WO2019216647A1

WO2019216647A1 - Method for predicting milk yields, tmr nutrient composition to achieve target milk yields, and tmr nutrient composition to achieve target cost, on basis of deep learning-based prediction models

Info

Publication number: WO2019216647A1
Application number: PCT/KR2019/005509
Authority: WO
Inventors: 박지환; 천선일
Original assignee: (주)씽크포비엘
Priority date: 2018-05-08
Filing date: 2019-05-08
Publication date: 2019-11-14
Also published as: KR102251643B1; KR20190128588A; KR102251643B9

Abstract

Provided is a method for predicting milk yields performed by a livestock farm management server that communicates with the outside through a communication network. The method for predicting milk yields of the present disclosure comprises the steps of: receiving, through the communication network, n data sets (n is an integer of 2 or more) retroactively from a reference date, accumulated from a past p-month prior date (p is an integer of 2 or more) to the reference date, wherein each data set of the n data sets includes state information data of a target cow to be managed, nutrient intake data of the target cow to be managed, and ambient state data on the basis of each specific date between the past p-month prior date and the reference date; and applying the received data sets and predicting milk yields that may be expected for the target cow to be managed for q months (q is an integer of 1 or more) from the reference date on the basis of a first prediction model. The state information data of the target cow to be managed includes the date of birth or the age in months of the target cow to be managed and the number of postpartum weeks on the specific date, the nutrient intake data of the target cow to be managed includes daily dry matter intake, water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake on the specific date, and the ambient state data includes average temperature and average humidity information of the specific date.

Description

Based on deep learning-based predictive models, yields, TMR nutritional composition to achieve target yields, and TMR nutritional composition prediction methods to achieve target costs

The present disclosure relates to a method for predicting future yield of livestock dairy cows based on deep learning based prediction models. The present disclosure is also based on a deep learning based prediction model to predict the nutritional composition for livestock dairy cows to achieve that target milk yield if there is a predetermined target milk yield, and further to establish the least cost TMR feed combination. It is about how to propose. The present disclosure also provides a method for producing livestock dairy cows based on a deep learning-based learning model that, if there is a predetermined target cost, provides the maximum milk yield expected from the cow within the target cost range and the cow for achieving the maximum milk yield. The present invention relates to a method for predicting a nutritional composition for a nutritional composition and further suggesting a minimum cost TMR feed combination for constructing the nutritional composition.

Yields for dairy cattle farmers are greatly influenced by milk production and feed costs, and milk production is greatly influenced by the nutritional composition of the cow (feed composition). Therefore, each farmer wants to predict what will be the future milk yield of each cow currently being raised, and may want to know what composition of feed is necessary to increase the milk yield of each cow.

However, conventionally, only information on the general TMR feed composition for production of milk yield is known without considering the condition of individual cows. There was no method for suggesting such a light.

Therefore, there is a need to provide a method for predicting future milk yield of individual cows or suggesting a TMR feed composition optimized for individual cows based on individual cow status information and other livestock environment information.

According to one aspect of the present disclosure, there is provided a production flow rate prediction method performed by a livestock farm management server that communicates with the outside through a communication network. The method for predicting oil flow rate according to the present disclosure includes n data sets accumulated from the reference date back to the previous p month (p is an integer of 2 or more) and accumulated from the reference date to the reference date through the communication network. Receive-each data set of the n data sets comprises status information data of the managed cow, nutritional intake data of the managed cow, and the surroundings, based on each specific date between the past p months ago and the reference date. Including state data, and based on a first predictive model, applying the received data set to expect for the managed cow for q months (q is an integer greater than or equal to 1) from the base date. Predicting the amount of milk produced. The state information data of the cow to be managed includes the birth date or month of the cow to be managed and postpartum parking at the specific date, and the nutritional intake data of the cow to be managed includes the daily building intake, water intake, Metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake, and the ambient state data includes average temperature and average humidity information for the specific date.

According to an embodiment of the present disclosure, the receiving of the accumulated n data sets comprises: receiving each data set from a livestock farm terminal capable of communicating through the communication network, and receiving each received data set. It may include accumulating and storing.

According to an embodiment of the present disclosure, the first prediction model is based on a Long-Short-Term Memory (LSTM) Recurrent Neural Network (RNN) algorithm, and is used for training using a plurality of milk flow data sets for each of a plurality of cows. May be generated by /. The plurality of milk yield data sets for each corresponding cow are accumulated for a predetermined period of time after the reference date, the nutrition information of the corresponding cow, the nutritional intake data of the corresponding cow, and the state data of the corresponding cow obtained at each of a plurality of reference days on a time series. And a milk production history data of the corresponding cow, wherein the status information data of the corresponding cow includes the birth date or month of the corresponding cow and postpartum parking at the reference day, and the nutrition intake data of the corresponding cow is a daily building on the reference day. Intake amount, water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake, the ambient state data may include the average temperature and average humidity information of the reference day.

According to an embodiment of the present disclosure, the number of cells of the LSTM RNN algorithm may be adjusted according to the p and q.

According to an embodiment of the present disclosure, the method may include receiving, from the livestock farm terminal connected to the communication network, an oil production rate prediction request for the management target cow, and the prediction target regarding the management target cow, through the communication network. The method may further include informing the livestock farm terminal of the milk production.

According to another feature of the present disclosure, there is provided a TMR proposal method for achieving a target oil flow rate performed by a livestock farm management server communicating with the outside through a communication network. The method of the present disclosure, through the communication network, from the livestock farm terminal, the status information of the management target cow on the reference date-the status information of the management target cow is the birth date or month of the management target cow, breeding, and postpartum at the reference date Receiving parking; receiving ambient state information from the outside through the communication network, wherein the ambient state information includes average temperature and average humidity information of the reference date; Receiving target milk flow rate information from a farm terminal, and based on a second prediction model, nutrition for achieving the target milk flow rate by applying the state information, the surrounding state information, and the target milk flow rate information of the management cow. Predicting the composition-The predicted nutritional composition, target building intake, target moisture intake, target Four energy consumption, including a target metabolic protein intake, intake MET goals, objectives LYS intake, calcium intake goals, and objectives of intake - including the.

According to one embodiment of the present disclosure, the method includes a daily nutritional intake at the reference date, the daily nutritional intake of the cattle to be managed, from the livestock farm terminal through the communication network, , Including water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake—and receiving milk yield information of the reference date. In addition, predicting the nutritional composition to achieve the target milk yield, based on the second prediction model, the state information of the management target cow, the surrounding condition information, the target milk yield information, the daily nutritional intake of the management target cow and the The method may include predicting a nutritional composition for achieving the target oil yield by applying the oil yield information of the reference date.

According to an embodiment of the present disclosure, the method may include receiving n feed types (n is an integer greater than or equal to 6) from the livestock farm terminal, and referring to the prepared feed nutrition / unit price table, the received n feeds A minimum cost feed combination to achieve the predicted nutritional composition, using a predetermined number of feeds selected from the variety, wherein the feed combination includes the predetermined number of feed types and the blending content of each feed type Comprising, and through the communication network, providing the configured feed combination to the livestock farm terminal.

According to an embodiment of the present disclosure, the second prediction model is a model generated by training according to a deep learning algorithm using a plurality of oil flow data sets for each of a plurality of cattle, and each of the plurality of oil flow data sets And status information of each corresponding cow on the basis of a specific date, surrounding condition information, nutritional intake amount of the corresponding cow, and milk yield information of the corresponding cow, wherein the status information of the corresponding cow includes the date of birth or month of birth of the corresponding cow, Postpartum parking on a date, and the nutritional intake of the corresponding cow includes daily dry matter intake, water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake on the specific day. The ambient state data may include average temperature and average humidity information of the specific date.

According to another feature of the present disclosure, there is provided a TMR proposal method for achieving the maximum oil flow rate performed by a livestock farm management server communicating with the outside through a communication network. According to the method of the present disclosure, the livestock farm terminal through the communication network, the status information of the management target cow on the reference date-The status information of the management target cow is the birth date or month of the management target cow, the breed and the Receiving postpartum parking; receiving ambient state information from the outside via the communication network, wherein the ambient state information includes average temperature and average humidity information of the reference date; Receiving target cost information from the livestock farm terminal, and based on the third predictive model, by applying the state information, the surrounding state information, and the target cost information of the management target cattle, the target cost range is achieved Estimating the maximum possible yield and the nutritional composition for achieving the maximum yield-the predicted zero It includes - composition, including the goals dry matter intake, water intake goals, objectives metabolizable energy intake, metabolic protein intake goals, objectives MET intake goals LYS intake, calcium intake goals, and objectives of intake.

According to one embodiment of the present disclosure, the daily nutritional intake on the reference date, the daily nutritional intake of the management target cattle, from the livestock farm terminal through the communication network, the daily nutritional intake of the cattle to be managed, water Including intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake—and receiving milk yield information on the reference date, achieved within the target cost range Predicting the maximum possible milk yield and the nutritional composition for achieving the maximum milk yield are based on the third prediction model, the status information of the managed cow, the surrounding condition information, the target cost information, the daily nutritional intake of the managed cow. And the maximum oil flow rate and the maximum oil flow rate by applying the oil flow rate information of the reference date. Predicting the nutritional composition to achieve.

According to an embodiment of the present disclosure, the method may include receiving n feed types (n is an integer greater than or equal to 6) from the livestock farm terminal, and referring to the prepared feed nutrition / unit price table, the received n feeds A minimum cost feed combination to achieve the predicted nutritional composition, using a predetermined number of feeds selected from the variety, wherein the feed combination includes the predetermined number of feed types and the blending content of each feed type Comprising, and through the communication network, providing the configured feed combination, the livestock farm terminal.

According to an embodiment of the present disclosure, the third predictive model is a model generated by training according to a deep learning algorithm using a plurality of oil flow rate data sets for each of a plurality of cattle, and each of the plurality of oil flow rate data sets And status information of each corresponding cow on the basis of a specific date, surrounding condition information, nutritional intake amount of the corresponding cow, and milk yield information of the corresponding cow, wherein the status information of the corresponding cow includes the date of birth or month of birth of the corresponding cow, Postpartum parking on a date, and the nutritional intake of the corresponding cow includes daily dry matter intake, water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake on the specific day. The ambient state data may include average temperature and average humidity information of the specific date.

According to the method according to an embodiment of the present disclosure, based on the status information of the individual cows and other environmental information such as temperature and humidity, etc. to predict the future milk production customized to the individual cows or to propose a TMR feed composition optimized for the individual cows A method is provided. According to the method of the present disclosure, it is possible not only to predict the amount of oil that can be obtained by maintaining the current nutritional status under a given environment, but also what nutritional composition is consumed to achieve the target amount of milk when the target amount of milk is determined. Furthermore, it can provide information on which TMR feed combinations will achieve such nutritional composition at the lowest cost. According to the method of the present disclosure, if there is a predetermined target cost, the maximum milk yield expected from the cow within the target cost range and what nutritional composition is to be ingested to the cow for achieving the maximum milk yield and further lower Within a range of costs, you can provide information on which TMR feed combinations will achieve such nutritional composition.

Therefore, according to the present disclosure, by optimizing the composition and providing nutrition for each cow by reflecting the characteristics of the individual cow or the surrounding environment, it is possible to maximize the production of milk and increase the profit of the livestock farm. According to the present disclosure, the livestock farmer can establish a total milk production plan and predict the farm income in advance, and can efficiently operate the farm.

1 is a diagram schematically showing the overall configuration of a livestock farm management system 100 according to an embodiment of the present disclosure.

FIG. 2 is a functional block diagram illustrating an exemplary functional configuration of the livestock farm management server 130 of FIG. 1.

FIG. 3 is a functional block diagram illustrating an exemplary functional configuration of the livestock farmhouse terminal 120 of FIG. 1.

4 is a diagram illustrating an exemplary simulation screen that may be displayed on the display unit 330 of the livestock farm terminal 120 of FIG. 2 according to an embodiment of the present disclosure.

FIG. 5 is a diagram generally illustrating input / output and utilization methods for each function of the livestock farm management server 130 according to an exemplary embodiment of the present disclosure.

Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In the following, when it is determined that there is a risk of unnecessarily obscuring the subject matter of the present disclosure, a detailed description of already known functions and configurations will be omitted. In addition, it should be understood that what is described below is only related to an embodiment of the present disclosure, but the present disclosure is not limited thereto.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. For example, a component expressed in the singular should be understood as a concept including a plurality of components unless the context clearly indicates the singular. In addition, in the specification of the present disclosure, the terms 'comprise' or 'having' are merely intended to designate that there exists a feature, a number, a step, an operation, a component, a part, or a combination thereof described on the specification. The use of the term is not intended to exclude the presence or addition of one or more other features or numbers, steps, actions, components, parts or combinations thereof.

In the embodiments described herein, 'block' or 'unit' refers to a functional part that performs at least one function or operation, and may be implemented in hardware or software or a combination of hardware and software. In addition, a plurality of 'blocks' or 'units' may be integrated into at least one software module and implemented as at least one processor, except for 'blocks' or 'units' that need to be implemented with specific hardware. .

In addition, all terms used herein, including technical or scientific terms, unless defined otherwise, have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. It is to be understood that commonly used terms are to be interpreted as having a meaning consistent with the contextual meaning of the related art, and are not to be construed as being excessively limited or extended unless expressly defined otherwise in the specification of the present disclosure. You should know

Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings.

1 is a diagram schematically showing the overall configuration of a livestock farm management system 100 according to an embodiment of the present disclosure. As illustrated, the livestock farm management system 100 includes a plurality of livestock farm terminal 110, a communication network 120, and a livestock farm management server 130.

According to one embodiment of the present disclosure, each livestock farm terminal 110 is a terminal for a livestock farm administrator, and may be any user electronic device having a wired or wireless communication function. In this figure, three livestock farmhouse terminals 110 are shown, but the present disclosure is not limited thereto. Each livestock farm terminal 110 may be a variety of wired or wireless communication terminals, including, for example, smart phones, tablet PCs, desktops, laptops, PDAs, digital TVs, and the like, and is not limited to specific forms.

According to an embodiment of the present disclosure, the farm manager may input and store / manage information related to each cow in the livestock farm, through the livestock farm terminal 110. According to one embodiment of the present disclosure, for example, on the livestock farm terminal 110, status information data about each cow in the farm, for example, the date of birth (monthly age), breed, postpartum parking of each cow (how many weeks have passed since birth) ), But not limited thereto, may be entered and stored. According to one embodiment of the present disclosure, on the livestock farmhouse terminal 110, for example, the environment information data of the livestock farmhouse surroundings, for example, average temperature or average humidity information, may be input and stored, but the present disclosure is limited thereto. It doesn't happen. According to one embodiment of the present disclosure, through the livestock farm terminal 110, nutritional intake information (e.g., TMR feed composition, ie, type and content information of each raw material feed, intake, Nutrition composition information and the like contained therein, but are not limited thereto, and the oil flow rate information (eg, the daily milk flow rate) may be input and stored / managed. According to an embodiment of the present disclosure, the nutritional intake information and the milk flow information of each cow input on the livestock farm terminal 110 may be continuously stored and managed in time series.

According to the exemplary embodiment of the present disclosure, the livestock farm terminal 110 may communicate with the livestock farm management server 130, that is, transmit and receive necessary information through the communication network 120. According to one embodiment of the present disclosure, the livestock farm terminal 110, through the communication network 120, information related to the status and environment of each cow in the farm, such as the age and postpartum parking information of each cow, temperature Or humidity information, nutritional intake information and milk flow information of each cow, can be transmitted to the livestock farm management server 130, and from the livestock farm management server 130, prediction information related to the cow, for example, milk production forecast information, Nutritional composition information, cost-optimized TMR feed combination information, etc., may be received to achieve a target yield. According to one embodiment of the present disclosure, the livestock farm terminal 110 may display various information received from the livestock farm management server 130 on the display.

According to an embodiment of the present disclosure, the communication network 120 may include any wired or wireless communication network, such as a TCP / IP communication network. According to one embodiment of the present disclosure, the communication network 104 may include, for example, a Wi-Fi network, a LAN network, a WAN network, an Internet network, and the like, but the present invention is not limited thereto. According to one embodiment of the present disclosure, the communication network 120 may include, for example, Ethernet, GSM, Enhanced Data GSM Environment (EDGE), CDMA, TDMA, OFDM, Bluetooth, VoIP, Wi-MAX, Wibro, and any other various wired or wireless. It may be implemented using a communication protocol.

According to an embodiment of the present disclosure, the livestock farm management server 130 may include a production flow prediction model that may be used to predict the production flow of the cow. According to one embodiment of the present disclosure, the livestock farm management server 130, the time series cumulative big data information related to cows collected in any of a variety of ways, such as the state information data (e.g. A circulating neural network based on big data information including postpartum parking, etc.), nutritional intake data (e.g., intake of each major nutrient, etc.), ambient state data (e.g., temperature and humidity information, etc.), and time series of milk yield information. It may include a model for predicting oil flow established according to the method. According to an embodiment of the present disclosure, the oil flow prediction model may be a model constructed based on the LSTM RNN algorithm.

According to an embodiment of the present disclosure, the livestock farm management server 130 may further include a nutritional composition prediction model that may predict the nutritional composition for achieving the target milk flow rate when the cow milk target milk flow rate is determined. According to one embodiment of the present disclosure, the livestock farm management server 130, the cow-related big data information collected by any of a variety of methods, such as state information data (for example, the age of the cow and postpartum parking, etc.), nutrition intake data (E.g., intake of each major nutrient, etc.), ambient condition data (e.g., temperature and humidity information, etc.), and big data information including oil flow rate information, and the like, It may include.

According to an embodiment of the present disclosure, the livestock farm management server 130 may also predict the maximum oil production that can be achieved within a range of the target breeding cost when a target breeding cost (eg, a target daily feed cost, etc.) is determined. Maximum oil flow prediction model. According to one embodiment of the present disclosure, the livestock farm management server 130, the cow-related big data information collected by any of a variety of methods, such as state information data (for example, the age of the cow and postpartum parking, etc.), nutrition intake data Based on big data information (e.g., intake of each major nutrient, etc.), ambient state data (e.g., temperature and humidity information, etc.), and oil flow rate information, a maximum oil yield prediction model constructed according to a deep learning method It may include.

FIG. 2 is a functional block diagram illustrating an exemplary functional configuration of the livestock farm management server 130 shown in FIG. 1. As shown, the livestock farm management server 130, the communication unit 210, the milk production flow rate prediction unit 220, the target oil production based nutrition composition prediction unit 230, feed nutrition / unit price table 240, minimum cost feed combination determination The unit 250, and the target cost-based maximum yield and nutritional composition predictor 260.

According to an embodiment of the present disclosure, the communication unit 210 may support the livestock farm management server 130 to communicate with the outside through the communication network 120. According to an embodiment of the present disclosure, the communication unit 210 may receive data from the communication network 120 according to a predetermined protocol, and data may be externally transmitted from the livestock farm management server 130 through the communication network 120. You can perform the necessary procedures to ensure that they are sent.

According to an embodiment of the present disclosure, the oil flow rate predictor 220 may include an oil flow rate prediction model. According to one embodiment of the present disclosure, the milk yield prediction model is based on a cyclic neural network method, in particular, a Recurrent Neural Network (RNN) Long-Short-Term Memory (LSTM) algorithm, based on a large number of cow-related data obtained by various methods. It may be generated by learning. As will be appreciated by those skilled in the art, the LSTM RNN algorithm is a model for causing the past data within a predetermined range of the input data accumulated in a time series to affect the result without forgetting. According to one embodiment of the present disclosure, the milk yield prediction model is a time series cumulative data set for a large number of cows, for example, each data set is obtained on each of a plurality of reference time series in time series for a given cow. Cow status information data (e.g., age and postpartum parking of cows on that reference day), nutrient intake data of the cow (e.g. daily building intake, water intake, metabolic energy intake, metabolic protein intake, MET intake) , Big data information including LYS intake, calcium intake, phosphorus intake, etc.), ambient state data (e.g., average temperature and humidity information for the reference day), and time series of milk yield information. It may be generated by learning. According to one embodiment of the present disclosure, the oil flow prediction model predicts a milk flow value for the next q months based on the cumulative data over the past p months, wherein the memory cell is determined according to the p period and the length of the q period. It can have a number.

According to an embodiment of the present disclosure, the milking flow predicting unit 220, together with the milking flow forecasting request for the predetermined cow received from the livestock farm terminal 110 through the communication network 120 and the communication unit 210, to the cow Cumulative data sets for a given period of time, e.g., data sets each obtained on any date in the middle, from a previous period up to the present (e.g., seven data obtained on the same day each month for the last seven months). Sets, etc., but are illustrative only and not limiting of this disclosure. Each data set belonging to the cumulative data set includes status information data (e.g., age and postpartum parking of the cow at that given date), nutritional data (e.g., the date) of the cow at each given date. Daily building intake, water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and intake of each of the major nutrients such as phosphorus intake, and ambient state data (e.g., average temperature for the day) And humidity information). According to one embodiment of the present disclosure, the milk yield estimator 220 applies the received cumulative data set to the milk yield forecasting model, so that the time series milk yield for a predetermined period of time (eg, 2 months, etc.) related to the cow. You can predict the flow.

According to an embodiment of the present disclosure, the target milk flow-based nutritional composition predictor 230 may include a target milk flow-based nutritional composition prediction model. According to one embodiment of the present disclosure, the target milk yield based nutritional composition prediction model may be generated by learning according to a deep learning algorithm based on a large number of cow-related data obtained by various methods.

According to one embodiment of the present disclosure, a target milk yield-based nutritional composition prediction model includes: cow-related big data information collected by any of a variety of methods, such as status information data (eg, age and postpartum parking of cows based on a specific date). Nutritional intake data (e.g. daily major daily nutrient intake, water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake, etc.), It may be a prediction model built according to a deep learning method based on big data information including ambient state data (eg, average temperature and humidity information of a specific date), and acid flow rate information.

According to an embodiment of the present disclosure, the target milk yield-based nutritional composition predicting unit 230, together with the target milk yield information about the predetermined cow, received from the livestock farm terminal 110 through the communication network 120 and the communication unit 210. , Status information (eg, age or date of birth, breed, postpartum parking, etc.) of the cow, and surrounding condition information (eg, average temperature and average humidity information of the livestock farm) may be received. . According to an embodiment of the present disclosure, the target milk yield based nutritional composition predicting unit 230 may include the above-mentioned predetermined cow related information and target milk yield information received from the livestock farm terminal 110 in the target milk yield based nutritional composition prediction model. By applying, the nutritional composition required to achieve the above target yield can be estimated. According to one embodiment of the present disclosure, the nutritional composition predicted by the target milk flow-based nutritional composition predicting unit 230 may include, for example, target dry matter intake, target water intake, target metabolic energy intake, target metabolic protein intake, target MET intake, Target LYS intake, target calcium intake, and target phosphorus intake.

According to one embodiment of the present disclosure, the feed nutrition / unit price table 240, each of the commercial feed raw materials for cows (eg, corn, soybeans, corn silly, Timothy dry, limestone, soybean meal, etc. 200 feed materials, respectively) It may be a database containing information on the amount of each ingredient and the unit price for each. According to an embodiment of the present disclosure, the livestock farm management server 130 may update the feed nutrition / unit price table 240 based on information received from the outside at a predetermined cycle.

According to one embodiment of the present disclosure, the minimum cost feed combination determination unit 250 may receive the nutritional composition prediction information obtained from the above-described target milk flow-based nutritional composition prediction unit 230. According to one embodiment of the present disclosure, the minimum cost feed combination determiner 250 may provide a minimum cost feed combination for achieving the above nutritional composition prediction information with reference to the feed nutrition / unit price table 240. In accordance with one embodiment of the present disclosure, the minimum cost feed combination determination unit 250, through the communication network 120 and the communication unit 210, the preferred feed material from the livestock farm terminal 110 (eg, mainly Information of a predetermined number of feed ingredients used) can be received. According to one embodiment of the present disclosure, the minimum cost feed combination determining unit 250, as described above, with reference to the feed nutrition / unit price table 240, a predetermined number selected from the predetermined number of feed ingredients received above Feed can be used to provide the least costly feed combination to achieve the predicted nutritional composition. The feed combination provided by the minimum cost feed combination determination unit 250 may be information including the blending content (%) of each feed type and each feed type. According to an embodiment of the present disclosure, the information on the feed combination determined by the minimum cost feed combination determination unit 250 may be transmitted to the livestock farm terminal 110 through the communication unit 210 and the communication network 120. .

According to one embodiment of the present disclosure, the target cost-based maximum yield and nutritional composition predicting unit 260 may include a target cost-based maximum yield and a maximum oil yield achieving nutritional composition prediction model. According to one embodiment of the present disclosure, the target cost-based maximum milk yield and maximum milk yield achieving nutrition composition prediction model may be generated by learning according to a deep learning algorithm based on a large number of cow related data obtained by various methods. have.

According to one embodiment of the present disclosure, a target cost-based maximum milk yield and maximum milk yield achieving nutritional composition prediction model includes: cow-related big data information, such as status information data (eg, based on a specific date) collected in any of a variety of ways. Age and postpartum parking of cows, nutritional data (e.g. daily dry matter, water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake) A prediction model built according to a deep learning method based on big data information including nutrient intake, etc.), ambient state data (eg, average temperature and humidity information of a specific date), and milk flow rate information. .

According to an embodiment of the present disclosure, the target cost-based maximum milk yield and maximum milk yield achieving nutrition composition predicting unit 250 may be provided to a predetermined cow received from the livestock farm terminal 110 through the communication network 120 and the communication unit 210. Status information about the cow (e.g., age or date of birth, breed, and postpartum parking), and surrounding condition information (e.g., average temperature and average humidity information of the livestock farm) And the like. According to an embodiment of the present disclosure, the target cost-based maximum milk yield and the maximum milk yield achieving nutrition composition predicting unit 250 may target the above-described predetermined cow-related information and target cost information received from the livestock farm terminal 110. Baseline maximum oil yield and maximum oil yield can be applied to a nutritional composition prediction model to predict the maximum achievable oil yield and the nutritional composition for achieving that maximum oil yield within the target cost range. According to the exemplary embodiment of the present disclosure, the nutritional composition predicted by the target cost-based maximum oil yield and maximum oil yield achieving nutrition composition predicting unit 250 may include, for example, target building intake, target water intake, target metabolic energy intake, and target metabolic protein. Intake, target MET intake, target LYS intake, target calcium intake, and target phosphorus intake.

According to one embodiment of the present disclosure, the above-described minimum cost feed combination determination unit 250 may receive nutritional composition prediction information obtained from the above-described target cost-based maximum oil yield and maximum oil yield achieving nutritional composition predictor 250. Can be. According to one embodiment of the present disclosure, the minimum cost feed combination determiner 250 may provide a minimum cost feed combination for achieving the above nutritional composition prediction information with reference to the feed nutrition / unit price table 240. In accordance with one embodiment of the present disclosure, the minimum cost feed combination determination unit 250, through the communication network 120 and the communication unit 210, the preferred feed material from the livestock farm terminal 110 (eg, mainly Information of a predetermined number of feed ingredients used) can be received. According to one embodiment of the present disclosure, the minimum cost feed combination determination unit 250, as described above, with reference to the feed nutrition / unit price table 240, a predetermined number selected from the predetermined number of feed ingredients received above ( For example, 5) feeds may be used to provide the least cost feed combination to achieve the above predicted nutritional composition. The feed combination provided by the minimum cost feed combination determination unit 250 may be information including the blending content (%) of each feed type and each feed type. According to an embodiment of the present disclosure, the information on the feed combination determined by the minimum cost feed combination determination unit 250 may be transmitted to the livestock farm terminal 110 through the communication unit 210 and the communication network 120. .

3 is a functional block diagram illustrating an exemplary functional configuration of the livestock farm terminal 110 shown in FIG. As illustrated, the livestock farm terminal 110 includes a communication unit 310, a data input unit 320, a display unit 330, and a history storage unit 340.

According to the exemplary embodiment of the present disclosure, the communication unit 310 may support the livestock farm terminal 110 to communicate with the outside through the communication network 120. According to the exemplary embodiment of the present disclosure, the communication unit 310 may receive data from the communication network 120 according to a predetermined protocol, and transmit data from the livestock farm terminal 110 to the outside via the communication network 120. You may perform the necessary procedures as much as possible.

According to an embodiment of the present disclosure, the data input unit 320 may include various types of data input devices including a predetermined data input device, such as a keyboard, a touch pad, a touch screen, a mouse, a scanner, a trackball, a camera, and the like. Can be. According to an embodiment of the present disclosure, the farm manager may input and store / manage information related to each cow in the farm, through the data input unit 320. According to one embodiment of the present disclosure, for example, through the data input unit 320, status information data about each cow, for example, the birth date (monthly age) of each cow, breed, postpartum parking (how many weeks have passed since birth), etc. Not limited) can be entered and stored. According to one embodiment of the present disclosure, through the data input unit 320, for example, barn environment information data of the corresponding livestock farm, for example, average temperature or average humidity information may be input and stored, but the present disclosure is limited thereto. It is not. According to one embodiment of the present disclosure, through the data input unit 320, nutritional intake information (e.g., TMR feed composition, namely, type and content information of each raw material feed, intake amount, and the like) of each cow being raised in the corresponding livestock farm The nutritional composition information may be included, but is not limited thereto, and the oil flow rate information (eg, the daily milk flow rate) may be input and stored / managed. According to one embodiment of the present disclosure, the status information, nutritional intake information, and milk yield information of each cow input on the data input unit 320 may be continuously stored and managed in time series in the history storage unit 340. have.

According to the exemplary embodiment of the present disclosure, the display unit 330 may include various display devices such as an LCD, an LED, a touch screen, and the like. According to the exemplary embodiment of the present disclosure, the display unit 330 may receive various information received from the livestock farm management server 130 through the communication network 120 and the communication unit 310, for example, on the livestock farm management server 130 described above. Obtained information such as the milk yield prediction value, the target milk yield-based nutritional composition, the feed composition (feedstock and content) to achieve the target milk yield-based nutritional composition, the target cost-based maximum milk yield prediction value, and the nutritional composition to achieve the maximum milk yield prediction value For example, it is possible to receive and display the feed composition for achieving the nutritional composition for achieving the maximum predicted milk flow rate.

According to the exemplary embodiment of the present disclosure, the history storage unit 340 may include information obtained in relation to each cow raised in the farm, for example, cow status information, surrounding condition information, and milk yield input through the data input unit 320. Information, such as nutritional intake information, from the livestock farm management server 130 through the communication network 120 and the communication unit 210, for example, a feed composition for achieving a milk production forecast value, a target milk production based nutrition composition, a target milk production based nutrition composition (Feedstock and content), the maximum cost forecast value based on the target cost, the nutritional composition to achieve the maximum milk yield forecast value, and the feed composition to achieve the nutritional composition to achieve the maximum yield forecast value can be accumulated and stored.

4 is a diagram illustrating an exemplary simulation screen that may be displayed on the display unit 330 of the livestock farm terminal 120 of FIG. 2 according to one embodiment of the present disclosure.

As shown, it can be seen that the current nutritional intake and the amount of milk are displayed on the left side of the screen. In the middle of the screen, there is a part of the large number of feed ingredients, which allows the farmer manager to input a desired number of feed ingredients, and at the bottom of the screen, two work request buttons, that is, a maximum oil yield optimum TMR formulation button ( Prediction of the maximum possible yield in the range of costs and the corresponding nutritional composition and optimum feed combination for achieving maximum yield, i.e. the button for requesting TMR composition, and the minimum cost and maximum effective TMR button (for nutritional composition to achieve a given target yield) The button for requesting the optimal feed combination, i.e. the TMR composition).

According to an embodiment of the present disclosure, the farm manager may input a predetermined number of feed ingredients preferred in the center of the screen of FIG. 3, select one of the two work request buttons at the bottom, and press the bottom simulate button. You can choose. According to one embodiment of the present disclosure, when the farm manager selects, for example, the maximum maximum oil yield optimal TMR combination button among two work request buttons, and selects the simulate button, the livestock farm terminal 110 includes a livestock farm management server ( 130, the maximum oil yield prediction result and the nutritional composition according to the cost range given in advance, and the optimum feed combination information for achieving the maximum yield (in accordance with one embodiment of the present disclosure, the feed combination information may be May be constituted by some feedstocks selected from among the selected feedstocks). According to an embodiment of the present disclosure, the received feed combination information may be displayed by being reflected in the amount of feed, the blending cost, and the blending cost next to each feed material at the top of the center of the screen of FIG. 3. On the right side of the screen of FIG. 3, the results of the simulation, for example, the maximum oil flow rate prediction result, the nutritional composition, and the cost information are displayed. It should be understood that the simulation screen of FIG. 4 is merely exemplary, and the present disclosure is not limited thereto.

FIG. 5 is a diagram generally illustrating input and output of each function of the livestock farm management server 130 according to an embodiment of the present disclosure, and a utilization plan. As shown, it will be easier to understand each of the above-described functions of the livestock farm management server 130 and related input and output data, and how to use each function.

As will be appreciated by those skilled in the art, the present disclosure is not limited to the examples described herein but may be variously modified, reconfigured and replaced without departing from the scope of the present disclosure. For example, the various techniques described herein may be implemented by hardware or software, or a combination of hardware and software. Thus, certain aspects or portions of an analysis machine for software safety analysis in accordance with the present disclosure may be implemented in one or more computer programs executable by a general purpose or dedicated microprocessor, micro-controller, or the like. A computer program according to an embodiment of the present disclosure may include a storage medium readable by a computer processor or the like, such as an EPROM, an EEPROM, a nonvolatile memory such as a flash memory device, a magnetic disk such as an internal hard disk and a removable disk, a magneto-optical disk, and It may be implemented in a form stored in various types of storage media, including a CDROM disk. Further, the program code (s) may be implemented in assembly or machine language, and may also be implemented in the form of transmission via electrical wiring or cabling, optical fiber, or any other form of transmission medium.

In the present specification, an exemplary embodiment has been mainly described with reference to various drawings, but other similar embodiments may be used. All modifications and changes that fall within the true spirit and scope of this disclosure are intended to be covered by the following claims.

Claims

An oil production forecasting method performed by a livestock farm management server communicating with the outside through a communication network,

Receive, through the network, n data sets accumulated from the date prior to the past p months (p is an integer of 2 or more) from the reference date back to the reference date (n is an integer of 2 or more)-of the n data sets. Each data set includes status information data of the managed cow, nutritional intake data of the managed cow, and ambient status data, based on each specific date between the past p months ago and the base date. , And

Based on a first predictive model, applying the received data set to predict milk yield that can be expected for the managed cow for q months (q is an integer of 1 or more) from the base date; ,

The state information data of the cow to be managed includes the date of birth or month of birth of the cow to be managed and postpartum parking at the specific date;

The nutritional intake data of the cows to be managed includes daily dry matter intake, water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake on the specific date,

The ambient state data includes average temperature and average humidity information of the specific date,

Production flow forecasting method.
The method of claim 1,

Receiving the accumulated n data sets,

Receiving each data set from a livestock farm terminal capable of communicating through the communication network, and

Accumulating and storing each received data set.
The method of claim 1,

The first prediction model is generated by training using a plurality of milk flow data sets for each of a plurality of cows according to a Long-Short-Term Memory (LSTM) Recurrent Neural Network (RNN) algorithm.

The plurality of milk yield data sets for each corresponding cow are accumulated for a predetermined period of time after the reference date, the nutrition information of the corresponding cow, the nutritional intake data of the corresponding cow, and the state data of the corresponding cow obtained at each of a plurality of reference days on a time series. Includes oil production history data of the corresponding cattle,

The state information data of the corresponding cow includes the date of birth or month of birth of the corresponding cow, and postpartum parking at the reference day,

The nutritional intake data of the corresponding cattle includes daily dry matter intake, water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake on the reference day,

The ambient state data includes the average temperature and average humidity information of the reference day, the oil quantity prediction method.
The method of claim 3,

And the number of cells of the LSTM RNN algorithm is adjusted according to the p and q.
The method of claim 1,

Receiving, through the communication network, a production flow forecasting request for the management target cow from a livestock farm terminal connected to the communication network, and

And a step of notifying the livestock farm terminal of the predicted milk yield with respect to the management target cow.
A TMR proposal method for achieving a target oil flow rate, which is performed by a livestock farm management server communicating with the outside through a communication network,

From the livestock farm terminal through the communication network, the status information of the management target cow on the reference date, wherein the status information of the management target cow includes the birth date or month of the management target cow, breed, and postpartum parking on the reference date. Receiving,

Receiving ambient state information from the outside through the communication network, wherein the ambient state information includes average temperature and average humidity information of the reference date;

Receiving target milking flow rate information from the livestock farm terminal through the communication network, and

Predicting a nutrition composition for achieving the target milk yield by applying state information, the surrounding condition information, and the target milk yield information of the management target cow, based on a second prediction model, wherein the predicted nutrition composition is a target A method of suggesting a TMR to achieve a target milk yield, including building uptake, target water intake, target metabolic energy intake, target metabolic protein intake, target MET intake, target LYS intake, target calcium intake, and target phosphorus intake.
The method of claim 6,

From the livestock farm terminal through the communication network, the daily nutritional intake on the reference date, the daily nutritional intake of the cow to be managed, the daily dry matter intake, water intake, metabolic energy intake, metabolic protein intake of the cow to be managed , Including MET intake, LYS intake, calcium intake, and phosphorus intake—and receiving milk yield information of the reference date;

Predicting the nutritional composition for achieving the target milk yield, based on the second prediction model, the state information of the management target cow, the surrounding condition information, the target milk yield information, the daily nutritional intake of the management target cow and the reference Predicting a nutritional composition to achieve the target yield by applying the date of flow information of the date, TMR proposal method for achieving the target yield.
The method of claim 6,

Receiving n feed types (n is an integer of 6 or more) from the livestock farm terminal,

Minimum cost feed combination to achieve the predicted nutritional composition using a predetermined number of feeds selected from the n received feed types, with reference to a previously prepared feed nutrition / unit price table, wherein the feed combination is the predetermined number of feeds Comprising the type and the blending content (%) of each feed type, and

Providing the configured feed combination to the livestock farm terminal through the communication network, TMR proposal method for achieving a target milk flow rate.
The method of claim 6,

The second predictive model is a model generated by training according to a deep learning algorithm using a plurality of flow rate data sets for each of a plurality of cattle,

Each of the plurality of milk flow rate data sets includes state information of each corresponding cow based on a specific date, surrounding state information, nutritional intake amount of the corresponding cow, and milk yield information of the corresponding cow,

The state information of the corresponding cow includes the date of birth or month of birth of the corresponding cow, and postpartum parking at the specific date.

The nutrient intake of the corresponding cattle includes daily dry matter intake, water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake on the specific date,

The ambient state information, the average temperature and average humidity information of the specific date, TMR proposal method for achieving a target oil flow rate.
As a method of suggesting a TMR for achieving a maximum oil production performed by a livestock farm management server communicating with the outside through a communication network,

Status information of the cow to be managed on a reference date from the livestock farm terminal through the communication network, wherein the status information of the cow to be managed includes a birth date or month of age of the cow to be managed, a breed, and postpartum parking on the reference date; Receiving the,

Receiving ambient state information from the outside through the communication network, wherein the ambient state information includes average temperature and average humidity information of the reference date;

Receiving target cost information from the livestock farm terminal through the communication network, and

On the basis of a third prediction model, by applying the state information, the surrounding state information, and the target cost information of the management target cow, to predict the maximum oil yield that can be achieved within the target cost range and the nutritional composition for achieving the maximum oil yield Step-the predicted nutritional composition includes a target dry matter intake, a target water intake, a target metabolic energy intake, a target metabolic protein intake, a target MET intake, a target LYS intake, a target calcium intake, and a target phosphorus intake ,

TMR proposal method for achieving maximum oil flow rate.
The method of claim 10,

From the livestock farm terminal through the communication network, the daily nutritional intake on the reference date, the daily nutritional intake of the cow to be managed, the daily dry matter intake, water intake, metabolic energy intake, metabolic protein intake of the cow to be managed , Including MET intake, LYS intake, calcium intake, and phosphorus intake—and receiving milk yield information at the reference date,

The step of predicting the maximum amount of oil that can be achieved within the target cost range and the nutritional composition for achieving the maximum amount of oil production is based on the third predictive model, the status information of the management cow, the surrounding condition information, the target cost information, Predicting the maximum yield and the nutritional composition for achieving the maximum yield by applying the daily nutritional intake of the cow and the reference date of the management target cow, the TMR proposal method for achieving the maximum yield.
The method of claim 10,

Receiving n feed types (n is an integer of 6 or more) from the livestock farm terminal,

Minimum cost feed combination to achieve the predicted nutritional composition, using a predetermined number of feeds selected from the n received feed types, with reference to a previously prepared feed nutrition / unit price table, wherein the feed combination is the predetermined number of feeds Comprising the type and the blending content (%) of each feed type, and

Providing the configured feed combination, through the communication network to the livestock farm terminal, TMR proposal method for achieving the maximum milk flow rate.
The method of claim 10,

The third prediction model is a model generated by training according to a deep learning algorithm using a plurality of oil flow data sets for each of a plurality of cattle,

Each of the plurality of milk flow rate data sets includes state information of each corresponding cow based on a specific date, surrounding state information, nutritional intake amount of the corresponding cow, and milk yield information of the corresponding cow,

The state information of the corresponding cow includes the date of birth or month of birth of the corresponding cow, and postpartum parking at the specific date.

The nutrient intake of the corresponding cattle includes daily dry matter intake, water intake, metabolic energy intake, metabolic protein intake, MET intake, LYS intake, calcium intake, and phosphorus intake on the specific date,

The ambient state data, the average temperature and average humidity information of the specific date, TMR proposal method for achieving the maximum oil flow rate.