KR101811640B1 - Prediction apparatus and method for production of crop using machine learning - Google Patents

Abstract

The present invention discloses an apparatus and method for predicting the production of crop using machine learning. The apparatus for predicting the production of crop using machine learning according to the present invention includes a communication part for receiving farmhouse data including at least one among the growth information, environmental information, and pest and disease information of a crop, and a control part for generating a growth model for predicting the growth of the crop by the machine learning of the farmhouse data received from the communication part and predicting the production of the crop by using the generated growth model.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for predicting crop production using machine learning,

More particularly, the present invention relates to an apparatus and method for predicting the yield of a crop using machine learning that predicts the yield of a crop using the growth information of the crop.

Recently, the production of crops has been increasing due to the activation of smart agriculture and facility horticulture facilities. Therefore, a device for predicting the yield of crops applicable in the domestic environment is being researched and developed in various fields.

On the other hand, the conventional production forecasting apparatus has a problem that the prediction result can not be relied on because it uses factors obtained from insufficient experiments, insufficient data, and biochemical theory as the input factors in a confined environment.

Therefore, there is a need for an apparatus for predicting the yield of crops which can be applied to the domestic environment while overcoming these problems.

Korean Patent Publication No. 10-2013-0019051 (Feb.

An object of the present invention is to provide an apparatus and method for predicting the yield of a crop using machine learning in which a growth model of a crop is generated using farmhouse data and a production amount of the crop is predicted using the generated growth model have.

It is another object of the present invention to provide an apparatus and method for predicting crop yield using machine learning, in which the reliability of the result of predicting the yield of crops is increased as the farmhouse data is accumulated.

In order to achieve the above object, an apparatus for predicting the yield of a crop using machine learning according to the present invention comprises a communication unit for receiving farmhouse data including at least one of growth information, environmental information and pest damage information of a crop, And a controller for generating a growth model for predicting the growth of the crop by using machine learning of the received farmhouse data and for predicting the yield of the crop using the generated growth model.

The control unit may collect the farmhouse data in accordance with the data standardization format, and filter the collected farmhouse data before generating the farming model of the crop.

The control unit may further include a data collecting unit for classifying the received farmhouse data by the same crop and species, classifying the classified farmhouse data by factors and collecting the classified farmhouse data according to the data standardization format, A data analysis unit for filtering the farmhouse data with data error and the farmhouse data whose data quality is lower than the reference value and filtering the filtered farmhouse data by performing a correlation analysis and a regression analysis on the filtered farmhouse data by the machine learning, A growth model generating unit for generating a growth model, and a production amount predicting unit for predicting a production amount with respect to the fresh weight of the crop using the generated growth model.

The data collecting unit may sort the collected farmhouse data by time zone to make a database.

In addition, the growth model generation unit may include a generation module for generating the growth model using the database farmed farm data, a method for analyzing the growth state of a current crop by comparing the filtered farmhouse data with the growth model, A prediction module for predicting a growth environment having the highest production yield of the crop using the state and a farmhouse data cluster having the highest production yield and quality of the crop among the filtered farmhouse data using the predicted growth environment And a learning module for providing the generated farmhouse data set to the generation module to update the farmhouse data.

The growth model generation unit may perform the machine learning by sequentially repeating the generation module, the prediction module, and the learning module.

The generation module may generate a plurality of patterns by combining different factors among the filtered farmhouse data, and generate a specific pattern of a specific crop and species among the generated plurality of patterns by performing the correlation analysis and the regression Extracting the specific pattern, and generating the growth model using the extracted specific pattern.

Also, the data standardization format is characterized in that at least one factor of the facility and cultivation information, the growth information, the environmental information, and the farmhouse other information are classified according to a predetermined standard data standard.

In addition, the growth information of the crop includes at least one of a production amount, a light weight, a total leaf area, a leaf area index (LAI), a leaf temperature, a stem thickness, a flower bed spacing, a growth rate, a growth intensity and a growth phase, The information includes at least one of a type of a greenhouse, an environmental facility, an operating temperature, a humidity, an instantaneous light amount, a carbon dioxide concentration, a medium water content, a medium EC and operational information. The pest- And at least one factor.

The method for predicting yield of a crop using machine learning according to the present invention includes the steps of receiving farm data including at least one of crop growth information, environmental information and pest damage information, Generating a growth model for predicting the growth of the crop, and estimating the yield of the crop using the generated growth model.

An apparatus and method for predicting crop production using machine learning according to the present invention can generate a crop growth model using farm data and predict the crop production using the generated growth model.

It is also possible to increase the reliability of the results of the prediction of crop yield by accumulating farm data.

FIG. 1 is a block diagram for explaining a cultivation facility control system according to an embodiment of the present invention.
2 is a block diagram for explaining a production forecasting apparatus according to an embodiment of the present invention.
3 is a block diagram for explaining the control unit of FIG.
FIG. 4 is a block diagram for explaining a growth model generation unit of FIG. 3. FIG.
FIG. 5 is a flowchart illustrating a method of estimating a production amount according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the drawings, the same reference numerals as used in the appended drawings denote like elements, unless indicated otherwise. In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather obvious or understandable to those skilled in the art.

FIG. 1 is a block diagram for explaining a cultivation facility control system according to an embodiment of the present invention.

Referring to FIG. 1, a cultivation facility control system 300 controls environmental changes of the cultivation facility 200, such as environmental conditions, crop growth, etc., to maintain environmental conditions optimized for crop growth. Here, the cultivation facility 200 may be a smart farm. The cultivation facility control system 300 includes a production amount prediction apparatus 100 and a cultivation facility 200. Here, the crops may be vegetables such as tomatoes, paprika, potatoes, peppers, tobacco, mushrooms, or fruits such as apples, pears, grapes, mandarins, persimmons, and bananas, or flowers that are all cultivated for viewing.

The production forecasting apparatus 100 predicts the production amount of crops cultivated in the cultivation facility 200. At this time, the production forecasting apparatus 100 generates a crop growth model and predicts the crop production amount by using the generated growth model, so that the user can confirm the production amount of the crop in advance.

The cultivation facility 200 is a farm that measures, analyzes and analyzes the temperature, humidity, light amount, carbon dioxide, and soil of the crop cultivation facility using the Internet of Thing (IoT) technology and maintains the optimum state according to the analysis result . To do this, the cultivation facility 200 includes a farm 210, a sensor unit 230, an actuator 250, and a farm control unit 270.

Farm 210 is a space where crops grow. Farms 210 may grow crops such as crops or flowers, preferably greenhouses.

The sensor unit 230 includes a plurality of sensors, and the sensors are installed at the farm 210 at regular intervals. At this time, the sensor unit 230 includes various kinds of sensors such as a temperature sensor, a humidity sensor, a carbon dioxide sensor, a light sensor, a soil moisture sensor, a leaf temperature sensor, Measure the status of the insect pests. The temperature sensor measures the temperature in the farm 210. The humidity sensor measures the humidity in the farm (210). The carbon dioxide sensor measures the amount of carbon dioxide in the farm 210. The light sensor measures the amount of sunshine within the farm 210. The soil moisture sensor measures the moisture contained in the soil within the farm (210). The leaf temperature sensor measures the temperature of the leaf of the crop in the farm (210).

Actuator 250 maintains the environmental conditions of farm 210. The actuator 250 is disposed within the farm 210 and creates the environmental conditions necessary for the crop to grow. The actuator 250 includes a plurality of actuators, and is driven by using electric, hydraulic, compressed air, or the like. For example, the actuator 250 may be an air conditioner, a fan unit, a sprinkler, a feed pump, a carbon dioxide generator, a lighting device, or an opening / closing device. The heating / cooling device controls the temperature in the farm 210. The fan arrangement regulates air circulation and air ventilation within the farm 210. The sprinkler and feed pump regulate the water supply within the farm (210). The carbon dioxide generator controls the concentration of carbon dioxide. The lighting device and the switchgear control the lighting within the farm (210). Here, the illuminating device can be configured as an LED illuminating device to illuminate the illumination including the spectrum required for the crop, and the opening / closing device can adjust the amount of sunshine by opening and closing the roof of the farm 210.

The farm control unit 270 transmits the farm data including the information of the farm 210 to the production forecasting apparatus 100 using the measurement information measured by the sensor unit 230. Accordingly, the farm control device 270 may include a communication module.

2 is a block diagram for explaining a production forecasting apparatus according to an embodiment of the present invention.

Referring to FIGS. 1 and 2, the production forecasting apparatus 100 generates a crop growth model using the farm data and predicts a crop production amount using the generated growth model. As the yield prediction apparatus 100 accumulates the farm data, the reliability of the result of prediction of the yield of the crop is improved. The yield prediction apparatus 100 may be a computer system such as a desktop, a laptop, a server computer, a cluster computer, or the like.

The production amount prediction apparatus 100 includes a communication unit 10, a control unit 20, an output unit 30, an input unit 40, and a storage unit 50.

The communication unit 10 receives farmhouse data including at least one of crop growth information, environmental information, and pest damage information. At this time, the communication unit 10 can receive farmhouse data remotely using wired / wireless communication. Herein, the growth information of the crop includes at least one of the production amount, light weight, general diameter, leaf area index (LAI), leaf temperature, stem thickness, flower height, growth rate, growth intensity and growth phase. The environmental information includes at least one of a type of a greenhouse, environmental facilities, operating temperature, humidity, instantaneous light intensity, carbon dioxide concentration, medium water content, medium EC and operational information. The pest insult information includes at least one of a pest-insect occurrence frequency and a plant physiological disorder information.

The communication unit 10 can transmit the simulated data in the crop production amount and the growth model predicted by the control unit 20 to the user terminal (not shown) and provide the user with the simulated data. This allows the user to view data about the crop at any location.

The control unit 20 collects the farmhouse data received by the communication unit 10 and the farmhouse data inputted from the input unit 40 to be described later according to the data standardization format. The control unit 20 can convert the collected farmhouse data into a database.

Here, the data standardization format is a standard data standard defined in advance, and it is a predetermined standard data standard, and can be used for various kinds of facilities such as greenhouses, vinyl houses, ventilating and heating facilities and cultivation information, growth information such as leaf area, leaf number, yield, growth rate, Environmental information such as rhizosphere environment, and other information such as foliage, pesticide use, fertilizer use, etc., are classified according to the standard. At this time, the control unit 20 can adjust the factor of the data standardization format according to the kind of the crop.

The control unit 20 selects and filters the collected farmhouse data. The control unit 20 generates a growth model for predicting the growth of the crop by machine learning the filtered farmhouse data. Here, the growth model can be predicted by simulating the process in which the crop grows according to the farmhouse data. The control unit 20 predicts the crop production amount using the generated growth model.

The output unit 30 is a display device including a monitor, a liquid crystal, a projector, and a printer. The output unit 30 outputs the farmhouse data received by the communication unit 10 and the farmhouse data inputted from the input unit 40. The output unit 30 outputs the amount of crop production predicted by the control unit 20. Further, the output unit 30 outputs the simulated data in the growth model of the control unit 20. Here, the output unit 30 may include a touch screen function, and when the touch screen function is included, the output unit 30 may serve as the input unit 40.

The input unit 40 receives the user input of the user. At this time, the user input may be farmhouse data. That is, the input unit 40 directly receives the farmhouse data, unlike the communication unit 10 receiving the farmhouse data via the communication network.

The storage unit 50 stores the farmhouse data received by the communication unit 10 and the farmhouse data input by the input unit 40. [ The storage unit 50 stores various analysis algorithms performed by the control unit 20, and stores analysis results analyzed through an analysis algorithm. The storage unit 50 stores the production amount of the crop estimated by the control unit 20. Here, the storage unit 50 may store the above-described data in a database, and may be a flash memory type, a hard disk type, a multimedia card micro type, Type memory (e.g., SD or XD memory), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read- ), A programmable read-only memory (PROM), a magnetic memory, a magnetic disk, and an optical disk.

FIG. 3 is a block diagram for explaining the control unit of FIG. 2, and FIG. 4 is a block diagram for explaining a growth model generation unit of FIG.

2 to 4, the control unit 20 includes a data collecting unit 21, a data analyzing unit 23, a growth model generating unit 25, and a production amount estimating unit 27.

The data collecting unit 21 collects the farmhouse data necessary for the initial operation. The data collection unit 21 can collect the farmhouse data received by the communication unit 10 by region and country. The data collecting unit 21 classifies the farmhouse data received by the communication unit 10 and the farmhouse data inputted from the input unit 40 by the same crop and species. The data collecting unit 21 classifies the classified farmhouse data by the factor and collects it according to the data standardization format. At this time, the data collecting unit 21 may sort the collected farmhouse data by time, such as time, day, and month, and may sort the data by time, preferably by time.

That is, the data collecting unit 21 collects the farmhouse data according to the time order, and converts the data into a database, thereby providing high quality farmhouse data having homogeneity and uniformity to the user.

The data analyzing unit 23 analyzes the farmhouse data collected by the data collecting unit 21 and selects the farmhouse data with the data error and the farmhouse data whose data quality is below the reference value. Here, the farm data whose data quality is less than the reference value may be erroneous data whose farm data exceeds the reference value range expected. The data analysis unit 23 filters the selected farmhouse data. In this way, the data analysis unit 23 can generate a growth model using only the farmhouse data with reliability.

The growth model generation unit 25 generates a growth model by machine learning the farmhouse data filtered by the data analysis unit 23. As a result, the growth model generation unit 25 can accumulate data on the learned experience over time and generate a growth model that accurately predicts the growth of the crop. The growth model generation unit 25 includes a generation module 61, a prediction module 63, and a learning module 65.

The generation module 61 generates a growth model flexibly. That is, at the beginning of the operation, since the farmhouse data provided to the generation module 61 is insufficient, the generation module 61 generates the farming model using the farmhouse data determined in advance or using the database farmhouse data, When the learned farmhouse data is updated in step 65, a new growth model can be generated using the learned farmhouse data.

The prediction module 63 compares the filtered farmhouse data with the farming model generated by the generation module 61 to analyze the present farming state. The prediction module 63 predicts the growth environment in which the crop yield is highest using the analyzed growth state.

The learning module 65 functions to process the farmhouse data and learn it. That is, the learning module 65 calculates a farmhouse data set cluster having the highest production quantity and quality among the farmhouse farming data filtered using the growth environment predicted by the prediction module 63. [ Here, the learning module 63 can re-filter below the threshold value of the current farmhouse data by using the growth model generated by the generation module 61 to select only the meaningful farmhouse data, and then calculate the farmhouse data set. The module 65 learns the calculated farmhouse data set and then provides the farmhouse data set learned to the generating module 61 to update the farmhouse data. In this way, the generation module 61 can apply the updated farmhouse data to the current farm model to generate a new farm model with a better predictability than the current farm model. Here, the farmhouse data learned in the learning module 65 may be updated in the storage section 50 and converted into a database.

That is, the growth model generation unit 25 performs machine learning by sequentially performing the generation module 61, the prediction module 63, and the learning module 65 in sequence. Thereby, the growth model generation unit 25 can gradually generate a growth model in which the production amount prediction of the crop is correct.

For example, the growth model generation unit 25 can perform various types of big data analysis by machine learning, and can preferably perform correlation analysis and regression analysis. That is, the generation module 61 of the growth model generation unit 25 can generate a plurality of patterns by combining different factors among the farm data and perform correlation analysis and regression analysis using the generated plurality of patterns have. The generation module 61 extracts specific patterns of specific crops and species by repeated correlation analysis and regression analysis, and generates a growth model using the extracted specific patterns. Here, the specific pattern may be a growth pattern in which the quality status of the crop is good among the plurality of patterns and the production amount is predicted to the maximum.

The production amount predicting unit 27 predicts the production amount of the crop using the growth model generated by the growth model generating unit 25. [ At this time, the production amount predicting unit 27 can predict the production amount with respect to the fresh weight of the crop. Here, the fresh weight is a weight measured in a non-dried biological state. Thus, the user can predict the yield of the actual crop rather than the dried crop.

FIG. 5 is a flowchart illustrating a method of estimating a production amount according to an embodiment of the present invention.

Referring to FIGS. 2 and 5, the method of predicting yield can generate a crop growth model using the farm data and predict the crop production using the generated growth model. The method of estimating the yield can increase the reliability of the result of the prediction of crop yield by accumulating the farm data.

In step S71, the production forecasting apparatus 100 receives farmhouse data including at least one of crop growth information, environmental information, and pest-insufficient information. The production forecasting apparatus 100 can receive farmhouse data through wired / wireless communication.

In step S73, the production forecasting apparatus 100 generates a growth model by machine learning the received farmhouse data. The production forecasting apparatus 100 can perform a big data analysis in various forms, and can preferably perform correlation analysis and regression analysis. That is, the production forecasting apparatus 100 may generate a plurality of patterns by combining different factors among the farm data, and may perform correlation analysis and regression analysis using the generated plurality of patterns. The production forecasting apparatus 100 repeatedly performs correlation analysis and regression analysis to extract specific patterns for specific crops and species, and generates a growth model using the extracted specific patterns.

Between step S71 and step S73, the production forecasting apparatus 100 may collect the received farmhouse data in accordance with the data standardization format, and may filter the collected farmhouse data.

In step S75, the production forecasting apparatus 100 predicts the crop production amount using the generated growth model. The production amount prediction apparatus 100 can predict the production amount of the crop with respect to the raw weight.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is clearly understood that the same is by way of illustration and example only and is not to be taken by way of limitation in the embodiment in which said invention is directed. It will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the scope of the appended claims.

10: communication unit 20:
21: Data collecting unit 23: Data analyzing unit
25: Growth model generation unit 27:
30: output unit 40: input unit
50: storage unit 61: generating module
63: prediction module 65: learning module
100: Production forecasting device 200: Cultivation facility
210: farm 230: sensor unit
250: Actuator 270: Farm control device
300: Plant control system

Claims (10)

A communication unit for receiving farmhouse data including at least one of crop growth information, environmental information, and pest control information; And
Wherein the farmhouse data collected from the communication unit is collected in accordance with a data standardization format, and the collected farmhouse data is selected and filtered, and the filtered farmhouse data is subjected to machine learning to predict the growth of the crop, And a control unit for predicting a production amount of the crop using the generated growth model,
Wherein,
The received farmhouse data is classified by the same crop and species, the classified farmhouse data is classified according to factors and collected according to the data standardization format, and the collected farmhouse data is sorted by time zone and converted into a database A data collecting unit;
A data analyzer for analyzing the database data of the farmhouse data, filtering the farmhouse data with data error and the farmhouse data whose data quality is lower than the reference value using the analyzed result, and filtering the selected farmhouse data;
A growth model generation unit for performing correlation analysis and regression analysis on the filtered farmhouse data by the machine learning to generate the growth model; And
And a production amount predicting unit for predicting a production amount with respect to the raw weight of the crop using the generated growth model,
The growth model generation unit generates,
Generating a plurality of patterns by combining different factors among the filtered farmhouse data, performing a correlation analysis and a regression analysis on the generated plurality of patterns, and using the performed correlation analysis and regression analysis results A generation module for extracting a specific pattern of a specific crop and species among the plurality of patterns and generating the growth model using the extracted specific pattern;
A prediction module that compares the filtered farmhouse data with the growth model to analyze the current state of the crop and predict the growth environment with the highest crop yield using the analyzed growth state; And
The farmhouse data set having the highest production amount and quality of the crop among the filtered farmhouse data is calculated using the predicted growth environment, and the calculated farmhouse data set is provided to the generation module, A learning module to update;
Wherein the machine-learning-amount estimating unit estimates the yield of the crop using the machine learning. The method according to claim 1,
Wherein the learning module comprises:
Filtering the farmhouse data that is less than the reference value among the current farmhouse data by using the growth model generated by the generation module, and then calculating the farmhouse data set. delete delete delete The method according to claim 1,
The growth model generation unit generates,
Wherein the machine learning is performed by sequentially repeating the generation module, the prediction module, and the learning module. delete The method according to claim 1,
The data standardization format includes:
Wherein at least one factor of plant and cultivation information, growth information, environmental information, and farmhouse other information is classified according to a predetermined standard data standard. The method according to claim 1,
Wherein the growth information of the crop includes at least one of a production amount, a light weight, a total length, a leaf area index (LAI), a leaf temperature, a stem thickness, a flower bed spacing, a growth rate, a growth intensity and a growth phase,
The environmental information includes at least one of a type of a greenhouse, environmental facilities, operating temperature, humidity, instantaneous light intensity, carbon dioxide concentration, medium water content, medium EC and operational information,
Wherein the pest-insoluble information includes at least one of a pest-insect-occurrence frequency and a plant physiological disorder information. The method comprising: receiving farm data including at least one of crop growth information, environmental information, and pest control information;
Collecting the received farm data in accordance with a data standardization format;
Filtering and collecting the collected farmhouse data;
Performing a correlation analysis and a regression analysis on the filtered farmhouse data by machine learning to generate a growth model for predicting the growth of the crop; And
And estimating a production amount of the crop with respect to the fresh weight using the generated growth model,
Wherein the collecting comprises:
Classifying the received farm data by the same crop and species;
Classifying the classified farmhouse data by a factor and collecting the classified farmhouse data according to the data standardization format; And
And sorting the collected farmhouse data by time zone to form a database,
Wherein the filtering comprises:
Analyzing the database farmhouse data;
Selecting the farmhouse data with the data error and the farmhouse data whose data quality is below the reference value using the analyzed result; And
Filtering the selected farmhouse data,
Wherein the step of generating the growth model comprises:
Generating a plurality of patterns by combining different factors among the filtered farmhouse data, performing a correlation analysis and a regression analysis on the generated plurality of patterns, and using the performed correlation analysis and regression analysis results Extracting a specific pattern of a specific crop and species among the plurality of patterns, and generating the growth model using the extracted specific pattern;
Comparing the filtered farmhouse data with the growth model to analyze a current crop growth state and predicting a growth environment with the highest production yield of the crop using the analyzed growth state; And
Calculating a farmhouse data set having the highest production yield and quality of the crop among the filtered farmhouse data using the predicted growth environment and providing the calculated farmhouse data set to the generated farm model to update the farmhouse data ;
And estimating the yield of the crop using the machine learning.

Images