drive), a floppy® disk, a magnetic strip, etc. The storage 1003 may also be called an auxiliary storage device. The above-mentioned storage medium may be, for example, a database, server, or other suitable medium including at least one of memory 1002 and storage 1003.
[0083] The communication device 1004 is hardware (transmitting / receiving device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as a network device, network controller, network card, communication module, etc. The communication device 1004 may be configured to include high-frequency switches, duplexers, filters, frequency synthesizers, etc., in order to implement at least one of frequency division duplex (FDD) and time division duplex (TDD). For example, the acquisition of sensing data may be implemented by the communication device 1004. The communication device 1004 may be implemented with physically or logically separated transmitting and receiving units.
[0084] The input device 1005 is an input device that accepts input from an external source (e.g., a keyboard, mouse, microphone, switch, button, sensor, etc.). The output device 1006 is an output device that outputs to an external source (e.g., a display, speaker, LED lamp, etc.). The input device 1005 and the output device 1006 may be configured as an integrated unit (e.g., a touch panel).
[0085] Furthermore, each device, such as the processor 1001 and the memory 1002, is connected by a bus 1007 for communicating information. The bus 1007 may be configured using a single bus, or different buses may be configured for each device.
[0086] Furthermore, the prediction device 100 and the learning device 150 may be configured to include hardware such as a microprocessor, a digital signal processor (DSP), an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), and an FPGA (Field Programmable Gate Array), and some or all of each functional block may be realized by such hardware. For example, the processor 1001 may be implemented using at least one of these hardware components.
[0087] Information notification is not limited to the embodiments described herein and may be carried out by other means. For example, information notification may be carried out by physical layer signaling (e.g., DCI (Downlink Control Information), UCI (Uplink Control Information)), upper layer signaling (e.g., RRC (Radio Resource Control) signaling, MAC (Medium Access Control) signaling, broadcast information (MIB (Master Information Block), SIB (System Information Block))), other signals, or combinations thereof. RRC signaling may also be called RRC messages, and may be, for example, RRC Connection Setup messages, RRC Connection Reconfiguration messages, etc.
[0088] The processing procedures, sequences, flowcharts, etc., of each aspect / embodiment described in this disclosure may be reordered, provided they do not contradict each other. For example, the methods described in this disclosure present various step elements using exemplary order and are not limited to the specific order presented.
[0089] Input and output information may be stored in a specific location (e.g., memory) or managed using a management table. Input and output information may be overwritten, updated, or appended to. Output information may be deleted. Input information may be transmitted to other devices.
[0090] The determination may be made by a value represented by one bit (0 or 1), by a boolean value (true or false), or by a numerical comparison (for example, a comparison with a predetermined value).
[0091] Each aspect / embodiment described in this disclosure may be used individually, in combination, or switched between as needed during implementation. Furthermore, notification of specific information (e.g., notification that "X is") is not limited to explicit notification, but may also be implicit (e.g., by not providing such notification).
[0092] Although the present disclosure has been described in detail above, it will be clear to those skilled in the art that the present disclosure is not limited to the embodiments described herein. The present disclosure can be implemented in modified and altered forms without departing from the intent and scope of the present disclosure as defined by the claims. Accordingly, the descriptions in the present disclosure are illustrative and not intended to be restrictive in any way.
[0093] Software should be broadly interpreted to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, execution threads, procedures, functions, and so on, whether they are called software, firmware, middleware, microcode, hardware description languages, or by any other name.
[0094] Furthermore, software, instructions, information, etc., may be transmitted and received via a transmission medium. For example, if software is transmitted from a website, server, or other remote source using at least one of wired technologies (such as coaxial cable, fiber optic cable, twisted pair, or digital subscriber line (DSL)) and wireless technologies (such as infrared or microwave), then at least one of these wired and wireless technologies is included in the definition of a transmission medium.
[0095] The information, signals, etc. described in this disclosure may be represented using any of the various different techniques. For example, the data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.
[0096] In addition, terms used in this disclosure and terms necessary for understanding this disclosure may be replaced with terms having the same or similar meanings. For example, at least one of the channel and symbol may be a signal (signaling). Also, a signal may be a message. Furthermore, a component carrier (CC) may be called a carrier frequency, cell, frequency carrier, etc.
[0097] Furthermore, the information, parameters, etc., described in this disclosure may be expressed using absolute values, relative values from a given value, or other corresponding information. For example, wireless resources may be indicated by an index.
[0098] The names used for the parameters described above are not restrictive in any way. Furthermore, the formulas and other expressions using these parameters may differ from those expressly disclosed in this disclosure. Various channels (e.g., PUCCH, PDCCH, etc.) and information elements can be identified by any suitable name, and therefore, the various names assigned to these various channels and information elements are not restrictive in any way.
[0099] In this disclosure, terms such as "Mobile Station (MS)," "user terminal," "User Equipment (UE)," and "terminal" may be used interchangeably.
[0100] A mobile station may also be referred to by those skilled in the art as a subscriber station, mobile unit, subscriber unit, wireless unit, remote unit, mobile device, wireless device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, handset, user agent, mobile client, client, or some other appropriate term.
[0101] As used in this disclosure, the terms “determining” and “determining” may encompass a wide variety of actions. “Determining” may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, searching, or inquiring (e.g., searching in a table, database, or other data structure), or ascertaining. “Determining” may also include, for example, receiving (e.g., receiving information), transmitting (e.g., sending information), inputting, outputting, or accessing (e.g., accessing data in memory). Furthermore, "judgment" and "decision" can include considering something as having been "judged" or "decided" after resolving, selecting, choosing, establishing, comparing, etc. In other words, "judgment" and "decision" can include considering something as having been "judged" or "decided" after some action. Also, "judgment (decision)" can be reinterpreted as "assuming," "expecting," or "considering."
[0102] The terms “connected,” “coupled,” or any variation thereof, mean any direct or indirect connection or coupling between two or more elements, and may include the presence of one or more intermediate elements between two elements that are “connected” or “coupled” with each other. The coupling or connection between elements may be physical, logical, or a combination thereof. For example, “connection” may be reinterpreted as “access.” As used in this disclosure, two elements may be considered to be “connected” or “coupled” with each other using at least one of one or more wires, cables, and printed electrical connections, and, in some non-limiting and non-exclusive examples, electromagnetic energy having wavelengths in the radio frequency domain, microwave domain, and optical (both visible and invisible) domain.
[0103] In this disclosure, the phrase "based on" does not mean "based solely on" unless otherwise specified. In other words, the phrase "based on" means both "based solely on" and "based at least on."
[0104] Any reference to elements using designations such as “first,” “second,” etc., as used in this disclosure does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient way to distinguish between two or more elements. Accordingly, references to first and second elements do not imply that only two elements may be employed, or that the first element must precede the second element in any way.
[0105] Where the terms “include,” “including,” and their variations are used in this disclosure, these terms are intended to be inclusive, as is the term “comprising.” Furthermore, the term “or” as used in this disclosure is not intended to be exclusive OR.
[0106] In this disclosure, if articles are added by translation, such as a, an, and the in English, this disclosure may include the fact that the noun following these articles is plural.
[0107] In this disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "combine" may be interpreted similarly to "different."
[0108] 100...Prediction device, 150...Learning device, 200...Sensing data storage device, 210...Agricultural land plot data storage device, 300...Weather data storage device, 101...Learning unit, 102...Learning data creation unit, 103...Model creation unit, 104...Model storage unit, 105...Prediction unit, 106...Prediction data creation unit, 107...Prediction unit, 108...Data storage unit.
Claims
A reception desk that accepts harvest date inquiry information, which includes at least information about the plot, An acquisition unit that acquires trend data showing the NDVI trend based on the harvest date inquiry information, An output unit that outputs a predicted harvest date for the crops to be harvested in the plot using a harvest date prediction model based on the aforementioned trend data and future weather forecast data. A prediction device equipped with the following features. The aforementioned harvest date prediction model, The model is generated by machine learning using a model generation device. The aforementioned model generation device is The harvest date prediction model is generated by machine learning, using NDVI trend data and weather forecast data obtained from multiple training sensing data of a field containing the crop to be harvested as explanatory variables, and the harvest date data of the crop to be harvested for training as the dependent variable. The prediction device according to claim 1. The sensing data consists of multiple sensing data taken at different times, capturing the plot containing the harvested material. The NDVI transition data is obtained from the plurality of sensing data. The prediction device according to claim 2. The aforementioned model generation device is The system further includes a plot-related data acquisition unit that acquires at least one of the plot-related data and crop / variety data as plot-related data. The model generation device further generates the harvest date prediction model using the plot-related data. The prediction device according to claim 2. The aforementioned model generation device is In addition to the aforementioned future weather forecast data, historical weather data is added to generate a harvest date prediction model. The prediction device according to claim 4. The aforementioned harvest date data is The correct values are identified using the aforementioned sensing data for training. The prediction device according to claim 2. The harvest date data is identified by performing a time-series comparison of the sensing data. The prediction device according to claim 6. NDVI is calculated for each measurement day. The harvest date data is identified by the measurement date on which the rate of decrease is minimized, based on the NDVI trend data. The prediction device according to claim 7. The aforementioned model generation device is During the training of the harvest date prediction model, a loss function is used for training, and the prediction performance of the harvest date prediction model is evaluated based on the loss function. Based on the above evaluation, the harvest forecast date is determined. The prediction device according to claim 2. In the method of the prediction device, A reception step that accepts harvest date inquiry information, which includes at least information about the plot, The acquisition step involves obtaining trend data showing the NDVI trend based on the harvest date inquiry information, An output step which outputs a predicted harvest date for the crops to be harvested in the plot using a harvest date prediction model based on the aforementioned trend data and future weather forecast data, A method for providing this.