KR102620199B1 - Server and method for operating farm agricultural products distribution price prediction information platform based on artificial-intelligent - Google Patents

Abstract

One aspect of the present invention provides an artificial intelligence-based field crop distribution price prediction information platform operation server. The artificial intelligence-based field crop distribution price prediction information platform operation server includes at least one processor and a memory that stores instructions that instruct the at least one processor to perform at least one step. can do. At least one step includes receiving weather information from a weather information server, receiving economic information based on economic indicators from an economic information server, receiving press information from a press information server, and receiving weather information. Classifying the received economic information into at least one category according to the properties of the information, classifying the received economic information into at least one category according to the properties of the information, classifying the received press information into at least one category according to the properties of the information Step, generating feature vectors for weather information, economic information, and media information classified into each category, calculating weather information scores, economic information scores, and media information scores from the weighted sum of the generated feature vectors, and calculated It may include predicting the crop price by multiplying the weighted sum of at least one of the weather information score, economic information score, and media information score by the maximum observable crop price.

Description

Artificial intelligence-based field crop distribution price prediction information platform operation server and method {SERVER AND METHOD FOR OPERATING FARM AGRICULTURAL PRODUCTS DISTRIBUTION PRICE PREDICTION INFORMATION PLATFORM BASED ON ARTIFICIAL-INTELLIGENT}

The present invention relates to an artificial intelligence-based field crop distribution price prediction information platform operation server and method. More specifically, an artificial intelligence-based open field crop distribution price prediction information platform operation that can predict the distribution price of crops according to weather and climate effects. It relates to servers and methods.

Agriculture is subject to many variables due to climate change, evolving consumer trends, fluctuating economic conditions, and fluctuating policy environments.

Crop production is greatly influenced by numerous factors, including weather, environmental and climatic conditions, and various economic indicators. However, traditional agriculture often fails to process multiple variables or sufficiently consider weather conditions to make data-based decisions. In particular, crops are produced and shipped using methods that do not reflect changes in policy and climate zones, which play an important role in determining the price and demand of crops, resulting in significant losses in productivity and economic efficiency.

Additionally, news about policies, events and public opinion have a significant impact on the agricultural sector and have a significant impact on consumer behavior, market dynamics and government regulatory actions. However, because most of this information exists in unstructured form such as text, there are limits to the ability of agricultural workers to efficiently collect it and respond in real time.

Lastly, accurately predicting the crop growing season using weather and environmental data has the problem of relying on predictions that are subject to limitations and uncertainties in weather forecasting technology, and production plans must be established primarily based on interpretation of past data. There is.

Accordingly, there is a need for research into methods that can provide analysis and guidance on crop production by analyzing real-time information in a specific region and time.

Domestic registered patent No. 10-2539662

The purpose of the present invention to solve the above problems is to provide an artificial intelligence-based field crop distribution price prediction information platform operation server and method.

One aspect of the present invention for achieving the above object provides an artificial intelligence-based field crop distribution price prediction information platform operation server.

The artificial intelligence-based field crop distribution price prediction information platform operation server includes at least one processor and a memory that stores instructions that instruct the at least one processor to perform at least one step. can do.

At least one step includes receiving weather information from a weather information server, receiving economic information based on economic indicators from an economic information server, receiving press information from a press information server, and receiving weather information. Classifying the received economic information into at least one category according to the properties of the information, classifying the received economic information into at least one category according to the properties of the information, classifying the received press information into at least one category according to the properties of the information Step, generating feature vectors for weather information, economic information, and media information classified into each category, calculating weather information scores, economic information scores, and media information scores from the weighted sum of the generated feature vectors, and calculated It may include predicting the crop price by multiplying the weighted sum of at least one of the weather information score, economic information score, and media information score by the maximum observable crop price.

In addition, the step of classifying the received weather information into at least one category according to the properties of the information includes classifying weather events including temperature, rainfall, wind direction, wind speed, humidity, frost, heat waves, and hail into the current weather condition category; , classification of seasonal rainfall, drought cycle, long-term temperature trend, El Nino and La Niña cycles as weather patterns, and classification of climate zone, average hours of sunshine, soil moisture content, air quality, and ultraviolet ray index as environmental and climate factors. Additionally, the step of classifying the received economic information into at least one category according to the properties of the information includes classifying gross domestic product (GDP), unemployment rate, inflation rate, interest rate, and exchange rate as macroeconomic indicators, and classifying the received economic information into at least one category according to the properties of the information. It further includes the step of classifying input costs, agricultural productivity index, fuel price, and food demand index into sectoral indicators, and classifying global trade policy, size of subsidies and government subsidies, and real estate price changes as policy factors, and received press information The step of classifying information into at least one category according to the nature of the information is to classify agricultural news, market trends, consumer trends, and product market reports into market and economic news, and policy changes, food safety incidents, and supply chain issues into policy news. Classification may further include categorizing social media sentiment, health and nutrition trends, expert opinions, promotions and advertisements into social and public opinion news.

Meanwhile, in the step of generating a feature vector, a weather information feature vector is generated based on the weather condition score, weather pattern score, and environmental score, and an economic information feature vector is created based on the macroeconomic score, sector indicator score, and policy factor score. It may include generating a press information feature vector based on the market news score, policy news score, and public opinion news score.

At this time, the weather condition score is calculated based on the weighted sum of the normalized values of temperature, rainfall, wind direction, wind speed, humidity, effective cumulative temperature of crops, and sunshine hours, and the weather pattern score is seasonal rainfall, drought cycle, and long-term temperature trend. and the weighted sum of the normalized values of the El Niño and La Niña cycles, the environmental score is calculated based on the weighted sum of the normalized values of the climate zone, average sunshine hours, soil moisture content, and ultraviolet ray index, and the macroeconomic score. is calculated as a weighted sum of the normalized values of gross domestic product, unemployment rate, inflation rate, interest rate, and exchange rate, and the sectoral indicator score is a weighted sum of the normalized values of agricultural production input costs, agricultural productivity index, fuel price, and food demand index. The policy factor score is calculated by subtracting the real estate price change rate from the common logarithm of the subsidy size and dividing it by the normalization index, and the market news score is calculated based on the weighted sum of market trends, consumer trends, and product market report scores. The policy news score is calculated based on a weighted sum of policy change, food safety incident, and supply chain issue scores, and the public opinion news score is calculated based on a weighted sum of social media sentiment, health and nutrition trends, and promotion and advertising scores. It can be characterized as:

At this time, the method may further include calculating a cultivation shortening factor using a weighted sum of the generated weather information scores and predicting the cultivation period by multiplying the calculated cultivation shortening factor by the observed maximum cultivation period.

In addition, a step of generating a guide indicating the planting date as the date calculated by calculating the predicted cultivation period from the shipment time of the crop may be further included.

When using the artificial intelligence-based field crop distribution price prediction information platform operation server and method according to the present invention as described above, market stability is increased and crop trading is stabilized by predicting the distribution price of crops based on artificial intelligence according to weather and climate conditions. can be secured.

The effects that can be obtained from the present invention are not limited to the effects mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the description below. will be.

The above-described and other aspects, features and benefits of certain preferred embodiments of the present invention will become more apparent from the following description taken in conjunction with the accompanying drawings.
Figure 1 is an exemplary diagram showing the operating environment of an artificial intelligence-based field crop distribution price prediction information platform operation server according to an embodiment of the present invention.
Figure 2 is an exemplary diagram showing an example of an artificial intelligence-based field crop distribution price prediction information platform operation server.
Figure 3 is an exemplary diagram to explain an example of extracting media information.
Figure 4 is an example diagram for explaining a crop price prediction model.
Figure 5 is a hardware configuration diagram of an artificial intelligence-based field crop distribution price prediction information platform operation server according to Figure 1.
It should be noted that throughout the drawings, like reference numerals are used to illustrate identical or similar elements, features and structures.

Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

In describing the embodiments, descriptions of technical content that is well known in the technical field to which the present invention belongs and that are not directly related to the present invention will be omitted. This is to convey the gist of the present invention more clearly without obscuring it by omitting unnecessary explanation.

For the same reason, some components are exaggerated, omitted, or schematically shown in the accompanying drawings. Additionally, the size of each component does not entirely reflect its actual size. In each drawing, identical or corresponding components are assigned the same reference numbers.

The advantages and features of the present invention and methods for achieving them will become clear by referring to the embodiments described in detail below along with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various different forms. The present embodiments are merely provided to ensure that the disclosure of the present invention is complete and to provide common knowledge in the technical field to which the present invention pertains. It is provided to fully inform those who have the scope of the invention, and the present invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout the specification.

At this time, it will be understood that each block of the processing flow diagrams and combinations of the flow diagram diagrams can be performed by computer program instructions. These computer program instructions can be mounted on a processor of a general-purpose computer, special-purpose computer, or other programmable data processing equipment, so that the instructions performed through the processor of the computer or other programmable data processing equipment are described in the flow chart block(s). It creates the means to perform functions. These computer program instructions may also be stored in computer-usable or computer-readable memory that can be directed to a computer or other programmable data processing equipment to implement a function in a particular manner, so that the computer-usable or computer-readable memory It is also possible to produce manufactured items containing instruction means that perform the functions described in the flowchart block(s). Computer program instructions can also be mounted on a computer or other programmable data processing equipment, so that a series of operational steps are performed on the computer or other programmable data processing equipment to create a process that is executed by the computer, thereby generating a process that is executed by the computer or other programmable data processing equipment. Instructions that perform processing equipment may also provide steps for executing the functions described in the flow diagram block(s).

Additionally, each block may represent a module, segment, or portion of code that includes one or more executable instructions for executing specified logical function(s). Additionally, it should be noted that in some alternative execution examples it is possible for the functions mentioned in the blocks to occur out of order. For example, it is possible for two blocks shown in succession to be performed substantially at the same time, or it is possible for the blocks to be performed in reverse order depending on the corresponding function.

At this time, the term '~unit' used in this embodiment refers to software or hardware components such as FPGA (field-programmable gate array) or ASIC (Application Specific Integrated Circuit), and '~unit' refers to what role perform them. However, '~part' is not limited to software or hardware. The '~ part' may be configured to reside in an addressable storage medium and may be configured to reproduce on one or more processors. Therefore, as an example, '~ part' refers to components such as software components, object-oriented software components, class components, and task components, processes, functions, properties, and procedures. , subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays, and variables. The functions provided within the components and 'parts' may be combined into a smaller number of components and 'parts' or may be further separated into additional components and 'parts'. Additionally, components and 'parts' may be implemented to regenerate one or more CPUs within a device or a secure multimedia card.

In describing the embodiments of the present invention in detail, the main focus will be on the example of a specific system, but the main point claimed in this specification is that the scope disclosed in this specification is applicable to other communication systems and services with similar technical background. It can be applied within a range that does not deviate significantly, and this can be done at the discretion of a person with skilled technical knowledge in the relevant technical field.

Figure 1 is an exemplary diagram showing the operating environment of an artificial intelligence-based field crop distribution price prediction information platform operation server according to an embodiment of the present invention.

Referring to Figure 1, the artificial intelligence-based field crop distribution price prediction information platform operation server 100 (hereinafter referred to as 'server 100') receives weather information from the weather information server 20 through a wired or wireless network. You can.

Here, weather information includes temperature, precipitation, humidity, wind direction, wind speed, frost/freeze/drought/hail, effective cumulative temperature of crops, sunlight hours, sunlight amount, soil moisture, atmospheric pressure, UV index, El Nino/La Niña intensity, and climate zone. zone), may include, but is not limited to, past weather data.

Additionally, when the server 100 receives weather information from the weather information server 20, the server 100 may classify the weather information into at least one category according to the received weather information.

Meanwhile, the server 100 may receive economic information based on economic indicators from the economic information server 30 through a wired or wireless network.

Here, economic information includes gross domestic product (GDP), inflation rate, unemployment rate, interest rate, exchange rate, consumer price index (CPI), commodity price, fuel price, wage rate, agricultural input cost, food demand index, global trade policy, and agricultural productivity. This may include, but is not limited to, indices, population growth rates, climate change policies, subsidies and government subsidies, real estate prices, consumer confidence, stock market performance and retail sales data.

Additionally, when the server 100 receives economic information from the economic indicator server 30, the server 100 may classify the economic information into at least one category according to the received economic information.

Meanwhile, the server 100 may receive press information from the press information server 40 through a wired or wireless network.

Here, media information includes agricultural news, policy changes, market trends, consumer trends, new technologies, social media sentiment, food safety incidents, global events, commodity market reports, climate change reports, industry analysis reports, health and nutrition trends, and consumer prices. Reports may include, but are not limited to, supply chain issue reports, agricultural investment trends, and labor and farm issue trends, and may include information affecting crop prices.

Additionally, when the server 100 receives economic information from the press information server 40, the server 100 may classify the press information into at least one category according to the received press information.

When weather, economic, and media information is classified, the server 100 can predict the price of crops by analyzing the classified weather, economic, and media information.

For example, the server 100 may organize data by preprocessing received weather, economic, and media information. As an example, the received information can be preprocessed to normalize and standardize the data, and categorical variables can be converted into quantitative variables (e.g., one-hot encoding).

Next, the server 100 can train a crop price prediction model. As an example, the server 100 extracts the characteristics of the received weather, economics, and media information, quantifies them, and generates a feature vector to analyze the correlation between the received weather, economic, and media information and crop prices, thereby creating a variable. It can be adjusted to the level necessary for crop price prediction, and crop prices can be predicted based on the generated feature vector.

For example, the server 100 may use a regression model based on a machine learning (ML) model. Regression models can predict persistent outcomes (prices), for example, crop prices through linear regression, decision tree regression, random forest regression, gradient boosting, or neural network analysis. For example, a learning model can be trained by dividing the received data into a training set and a validation set, training the model in the training set, and then verifying the prediction result with data in the validation set.

Next, the server 100 may evaluate the learned crop price prediction model. For example, if the crop prediction model generated by the server 100 predicts the price of apples, based on the mean absolute error (MAE), mean square error (MSE), or root mean square error (RMSE) for the regression model The learning model can be evaluated by measuring the error.

If it is determined that the learning model has been learned within a predetermined reliability, the server 100 can predict the price of the crop.

For example, if the weather, economic, and media data received by the server 100 have a stable economic environment and stable media sentiment for crops (e.g., apples) under favorable meteorological cultivation conditions, the server 100 The learning model can predict that apple prices will steadily drop due to high supply and steady demand.

Meanwhile, when the price of a crop is predicted, the server 100 generates a guide to provide guidance to users about the demand (e.g., wholesale or retail) or supply (e.g., producer and seller) of the crop according to the predicted price level. And this can be sent to the user terminal 10 as a notification.

Figure 2 is an exemplary diagram showing an example of an artificial intelligence-based field crop distribution price prediction information platform operation server. Referring to Figure 2, the artificial intelligence-based field crop distribution price prediction information platform operation server 200 (hereinafter referred to as 'server 200') includes a weather information analysis unit 210, an economic indicator analysis unit 220, It may include a press information analysis unit 230 and a crop price prediction unit 240.

For example, the weather information analysis unit 210 may receive weather information through a wired or wireless network.

Here, weather information includes temperature, precipitation, humidity, wind direction, wind speed, frost/freeze/drought/hail, sunshine hours, sunlight amount, soil moisture, atmospheric pressure, UV index, El Niño/La Niña intensity, climate zone, and past weather data. It may include, but is not limited to.

After receiving weather information, the weather information analysis unit 210 may classify each received weather information into at least one category according to the properties of the information.

For example, the weather information analysis unit 210 may classify the received weather information as current weather information. As an example, current weather conditions may be time-sensitive and information related to the real-time status of crops and agricultural operations. For example, current weather conditions may include weather information that reflects real-time or daily weather conditions, and current weather conditions have a direct impact on the costs and yields of agricultural production, i.e. prices, agricultural operations such as harvesting and sowing. It may be information related to . For example, the weather information analysis unit 210 may classify weather events such as daily temperature, daily rainfall, wind direction, wind speed, humidity, frost/heat wave/hail, and effective integrated temperature of crops into current weather condition categories.

Additionally, the weather information analysis unit 210 may classify the received weather information into weather patterns. For example, long-term weather patterns are information about annual or multi-year weather patterns that affect crop selection, yield estimates, and the risk of crop disease or pest infestation, which can have a significant impact on crop prices. It can be an influencing factor. Weather patterns can be information that can be used to estimate weather patterns over an annual or multi-year extended period, and to plan and predict impacts on crop production and prices. For example, the weather information analysis unit 210 may classify seasonal rainfall, drought cycles, long-term temperature trends, and El Niño/La Niña cycles as weather patterns.

Additionally, the weather information analysis unit 210 may classify the received weather information into environmental and climate factors. For example, environmental and climatic factors may vary depending on geographic location, may affect the type of crop grown, its yield and susceptibility to disease, and may be factors that affect the overall health and growth potential of the crop. may be, and the weather information analysis unit 210 may classify climate zone, average sunshine hours (latitude), soil moisture content (ratio of precipitation and evaporation rate), air quality, and ultraviolet ray index into environmental and climate factors. there is.

The weather information analysis unit 210 may calculate a weather condition score based on current weather information. For example, the weather information analysis unit 210 may calculate a weather condition score based on temperature, rainfall, wind direction, wind speed, humidity, effective integrated temperature, and sunshine hours classified as current weather information from the received weather information. and can be normalized through min-max normalization.

Meanwhile, the weather information analysis unit 210 can calculate the effective integrated temperature based on the cultivated crops and temperature. For example, the weather information analysis unit 210 determines the basic temperature (e.g., the temperature at which crop growth becomes 0) according to the maximum temperature, minimum temperature, and species of cultivated crops classified by temperature in the received weather information. The effective integrated temperature can be calculated based on 5°C for temperate crops and 10°C for temperate crops. For example, if the daily maximum temperature is 28°C, the daily minimum temperature is 18°C, and the basic temperature is 10°C for temperate crops, the effective integrated temperature can be calculated using the equation below.

The effective integrated temperature is calculated based on the basic temperature, and the values within the range of the basic temperature and maximum temperature are accumulated and compared with the integrated temperature, which is the cumulative sum of the temperatures required from the sowing to the harvest stage of the crop, and is an indicator that can predict the crop cultivation period. It can be used as

For example, when the calculated effective integrated temperature is less than the basic temperature, the weather information analysis unit 210 uses one of the basic temperature, the calculated effective integrated temperature, or 0°C as the effective integrated temperature depending on the species or cultivation method of the cultivated crop. You can choose.

In addition, if the calculated effective integrated temperature exceeds the maximum temperature range, the weather information analysis unit 210 effectively uses 30°C, which is generally known to achieve maximum growth, or the maximum value according to the species and cultivation method of the cultivated crop. You can select the integrated temperature. For example, if the temperature is 27°C, rainfall is 0, wind speed is 10m/s, humidity is 80%, sunshine time is 13 hours, the range for each factor is temperature (0-50 °C), rainfall (0-100 mm), wind speed (0-50 m/s), humidity (0-100%), effective cumulative temperature of 13°C, and sunshine hours (0-24 h). Climate condition score (CCS) can be calculated as shown in the equation below.

Meanwhile, the weather information analysis unit 210 may assign predetermined weights to weather elements important to each crop in the current weather information. For example, in the example above, each quantitative element is multiplied by predetermined weights w1 to w5 to determine their CCS can be calculated from the weighted sum, and the total sum of predetermined weights w1 to w5 may be 1.

CCS ranges from 0 to 1, for example, the best weather conditions for crop growth can be 1, and the worst weather conditions can be 0. For example, the weather information analysis unit 210 may estimate that if the CCS is high, the current weather conditions are very favorable for crop growth, thus increasing supply and potentially lowering prices.

Meanwhile, the weather information analysis unit 210 may calculate a weather pattern score based on the weather pattern. For example, the weather information analysis unit 210 may calculate a weather pattern score based on seasonal rainfall, drought cycle, long-term temperature trend, and El Niño/La Niña cycle classified into weather patterns from the received weather information, and may calculate a weather pattern score based on a minimum -Can be normalized through maximum normalization.

For example, seasonal rainfall is 700mm, very severe droughts (average annual rainfall of 500mm or less) occur every 7 years, long-term temperature trends include an average annual temperature increase of 0.01°C over the past 50 years, and four El Niño events over the past 20 years. and two La Niña events occurred, seasonal rainfall was 200-1200 mm, drought cycle was 2-10 years, long-term temperature trend was -0.02 ~ +0.02°C per year, and El Niño/La Niña cycle was 0 to 5 times per decade. It has an occurrence range, and as a weight for the growth of crops, if the predetermined weights w1 to w4 are 0.4, 0.2, 0.3, and 0.1, and the total of these is 1, the weather pattern score (WPS) is as follows. It can be calculated as a mathematical equation.

That is, a high WPS could mean that long-term weather patterns are generally favorable for crop growth, potentially leading to higher yields and lower prices, while a lower score indicates unfavorable conditions, which could mean lower yields and higher prices. This could mean that it could lead to .

Meanwhile, the weather information analysis unit 210 may calculate an environmental score based on environmental and climate factors. For example, the weather information analysis unit 210 may calculate an environmental score based on climate zone, average sunshine time, soil moisture content, and ultraviolet ray index classified as environmental and climate factors from the received weather information, and may calculate an environmental score based on a minimum -Can be normalized through maximum normalization.

For example, the climate zone is tropical (2), the average sunshine time is 6 hours per day, the soil moisture content is defined as 600/700=0.857 with the average annual precipitation being 600 mm and evaporation being 700 mm, and the UV index is 6. Predetermined climate zones range from 1 to 3, with average sunlight hours ranging from 0 to 24 hours, soil moisture content ranging from 0 to 1, and ultraviolet ray index ranging from 1 to 11, with predetermined values assigned to each element. When the weights w1 to w4 are 0.4, 0.2, 0.3, and 0.1, and their total sum is 1, the environment score (ES) can be calculated as in the equation below.

In other words, higher ES may generally be beneficial to crop growth, which can lead to higher crop yields and potentially lower prices. Meanwhile, in the predetermined climate zone, polar and desert climates can be predetermined as 1, temperate and tropical climates as 2, and subtropical climates as 3.

Meanwhile, the economic indicator analysis unit 220 may receive economic information based on economic indicators from an economic information server through a wired or wireless network.

Here, economic information includes gross domestic product (GDP), inflation rate, unemployment rate, interest rate, exchange rate, consumer price index (CPI), commodity price, fuel price, wage rate, agricultural input cost, food demand index, global trade policy, and agricultural productivity. This may include, but is not limited to, indices, population growth rates, climate change policies, subsidies and government subsidies, real estate prices, consumer confidence, stock market performance and retail sales data.

After receiving economic information, the economic indicator analysis unit 220 may classify each received economic information into at least one category according to the properties of the information.

For example, the economic indicator analysis unit 220 may classify the received economic information into macroeconomic indicators. As an example, macroeconomic indicators may include a broad range of economy-wide indicators that provide an overview of the health and direction of the overall economy, which indirectly affects crop prices by influencing overall economic conditions and consumer purchasing power. It can go crazy. Additionally, macroeconomic indicators can be broad indicators that measure the overall state of the macroeconomy. For example, the economic indicator analysis unit 220 can classify gross domestic product (GDP), unemployment rate, inflation rate, interest rate, and exchange rate as macroeconomic indicators.

Additionally, the economic indicator analysis unit 220 may classify the received economic information into sector-specific indicators. As an example, sectoral indicators are economic indicators directly related to the agricultural sector or closely related sectors, and may be indicators that are more specific and likely to have a direct impact on crop distribution prices, and are directly linked to agricultural production, supply and demand. May include indicators. For example, the economic indicator analysis unit 220 can classify product prices, agricultural input costs, agricultural productivity index, fuel price, and food demand index into sectoral indicators.

Additionally, the economic indicator analysis unit 220 may classify the received economic information into policy factors. As an example, policy factors may include economic indicators shaped by government policies, international trade agreements, and regulatory decisions that affect crop prices by changing production costs, modifying import/export dynamics, or providing subsidies. may be included. Additionally, policy factors may include indicators related to government and international trade policies that affect the agricultural economy. For example, the economic indicator analysis unit 220 can classify global trade policy, the size of subsidies and government subsidies, and real estate price changes as policy factors.

Meanwhile, the economic indicator analysis unit 220 may calculate a macroeconomic score based on macroeconomic indicators. For example, the economic indicator analysis unit 220 may calculate a macroeconomic score based on gross domestic product, unemployment rate, inflation rate, interest rate, and exchange rate classified as macroeconomic indicators from the received economic information, with min-max normalization. It can be normalized through .

For example, GDP is 200 trillion won, unemployment rate is 4%, inflation rate is 2%, interest rate is 1.5%, exchange rate (USD/KRW) is 1200, and the category of each indicator over the past 10 years is GDP 160-240 trillion won. , unemployment rate is 3-10%, inflation rate is 1-3%, interest rate is 0.25-3%, exchange rate is 1000-1400, and predetermined weights w1 to w5 may each be 0.2. At this time, the economic indicator analysis unit 220 can calculate a macroeconomics score (MS) as shown in the equation below.

In other words, for a specific crop, GDP, unemployment rate, inflation rate, interest rate, and exchange rate may have a complementary relationship. For example, GDP may have a positive effect on crop prices within a certain range, after which further increases in GDP may have little or no effect on prices.

Additionally, changes in unemployment rates may not have an immediate effect on crop prices, but may lag in the future (e.g. months or years) and have a negative impact on GDP. Additionally, the impact of inflation on crop prices may vary depending on other factors. For example, if inflation occurs due to rising energy prices, production costs may increase and crop prices may rise.

In addition, exchange rates may vary depending on many other factors, such as trade policies, tariffs, the relative strength of the economies of trading partners, and global raw material price trends. In general, an increase in exchange rates will require an increase in crop prices due to an increase in the price of imported crops. You can. In other words, the higher the MS, the higher the price of crops can be.

Meanwhile, the economic indicator analysis unit 220 can calculate sector-specific indicator scores based on sector-specific indicators. For example, the economic indicator analysis unit 220 generates sector indicator scores (IIS, Indicators index) based on agricultural production input costs, agricultural productivity index, fuel price, and food demand index classified as sector indicators in the received economic information. score) can be calculated and normalized through min-max normalization.

For example, the agricultural production input cost, which is the cost of agricultural production, including seeds, fertilizers, pesticides, agricultural equipment, and labor costs, is 5,000 won, the agricultural productivity index based on 100 is 105, and agricultural machinery It is assumed that the price of diesel fuel for operation is 1,400 won per liter, with 100 as the standard, and that the food demand index, which is determined by factors of population, income level, and eating habits, is 110. In addition, when the maximum observed values of agricultural production input cost, agricultural productivity index, fuel price, and food demand index are 10,000 won, 130, 2,000 won, and 120, respectively, and the predetermined weights w1 to w4 are each 0.25, the index score for each sector can be calculated as in the equation below.

In other words, the sectoral indicator score calculated based on the sectoral indicator can predict the price of the crop according to its value, and if the maximum observed price per 1 kg of rice is 10,000 won, the predicted price based on the sectoral indicator score can be predicted as 10000*0.73=7300 won.

Meanwhile, the agricultural productivity index can be determined by the input-to-output ratio, and can be determined by the agricultural production input cost compared to the crop shipments of the previous year in a specific region. In addition, the food demand index is estimated based on population, income level, and dietary trends. It can be assumed that the average rice consumption in the previous year was 100 kg, the population was 50 million, and the average income was 30 million won. In addition, if per capita consumption increases to 105 kg this year, the population increases to 52 million, and average income increases to 31 million won, the food demand index will increase. It can be calculated as, and can be divided by the observed maximum value to normalize it.

Meanwhile, the economic indicator analysis unit 220 can calculate policy factor scores based on policy factors. For example, the economic indicator analysis unit 220 may calculate a policy index score (PIS) based on the subsidy size and real estate price changes classified as policy factors in the received economic information.

For example, it can be assumed that the government provides an agricultural subsidy of 1 million won per hectare for a specific crop, and the annual change rate of land used for farming increases by 5% every year. At this time, the policy factor score is calculated as shown in the equation below.

Meanwhile, when the maximum possible agricultural subsidy per 1 ha is 1.5 million won and the maximum annual land change rate of agricultural land is 10% per year, PIS can be normalized based on the normalization index as shown in the equation below.

In other words, the closer PIS is to 1, the greater the subsidy and the lower the real estate price fluctuation, which can mean that stable crop production is possible.

The press information analysis unit 230 can receive press information from a press information server through a wired or wireless network.

Here, media information includes agricultural news, policy changes, market trends, consumer trends, new technologies, social media sentiment, food safety incidents, global events, commodity market reports, climate change reports, industry analysis reports, health and nutrition trends, and consumer prices. Reports may include, but are not limited to, supply chain issue reports, agricultural investment trends, and labor and farm issue trends, and may include information affecting crop prices.

Additionally, when the press information analysis unit 230 receives economic information from the press information server, it may classify the press information into at least one category according to the received press information.

For example, the media information analysis unit 230 may classify agricultural news, market trends, consumer trends, and product market reports into market and economic news.

In addition, the media information analysis department (230) can classify policy changes, food safety incidents, and supply chain issues as policy news, and social media sentiment, health and nutrition trends, expert opinions, promotions, and advertisements as social and public opinion news. can do.

Meanwhile, Figure 3 is an exemplary diagram to explain an example of extracting press information.

Referring again to FIGS. 2 and 3 , the media information analysis unit 230 may train a learning model for extracting weather and climate damage information from news articles. For example, if the information to be extracted is the type and intensity of weather and climate phenomena that caused crop damage/crop damage area, area, amount, number of farms, and area of damaged crops, the press information analysis department 230 You can crawl news articles related to crop damage caused by climate events.

For example, the press information analysis unit 230 can configure a crawling module using Python's BeutifulSoup, request, pandas, and newspaper3k modules, and can crawl data by removing duplicate contents of crawled data and loading it into Elasticsearch DB.

Next, the media information analysis unit 230 normalizes words in the text content, processes stopwords (e.g., KoNLPy module), re-tokenizes (syllable units) using the Sentencepiece tokenizer, and trains the tokenized data in the KoBERT model. Through this, it is possible to implement the MRC model for extracting crop damage information based on the learned KoBERT model.

Meanwhile, the media information analysis unit 230 can calculate a market news score based on market and economic news. For example, the press information analysis unit 230 may calculate a market news score (MNS, Marker news score) based on market trends, consumer trends, and product market reports.

As an example, market trends are an indicator of the general direction of product prices in the market, where phrases revealed in market trend-related news can be scored as -1 (very negative) and +1 (very positive) using a sentiment analysis tool. You can. Additionally, consumer trends can take into account trend measures, for example, and Google Trends scores can be used.

Additionally, a product market report may include sentiment about the product's price, supply, and demand, which can be assigned a score of -1 (very negative) and +1 (very positive). Meanwhile, as sentiment analysis tools for news articles, Google Cloud Natural Language API, IBM Watson Natural Language Understanding, Microsoft Azure Text Analytics API, NLTK (Natural Language Toolkit), VADER (Valence Aware Dictionary and sEntiment Reasoner), Lexalytics, and RapidMiner can be used. However, it is not limited to this.

For example, assuming that the market trend is 0.7 points, the consumer trend is 0.75 points (e.g., Google Trends 75 points), and the product market report is 0.6 points, the press information analysis unit 230 can calculate the MNS as shown in the equation below.

Here, the predetermined weights w1 to w3 may be determined in advance according to importance, and their total may be 1. MNS represents current market dynamics based on recent market trends, consumer trends and product market reports, and higher scores can indicate favorable conditions for rising crop prices.

Meanwhile, the media information analysis unit 230 may calculate a policy news score based on policy and event news. For example, the press information analysis unit 230 may calculate a policy news score (PNS) based on policy changes, food safety accidents, and supply chain issues.

For example, policy changes may be determined by their impact on crop prices, and policies that encourage farmers and agriculture may lead to increased production, which may lead to lower prices due to increased supply. Additionally, food safety incidents can be scored by assigning -1 point when an incident occurs using news feeds and machine learning models for event detection. Additionally, supply chain issues can be determined by crawling articles about price changes due to oversupply or undersupply, giving +1 for undersupply and -1 for oversupply, and a score given for each word crawled. The score can be calculated by dividing the sum by the total number of words.

For example, assuming that the policy change is 0.5 points, the food safety incident is 0, and the supply chain problem is 0.4 points, the media information analysis unit 230 can calculate the PNS as shown in the equation below.

Here, the predetermined weights w1 to w3 may be determined in advance according to importance, and their total may be 1. PNS refers to recent policies and events regarding crops, and the closer it is to 1, the higher the price of crops can be.

Additionally, the media information analysis unit 230 can calculate a public opinion news score based on social and public opinion news. For example, the media information analysis unit 230 may calculate a general opinion news score (GNS) based on social media sentiment, health and nutrition trends, promotions, and advertisements.

As an example, social media sentiment can be determined through analysis of popular social media platforms where agriculture-related discussions take place. Emotions can be positive, negative, or neutral and can be expressed as a numeric score: +1 for positive, -1 for negative, and 0 for neutral.

Additionally, health and nutrition trends are factors that can affect demand for certain crops. For example, if current trends are toward a high-protein, low-carbohydrate diet, demand for corn, a carbohydrate-rich grain, may decline. Health and nutrition trends can be quantified using sentiment analysis tools that can rate trends on a scale of -1 to +1. Additionally, for a promotion or advertisement, +1 may indicate a very effective promotion leading to increased demand, -1 may indicate an advertisement that is ineffective or received negatively, etc.

For example, assuming that social media sentiment is 0.3, health and nutrition trends are 0.7, and promotions and advertisements are 0.3, the press information analysis unit 230 can calculate GNS as shown in the equation below.

Here, the predetermined weights w1 to w3 may be determined in advance according to importance, and their total may be 1. GNS represents the current emotions of general consumers, and the closer it is to 1, the more positive it is, indicating that it can attract consumers even at high prices.

The crop price prediction unit 240 can predict the price of crops based on analyzed weather information, economic information, and media information.

Figure 4 is an example diagram for explaining a crop price prediction model.

Referring to FIGS. 2 and 4 , the crop price prediction unit 240 may train a learning model for predicting the price of crops. For example, the crop price prediction unit 240 generates exponential moving average (EMA), moving average convergence divergence index (MACD), and relative strength index (RSI) as derived variables for the price of a specific crop, and generates price, distribution, and , prices, weather, and crop impact data can be integrated.

Next, the crop price prediction unit 240 can learn a price prediction model. As an example, the crop price prediction unit 240 may train a dual attention LSTM price prediction model. For example, in the attention layer, a layer divided into time series (Time attention) and feature (Feature Attention) layers can be extracted, and LSTM crop price predictions can be made by reflecting the results of the attention layer.

Meanwhile, the learned model evaluates the prediction reliability of the learning model based on RMSE (Root Mean Square Error) or MAPE (Mean Absolute Percentage Error), and uses the learned model above a predetermined threshold to predict crop prices. can do.

In addition, the crop price prediction unit 240 includes a climate condition score (CCS), a weather pattern score (WPS), an environment score (ES) calculated based on weather information, and Macroeconomics score (MS) calculated based on economic information, sector indicator score (IIS), policy index score (PIS), and market news calculated based on media information. Crop prices can be predicted from the weighted sum of the Marker news score (MNS), Policy news score (PNS), and General opinion News score (GNS).

For example, crop prices can be predicted for pre-specified crops using weather, economic, and media information at a pre-specified time and location.

For example, to predict the price of 1 kg of rice to be shipped from Naju City, Jeollanam-do on August 10, 2023, the crop price prediction unit 240 can collect and analyze weather information, economic information, and media information.

The crop price prediction unit 240 may calculate weather information scores, economic information scores, and media information scores from a feature vector characterized by scores calculated as a result of analysis of weather information, economic information, and media information.

For example, Climate condition score (CCS), Weather pattern score (WPS), and Environment score (ES) are characteristics of weather information calculated as a result of analysis of weather information, for example, The crop price prediction unit 240 uses weather information score = [0.8, 0.7] as a feature vector of a climate condition score (CCS), a weather pattern score (WPS), and an environment score (ES). , 0.5] can be generated.

In addition, the Macroeconomics Score (MS), Indicators index score (IIS), and Policy index score (PIS) are characteristics of economic information calculated as a result of analysis of economic information, For example, the crop price prediction unit 240 uses a feature vector of a macroeconomics score (MS), sector indicator score (IIS), and policy index score (PIS), with economic information score = [0.4, 0.2, 0.6] can be generated.

In addition, the market news score (MNS, Marker news score), policy news score (PNS), and general opinion news score (GNS) are characteristics of media information calculated as a result of analysis of media information. For example, the crop price prediction unit 240 uses press information as a feature vector of market news score (MNS, Marker news score), policy news score (PNS, Policy news score), and general opinion news score (GNS). Score=[0.4, 0.8, 0.9] can be generated.

When each feature vector is generated, the crop price prediction unit 240 can calculate a weather information score, an economic information score, and a media information score from the weighted sum of each feature vector element.

For example, when the weather information score feature vector is [0.8, 0.7, 0.5] and the predetermined weight matrix [w1, w2, w3] = [0.3, 0.2, 0.5], the weather information score is calculated as follows.

Additionally, when the economic information score feature vector is [0.4, 0.2, 0.6] and the predetermined weight matrix [w1, w2, w3] = [0.4, 0.3, 0.3], the economic information score is calculated as follows.

In addition, when the press information score feature vector is [0.4, 0.8, 0.9] and the predetermined weight matrix [w1, w2, w3] = [0.3, 0.3, 0.4], the press information score is calculated as follows.

Next, the crop price prediction unit 240 can predict crop prices by multiplying the weighted sum of at least one of the calculated weather information score, economic information score, and media information score by the maximum observable crop price.

Here, the maximum observable price may refer to the maximum price that the specific crop whose price you are trying to predict has over a predetermined period of time (e.g., 5 years, 10 years, etc.).

For example, the maximum observable price of 1 kg of rice may be 12,000 won for 10 years. At this time, the crop price prediction unit 240 can predict the price of 1 kg of rice from the weighted sum of the calculated weather information score and economic information score. For example, if the current (e.g. August 10, 2023) weather information score calculated based on Naju, Jeollanam-do is 0.63, the economic information score is 0.56, and the predetermined weights w1 and w2 are each 0.5, the predicted price is calculated using the math below: It is calculated as in the formula.

In addition, the forecast price is calculated based on the calculated weather information score, economic information score, and media information score. Currently (e.g., August 10, 2023), the calculated weather information score based on Naju, Jeollanam-do is 0.63, and the economic information score is 0.63. If the information score is 0.56, the media information score is 0.72, and the predetermined weights w1 to w3 are 0.2, 0.3, and 0.5, the predicted price is calculated as in the equation below.

In this way, the crop price prediction unit 240 can predict crop prices based on weather, economic, and media information in a specific time period in a specific region and provide accurate forecast values.

Figure 5 is an exemplary diagram showing an example of an artificial intelligence-based field crop distribution price prediction information platform operation server. Referring to Figure 5, the artificial intelligence-based field crop distribution price prediction information platform operation server 500 (hereinafter referred to as 'server 500') includes a weather information analysis unit 510, an economic indicator analysis unit 520, It may include a press information analysis unit 530, a crop cultivation period prediction unit 540, a crop price prediction unit 550, and a guide provision unit 560. Here, the weather information analysis unit 510, the economic indicator analysis unit 520, the media information analysis unit 530, and the crop price prediction unit 550 are the weather information analysis unit 210 and the economic information analysis unit 210 previously described with reference to FIG. 2. Since it may have the same configuration as the index analysis unit 220, the media information analysis unit 230, and the crop price prediction unit 240, a detailed description will be given below focusing on non-overlapping configurations.

The crop cultivation period prediction unit 540 may predict the crop cultivation period based on the weather information analysis of the weather information analysis unit 510.

For example, the weather information score is a feature vector of the climate condition score (CCS), weather pattern score (WPS), and environment score (ES) calculated through analysis results of weather information. When =[0.8, 0.7, 0.5] is generated, the crop cultivation period prediction unit 540 can predict the cultivation period by multiplying the observed maximum cultivation period by the cultivation shortening factor calculated as a weighted sum of weather information scores.

As an example, in Naju-si, Jeollanam-do, the weather information score feature vector on August 10, 2023 is [0.8, 0.7, 0.5], predetermined weights w1 to w3 are 0.3, 0.3, and 0.4, respectively, and the maximum observed cultivation period is 110 days. , cultivation shortening factors and predicted cultivation period can be calculated using the equation below. Meanwhile, here, the total sum of the predetermined weights may be set to 1 or more and less than 2, and depending on weather conditions, the total sum of the weights may be set to exceed 1.

Here, the observed maximum cultivation period refers to the time taken from sowing a specific crop to harvesting during a predetermined period (e.g., 10 years). Meanwhile, the crop cultivation period prediction unit 540 predicts the cultivation period of a specific crop based on weather information analyzed at a specific time in a specific location, thereby predicting the sowing time and cultivation period of the crop to provide a smooth supply of crops. You can. For example, the crop cultivation period prediction unit 540 predicts the crop cultivation period based on the weather information at a specific time at a specific location, the weather forecast, and the effective integrated temperature of the cultivated crop, thereby predicting the cultivation period according to the sowing and harvest time of the crop. Therefore, it can be used as an indicator for the smooth supply of crops.

Meanwhile, referring to FIGS. 1 and 5, the guide providing unit 560 provides a guide for crop breeding to the user terminal 10 according to the prediction results of the crop cultivation period prediction unit 540 and the crop price prediction unit 550. You can create a notification containing .

For example, the guide provider 560 provides a notification containing a guide for crop breeding to the user terminal 10 by comparing at least one of the weather information score, economic information score, and media information score with a predetermined threshold. can be created.

At this time, the guide may include sowing date and time, irrigation schedule, harvest time, weather alert, market price change alert, policy change alert, and subsidy change alert.

As an example, the guide provider 560 shortens the predicted cultivation period from 110 days, the maximum observed cultivation period, to 71.5 days for users who farm rice in Naju City, Jeollanam-do in 2023, according to the prediction results of the crop cultivation period prediction unit 540. You can notify that it has been done.

Additionally, the guide provider 560 may generate a guide informing the sowing date of the date calculated backward from the shipment date of the crop to the predicted cultivation period and provide the guide to the user terminal 10 . For example, if the general shipment date for a specific crop is October 1, the maximum observed cultivation period is 110 days, and the predicted cultivation period is 71 days, sowing should occur around May 20 according to the maximum observation period. However, if the predicted cultivation period is shortened to 71 days due to weather conditions, sowing only needs to be done on about July 20th.

Accordingly, the guide provider 560 may calculate the predicted cultivation period (e.g., 71 days) from the shipment time of the crop and provide a notification and guide for sowing on July 20 of the same year.

Figure 6 is a hardware configuration diagram of an artificial intelligence-based field crop distribution price prediction information platform operation server according to Figure 1.

Referring to FIG. 6, the artificial intelligence-based field crop distribution price prediction information platform operation server 600 instructs at least one processor 610 and at least one processor 6610 to perform at least one step. It may include a memory 620 that stores instructions.

Here, the at least one processor 610 may mean a central processing unit (CPU), a graphics processing unit (GPU), or a dedicated processor on which methods according to embodiments of the present invention are performed. You can. Each of the memory 620 and the storage device 660 may be comprised of at least one of a volatile storage medium and a non-volatile storage medium. For example, the memory 620 may be comprised of at least one of read only memory (ROM) and random access memory (RAM).

In addition, the artificial intelligence-based field crop distribution price prediction information platform operation server 600 may include a transceiver 630 that performs communication through a wireless network. In addition, the artificial intelligence-based field crop distribution price prediction information platform operation server 600 may further include an input interface device 640, an output interface device 650, a storage device 660, etc. Each component included in the artificial intelligence-based field crop distribution price prediction information platform operation server 600 is connected by a bus 670 and can communicate with each other.

Methods according to the present invention may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. Computer-readable media may include program instructions, data files, data structures, etc., singly or in combination. Program instructions recorded on a computer-readable medium may be specially designed and constructed for the present invention or may be known and usable by those skilled in the computer software art.

Examples of computer-readable media may include hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, etc. Examples of program instructions may include machine language code such as that created by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The above-described hardware device may be configured to operate with at least one software module to perform the operations of the present invention, and vice versa.

Additionally, the above-described method or device may be implemented by combining all or part of its components or functions, or may be implemented separately.

Although the present invention has been described above with reference to preferred embodiments, those skilled in the art may make various modifications and changes to the present invention without departing from the spirit and field of the present invention as set forth in the claims below. You will understand that you can do it.

Claims (5)

As an artificial intelligence-based field crop distribution price prediction information platform operation server,
at least one processor; and
a memory storing instructions that instruct the at least one processor to perform at least one step;
The at least one step is,
Receiving weather information from a weather information server;
Receiving economic information based on economic indicators from an economic information server;
Receiving press information from a press information server;
Classifying the received weather information into at least one category according to information properties;
Classifying the received economic information into at least one category according to information properties;
Classifying the received press information into at least one category according to information properties;
generating feature vectors for the weather information, the economic information, and the media information classified into each category;
calculating a weather information score, an economic information score, and a media information score from the weighted sum of the generated feature vectors; and
Comprising: predicting the crop price by multiplying the calculated weighted sum of at least one of the weather information score, the economic information score, and the media information score by the maximum observable crop price;
Classifying the received weather information into at least one category according to the properties of the information,
Weather events including temperature, rainfall, wind direction, wind speed, effective cumulative temperature, humidity, frost, heat waves and hail are categorized into current weather condition categories, and seasonal rainfall, drought cycles, long-term temperature trends, and El Nino and La Niña cycles are categorized into weather patterns. Classifying and classifying climate zone, average sunlight hours, soil moisture content, air quality, and ultraviolet ray index as environmental and climate factors; further comprising:
The step of classifying the received economic information into at least one category according to the properties of the information,
Gross domestic product (GDP), unemployment rate, inflation rate, interest rate, and exchange rate are classified as macroeconomic indicators; commodity prices, agricultural input costs, agricultural productivity index, fuel price, and food demand index are classified as sectoral indicators; and global trade policy. , classifying the size of subsidies and government subsidies and real estate price changes as policy factors;
The step of classifying the received press information into at least one category according to the properties of the information,
Agricultural news, market trends, consumer trends, and product market reports are categorized into market and economic news, policy changes, food safety incidents, and supply chain issues are categorized into policy news, social media sentiment, health and nutrition trends, expert opinions, and promotions. and classifying advertisements into social and public opinion news,
The step of generating the feature vector is,
Generate weather information feature vectors based on weather condition scores, weather pattern scores, and environment scores,
Generates an economic information feature vector based on the macroeconomic score, sector indicator score, and policy factor score,
Generating a media information feature vector based on the market news score, policy news score, and public opinion news score;
The weather condition score is calculated based on a weighted sum of normalized values of temperature, rainfall, wind direction, wind speed, humidity, effective integrated temperature, and sunshine hours,
The weather pattern score is calculated based on a weighted sum of normalized values of seasonal rainfall, drought cycles, long-term temperature trends, and El Niño and La Niña cycles,
The environmental score is calculated based on a weighted sum of normalized values of climate zone, average sunshine hours, soil moisture content, and ultraviolet index,
The macroeconomic score is calculated as a weighted sum of normalized values of gross domestic product, unemployment rate, inflation rate, interest rate, and exchange rate,
The sector indicator score is calculated as a weighted sum of the normalized values of agricultural production input costs, agricultural productivity index, fuel price, and food demand index,
The policy factor score is calculated by subtracting the real estate price change rate from the common logarithm of the subsidy size and dividing it by the normalization index,
The market news score is calculated based on the weighted sum of market trends, consumer trends, and product market report scores,
The policy news score is calculated based on the weighted sum of policy change, food safety incident, and supply chain issue scores,
Characterized in that the public opinion news score is calculated based on a weighted sum of social media sentiment, health and nutrition trends, and promotional and advertising scores.
server. delete delete In claim 1,
The at least one step is,
Calculating a cultivation shortening factor using a weighted sum of the generated weather information scores; and
Further comprising: predicting the cultivation period by multiplying the calculated cultivation shortening factor by the observed maximum cultivation period,
server. In claim 4,
The at least one step is,
Further comprising: generating a guide indicating the sowing date as the date calculated backward from the cultivation period predicted from the shipment time of the crop,
server.

Images