JP2018007630A - Cultivation support device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide plant cultivation support devices which can obtain a harvest result close to an intended harvest result.SOLUTION: A cultivation support device for supporting plant cultivation comprises: a stress information measuring part that measures stress information of a cultivation target; an environmental information measuring part that measures environmental information of the cultivation target; a state estimating part that estimates an internal state of the cultivation target from the stress information and the environmental information; a growth state estimating part that estimates a growth situation of the cultivation target based on the estimation result of the internal state; and a control part that controls cultivation contents of the cultivation target based on the estimation result of the growth situation.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a cultivation support apparatus.

  As a technique for controlling plant cultivation, there is a technique for automatically diagnosing the state of a plant and controlling environmental conditions to a state suitable for plant growth. This technology is called Speaking Plant Approach (SPA), and it is easy to save labor of cultivation through recording of plant growth status and remote pathological diagnosis, and to realize high yield and high quality of agricultural products by applying stress. In recent years, it has attracted attention as something to make.

  Plant stress refers to the load placed on a plant by placing it in an environment different from the ideal growth environment for the plant. Within a plant placed under a stressed environment, various stresses can be applied depending on the environment. It is known that physiological changes occur.

  For example, fruit vegetables may be reduced in fruit size and yield when subjected to moisture stress due to insufficient moisture after fruit set. On the other hand, fruits given moderate water stress may have a high sugar content. There is an agricultural method that uses this property to produce high-value-added crops through moisture control. In recent years, various attempts have been made to automate this farming method.

  Japanese Patent Application Laid-Open No. 2007-306846 (Patent Document 1) discloses that an irrigation signal is generated when a ratio (projection area ratio) between a reference value of a projected area of a plant leaf and a current projected area falls below a threshold value. Techniques for controlling are disclosed. In this technology, plant stress is acquired from the image of the plant itself, and the information is fed back to the cultivation control to constantly monitor whether or not the intended stress is being applied, and give the plant appropriate stress. Cultivation control is performed.

  Japanese Patent Application Laid-Open No. 2012-55207 (Patent Document 2) discloses a technique for performing cultivation control based on a growth state in which a growth state of a plant is determined using a growth model that has acquired information as an input. In this technique, a harvesting plan is created from a growth model that uses an acquired image as an input, and a harvesting part is controlled based on the harvesting plan, thereby performing cultivation control in consideration of the current growth state of the plant.

  However, the existing growth model does not consider the stress on the crop, so even if you try to obtain the intended harvest result by cultivation control using the existing growth model, the actual harvest result is not the intended harvest result. It will deviate from the result. For this reason, there is a problem that crops with the intended yield and quality cannot be obtained by cultivation control using existing growth models.

  The objective of this invention is providing the plant cultivation assistance apparatus which can obtain the harvest result close | similar to the intended harvest result.

  In order to solve the above-described problem, one aspect of the cultivation support device of the present invention is a cultivation support device that supports plant cultivation, a stress information measurement unit that measures stress information of a cultivation target, and the environment of the cultivation target. Based on the environmental information measuring unit that measures information, the stress information, the state estimation unit that estimates the internal state of the cultivation target from the environmental information, and the growth state of the cultivation target based on the estimation result of the internal state It has a growth condition estimation part to estimate and a control part which controls the cultivation contents of the cultivation object based on the estimation result of the growth situation.

  According to one aspect of the present invention, it is possible to provide a plant cultivation support device that can obtain a harvest result close to the intended harvest result.

It is the schematic of the cultivation assistance system containing an example of the cultivation assistance apparatus which is embodiment of this invention. It is a block diagram of an example of the cultivation assistance apparatus which is embodiment of this invention. It is a functional block diagram of an example of the cultivation assistance apparatus which is embodiment of this invention. It is a flowchart of the process by an example of the cultivation assistance apparatus which is embodiment of this invention. It is a graph which shows an example of the change of the stress information in the modification of Example 2, and control information. It is a graph which shows an example of the estimation result of the internal state of the plant in the modification of Example 2. It is a graph which shows another example of the change of the stress information in the modification of Example 2, and control information. It is a graph which shows another example of the estimation result of the internal state of the plant in the modification of Example 2.

  Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. In the drawings, the same or corresponding components may be denoted by the same reference numerals and description thereof may be omitted.

  FIG. 1 is a schematic configuration diagram of a cultivation support system including an example (Example 1) of a cultivation support apparatus according to an embodiment of the present invention. In FIG. 1, the code | symbol 100 is the assistance apparatus which is an example of the cultivation assistance apparatus of this invention. The code | symbol PL is a plant which is an example of a cultivation object. The symbol SO is soil for growing the plant PL. The symbol WS is a part of a control device (irrigation device) that controls the cultivation of plants.

  The plant means a plant when the organism is classified into an animal, a plant, and a microorganism. In addition to plants that grow on the earth's surface, plants that grow in the sea and water can also be targeted.

  The main part of the plant cultivation support apparatus 100 is composed of a computer 110 and a sensor unit 120. In the plant cultivation support apparatus 100, as shown in FIG. 1, the sensor part 120 is installed in the above-ground part and the underground part with respect to the plant PL and the soil SO. Various types of information obtained by the sensor unit 120 are processed by the computer 110 to determine plant control information. The obtained control information is sent to the irrigation device WS, and based on this control information, the irrigation device WS is controlled to supply water to the plant PL.

  FIG. 2 is a configuration diagram of the plant cultivation support device 100. As shown in FIG. 1, the plant cultivation support apparatus 100 includes a computer 110 and a sensor unit 120 as main parts. The computer 110 is an example of a processing unit in the cultivation support device of the present invention.

  As shown in FIG. 2, the computer 110 includes an interface 111 for communicating with the sensor unit 120, a storage unit 112 that stores data obtained from the sensor unit 120 through the interface 111, and the storage unit 112. It has the calculating part 113 which processes data, and the recording part 114 which stores the obtained data and the calculation result in the calculating part 113. The interface 111, the storage unit 112, the calculation unit 113, and the recording unit 114 are connected by a bus 115.

  The sensor unit 120 includes at least a stress information acquisition sensor 121 and an environment information acquisition sensor 122. The sensor unit 120 acquires stress information and environmental information of the observation target plant PL in response to a predetermined time interval or an instruction from the computer 110.

  Specifically, the stress information acquisition sensor 121 in the sensor unit 120 can acquire information obtained by measuring the stress of the cultivation target as stress information. Plant stress is a load applied to a plant by being placed in an environment different from the ideal growth environment for the plant.

  As the stress information, for example, a difference between an ideal growth environment known in advance and an actual growth environment can be used. Specifically, the difference between the temperature for ideal growth and the current temperature can be used. In addition, it is known that various physiological changes occur in response to the environment in a plant placed in a stressed environment, so the amount of physiological change is measured and this is used as stress information. It may be used. In this case, for example, the degree of leaf wilting of the plant can be used as stress information.

  In this example, the stress information acquisition sensor 121 shoots the leaves of a plant PL to be cultivated as shown in FIG. 1, for example, measures the degree of withering and the degree of withering, and uses the measured information as stress information. Can be acquired. The stress information acquisition sensor 121 is an example of a stress information measurement unit in the cultivation support device of the present invention.

  Moreover, the environment information acquisition sensor 122 in the sensor part 120 can acquire the information which measured the surrounding environment of cultivation object as environmental information.

  The environmental information is data that can be measured by a sensor or the like from the above-ground part or the underground part in the place where the plant is grown. For example, data such as temperature and soil humidity. In addition to data measurable by a sensor or the like, information such as a weather forecast may be used. In addition, information on cultivation control such as the amount of watering or the set temperature of heating in house cultivation may be used as environmental information.

  In this example, as shown in FIG. 1, the environment information acquisition sensor 122 measures the air temperature in the ground part, measures the humidity of the soil SO in the underground part, and obtains the measured information as environment information. Can do. The environment information acquisition sensor 122 is an example of an environment information measurement unit in the cultivation support device of the present invention.

  The acquired stress information and environment information are transmitted to the computer 110 and stored in the storage unit 112 via the interface 111. The stored stress information and environment information are read from the storage unit 112 by the calculation unit 113. The internal state of the plant PL is estimated from the read stress information and environmental information, and the growth state is estimated based on the estimated internal state.

  And the control content of cultivation (henceforth the cultivation control content) is judged based on the estimated growth condition. When it is determined that the cultivation control is necessary, the calculation unit 113 transmits the contents of the cultivation control as a control signal to the control device or notifies a preset contact using a not-shown notification unit. Thereby, the cultivation content of plant PL is controlled.

  Here, as an example of the internal state of the plant, sensitivity to environmental information of the plant can be used. Such sensitivity is, for example, information on how much the amount of plant stress changes when the temperature changes by 1 ° C. In addition, a part of the existing growth model may be used as the internal state of the plant. For example, when an existing growth model includes a ratio of the balance between the production of a plant photosynthesis product and the growth of leaves and stems, the ratio can be considered as the internal state of the plant.

  The growth status indicates, for example, the total leaf area of the plant, the height of the shoot apex (the tip of the stalk), the total number of buds, and the like.

  FIG. 3 is a functional block diagram of the plant cultivation support apparatus 100. As shown in FIG. 3, the plant cultivation support device 100 includes a computer 110 and a sensor unit 120. The computer 110 includes an internal state estimation unit 116, a growth simulation unit 117, and a cultivation control unit 118.

  The internal state estimation unit 116 can estimate the internal state of the plant PL by using information from the sensor unit 120 as an input. The growth simulation unit 117 can estimate the growth status of the plant PL by using the estimation result of the internal state estimation unit 116 and the environment information obtained by the sensor unit 120 (environment information acquisition sensor 122) as inputs. Furthermore, the cultivation control unit 118 can control the cultivation content of the plant PL by determining the cultivation control content of the plant PL based on the estimation result of the growth simulation unit 117.

  In FIG. 3, the sensor unit 120 first acquires stress information and environmental information of the plant PL to be cultivated. The acquired stress information and environmental information are transmitted to the computer 110. In the computer 110, the internal state estimation unit 116 estimates the internal state of the plant PL to be cultivated using information from the sensor unit 120 as an input. In addition, as an input for estimating the internal state in the internal state estimation unit 116, a past output history of the growth simulation unit may be used.

  The output from the internal state estimation unit 116 or the environment information acquired by the environment information acquisition sensor 122 is input to the growth simulation unit 117, and the growth simulation unit 117 estimates the growth status.

  And the cultivation control part 118 judges the cultivation control content by using the estimation result of the growth simulation in the growth simulation part 117 as an input. When the cultivation control unit 118 determines that the cultivation control is necessary, the notifying unit (not shown) installed inside or outside the computer 110 transmits the cultivation control content as a control signal to the control device or a preset communication. Notify first. Thereby, cultivation control of plant PL which is a cultivation object is performed.

  In addition, the internal state estimation part 116, the growth simulation part 117, and the cultivation control part 118 are each examples of a state estimation part, a growth condition estimation part, and a control part in the cultivation assistance apparatus of this invention.

  FIG. 4 is a diagram (flow chart) illustrating a processing flow by the plant cultivation support apparatus 100 according to the first embodiment. In FIG. 4, first, stress information and environmental information are acquired (step S1201). Specifically, the sensor unit 120 acquires stress information and environmental information of the plant PL at a predetermined timing. The predetermined timing is a timing at which a trigger signal is received from the computer 110 or a timing determined based on a preset time or time interval.

For example, the environment information x _t acquired at time t can be expressed by the following equation.

この式において、ｘ_ｔ ^ｎ（ｎは１以上の整数）は、例えば、気温、湿度のような大気の情報や、土壌温度などの土壌の情報等とすることができる。

In this equation, x _t ⁿ (n is an integer of 1 or more) can be air information such as air temperature and humidity, soil information such as soil temperature, and the like.

  Next, the internal state of the plant is estimated from stress information and environmental information (step S1202). In this step, the internal state estimation unit 116 in the computer 110 estimates the internal state of the plant PL using the stress information and environment information received from the sensor unit 120. Regarding the estimation of the internal state, the relationship between the stress information and the internal state can be defined in advance, and the stress information acquired in this loop can be applied.

  In addition, when estimating the internal state of the plant PL in this loop, the data information acquired in step S1201 has all the predetermined items (for example, the amount of watering, the set temperature of heating, the amount of leaves taken, etc.). There is no need. That is, past data may be used, or a modeling technique that allows missing values may be used. Further, in addition to the stress information, the internal state may be estimated by using the environmental information and the estimation result of the past growth state.

  Next, the growth status is estimated using the growth model from the stress information, the environmental information, and the estimation result of the internal state of the plant PL (step S1203). In this step, the growth simulation unit 117 in the computer 110 estimates the growth status using a predetermined growth model from the environmental information received from the sensor unit 120 and the internal state of the plant PL estimated in step S1202.

  Here, the growth model is a mathematical expression of how a plant grows or grows. For example, the growth model is represented by a function of a daily growth rate using information on the average daily temperature as a variable.

  Next, the cultivation contents of the plant PL to be cultivated are controlled based on the growth situation estimation result (hereinafter also referred to as the growth situation estimation result). Specifically, the cultivation control content is first determined from the growth status estimation result (step S1204). In this step, the cultivation control unit 118 in the computer 110 determines whether or not the cultivation control for the plant PL is necessary based on the estimated growth state. The cultivation control content can be judged by the following method.

For example, it is assumed that there is a harvest result using an output of an existing growth model in which stress or the like is not considered as the harvest result H to be achieved. On the other hand, when the time in this loop is t, the current predicted H _t ^′ of the harvest result is calculated from the growth state estimation result in step S1203. There is a high possibility of a difference between H and H _t ^′ . If this difference is diff (H, H _t ^′ ), diff (H, H _t ^′ ) is a function such as stress and environmental information. The cultivation control content is determined so that the value of diff (H, H _t ^′ ) becomes smaller in the next step.

As a specific method, a method of selecting a case where diff (H, H _t ^′ ) becomes small can be used from among combinations of cultivation control contents that can be implemented.

  When the content of cultivation control is C, C can be expressed as follows. Specifically, the cultivation control content C is represented by the amount of operation for each item of the cultivation control, and the operation amount can be summarized for each item one by one and expressed as follows.

Ｃ_ｉとしては、例えば、水遣りの量、暖房の設定温度、葉を取る量など、栽培制御内容を示すことができる。栽培制御内容には、想定し得る内容が複数通りある。例えば、水遣りの量を０，０．５，１の３通りとし、暖房の設定温度を２０〜２５℃までの６通りとすると、このときの栽培制御内容は１８通り想定できる。これらを全て列挙すると、以下のように表すことができる。

As C _i , for example, it is possible to indicate the contents of cultivation control such as the amount of watering, the set temperature of heating, and the amount of leaves taken. There are a plurality of possible contents of the cultivation control contents. For example, if the amount of watering is 0, 0.5, 1 and the heating set temperature is 6 to 20 to 25 ° C., 18 types of cultivation control contents can be assumed. All of these can be expressed as follows.

このＣ^ｋの中から、diff（Ｈ，Ｈ_ｔ ^´）を最小とする値を選択する。なお、Ｃ^ｋの要素については、栽培施設において制御可能な項目の内容や、栽培を行っている季節によって、随時許容できる要素のみに制限しても良い。

A value that minimizes diff (H, H _t ^′ ) is selected from C ^k . It is to be noted that the elements of C ^k, and the contents of the control items that can be in cultivation facility, depending on the season, which is doing the cultivation, may be limited only to the elements that can be tolerated at any time.

C ^k has a data structure as shown in Table 1 below, for example.

次にステップＳ１２０５に進み、栽培制御が必要か否か判断される。栽培制御が必要（Ｙｅｓ）と判断されれば、ステップＳ１２０６に進み、栽培制御が不要（Ｎｏ）と判断されればステップＳ１２０１に進む。

Next, it progresses to step S1205 and it is judged whether cultivation control is required. If it is determined that cultivation control is necessary (Yes), the process proceeds to step S1206. If it is determined that cultivation control is not necessary (No), the process proceeds to step S1201.

  If it is determined in step S1205 that the cultivation control is necessary, then the cultivation control content is transmitted as a notification or control signal (step S1206). The cultivation control unit 118 in the computer 110 can transmit the contents of cultivation control to the control device as a control signal or notify a preset contact using a not-shown notification means.

  The transmission contents of the control signal include, for example, a set value of a device (such as the irrigation device WS in FIG. 1) installed in the cultivation facility, an instruction to perform irrigation (hereinafter also referred to as “watering”), and harvesting. As a notification method, in addition to notification by email or telephone, notification may be performed visually using a smartphone dedicated application or display. Moreover, you may employ | adopt the notification method of sounding a buzzer or making a revolving light shine.

  As described above, steps S1201 to S1206 are a series of processing flows performed in association with the acquisition of stress information and environmental information once. When these steps are completed, it is determined whether or not to end the control (step S1207). If not, the process returns to step S1201, and after a predetermined time or time interval, steps S1201 to S1206 are repeated.

  Thus, in the plant cultivation assistance apparatus 100 of Example 1, the growth condition in consideration of stress is estimated from the stress of the plant to be cultivated and the surrounding environment information, and the control content of the cultivation is based on the estimated growth condition. By determining, cultivation control can be performed appropriately. Therefore, it is possible to stably and easily produce intended high yield and high quality crops. Therefore, according to the plant cultivation support apparatus 100 of Example 1, the cultivation control for obtaining the harvest result close to the intended yield and quality becomes possible.

  Specifically, it is known that crops cultivated under water stress by intentionally deficient water content are smaller in size than crops that do not. Since cultivation control in consideration of such stress information can be performed, an intended high-yield and high-quality crop can be obtained.

  Next, another example (Example 2) of the cultivation support apparatus according to the embodiment of the present invention will be described. Note that in the second embodiment, description of portions common to the first embodiment may be omitted. In Example 2, as the internal state of the plant PL, the content of cultivation control when the sensitivity of the plant PL to the environment is used is determined. By determining the content of cultivation control based on the sensitivity of the plant to the environment, it is possible to give the stress necessary to obtain the desired result.

The environment indicates items of environmental information acquired from the environmental information acquisition sensor 122 (for example, temperature, humidity of soil SO, etc.). The sensitivity of the plant to the environment can be determined as follows. First, the measured value of stress at time t is expressed as S _t. The stress value St is expressed as a function using environmental information or the like. In other words, the use of notation environmental information x _t in the first embodiment, can be represented by the following relation.

環境情報のうちの一つｘ^ｊの、時刻ｔにおける値をｘ_ｔ ^ｊと表す。このとき、ｘ_ｔ ^ｊの変化量とＳ_ｔの変化量の関係を、環境情報ｘ^ｊに対する感度ａ_ｔ ^（ｘｊ）とする。

One ^{x j} of the environmental information, the value at time t is expressed as _x ^{t j.} At this time, the relationship between the change amount of x _t ^{j and} the change amount of _St is defined as sensitivity a _t ^(xj) with respect to the environment information x ^j .

  Under this condition (the internal state of the plant is estimated from the stress information and environmental information acquired in step S1201 (step S1202). In this example, the sensitivity of the plant to the environment is estimated in step S1202. For example, a method of estimating using a regression equation can be used.

In this example, the stress explanatory variables, the function f _a as follows for the purpose variable internal state in advance derived by a preliminary experiment or the like, fitting the obtained stress information at the same time.

ｆ_ａとしては、例えば、Ｓのｘ^ｊに関する逆関数や、ｘ^ｊに関する偏微分などを用いることができる。なお、ストレスに加えて、環境情報などを説明変数として用いても良い。上記の説明変数における「Ｓ，_・・・」の表記は、Ｓを説明変数として用いることは必須であるが、他に環境情報等を用いても良いことを表す。この場合、説明変数に欠損があると、回帰式への代入ができない。そのため、取得した情報の欠損を予め補っておくことや、直前の時刻の値を代わりに用いることなどが好ましい。

The f _a, for example, and the inverse function for ^{x j} of S, or the like can be used partial differential related ^{x j.} In addition to stress, environmental information or the like may be used as an explanatory variable. The notation “S, _... ” In the above explanatory variables indicates that it is essential to use S as an explanatory variable, but other environmental information or the like may be used. In this case, if there is a missing explanatory variable, it cannot be substituted into the regression equation. For this reason, it is preferable to compensate for missing information in advance, or to use the value of the previous time instead.

  Next, a growth state is estimated using a growth model from stress information, environmental information, and the internal state estimation result of a plant (step S1203). In this example, in step S1203, the simulation result of the growth model when there is no stress is R, and the simulation result when stress is applied is R ′. The following relational expression is derived in advance by a preliminary experiment or the like, and by applying the stress information acquired at the same time, a growth state estimation result considering the stress can be obtained.

なお、ストレスに加えて、Ｇや環境情報などを説明変数として用いても良い。上記の説明変数における「・・・」の表記は、Ｓを説明変数として用いることは必須であるが、他にＧや環境情報等を用いても良いことを表す。

In addition to stress, G and environmental information may be used as explanatory variables. The notation “...” In the above explanatory variables indicates that it is essential to use S as an explanatory variable, but G, environmental information, etc. may be used in addition.

The cultivation control content is determined from the growth state estimation result (step S1204). In this example, based on the internal state estimated in step S1202, the control content is determined such that diff (H, H _t ^′ ) becomes small. As a method for determining the control content, for example, a method of selecting a case where diff (H, H _t ^′ ) is small from among combinations of cultivation control content that can be implemented can be used.

This method is based on the premise that the amount of change in environmental information is clarified for each control content of cultivation. That is, the variation of environmental information in one C _i of cultivation the control items can be expressed by the following equation.

ステップＳ１２０２で推定した内部状態を用いることで、各々のＣ^ｋに対し、ストレスの予測値Ｓ_ｔ＋１が算出できる。これを用いることで、各々のＣ^ｋに対し、diff（Ｈ，Ｈ_ｔ＋1 ^´）が計算できる。この中から、が最小となるようなＣ^ｋを選択する。そして、生育状況（ステップＳ１２０４、Ｓ１２０５）、栽培制御内容を通知ないし制御信号として送信する（ステップＳ１２０６）。

By using the internal state estimated in step S1202, the for each of the ^{C k,} the predicted value _{S t + 1} of the stress can be calculated. By using this, diff (H, H _{t + 1} ^′ ) can be calculated for each C ^k . From these, C ^k that minimizes is selected. Then, the growth status (steps S1204 and S1205) and the contents of cultivation control are transmitted as notifications or control signals (step S1206).

  Thus, in the form of Example 2, since the control content of cultivation is determined using the sensitivity to the environment of the plant, it becomes possible to give an appropriate stress to the plant, and the harvest is closer to the intended yield / quality. The result can be obtained. Moreover, since the internal state of the cultivation object can be estimated by preliminary experiments and simple data analysis, the cultivation object can be easily controlled.

As a modification of the second embodiment, a model having a hidden state may be used for estimating the internal state of the second embodiment. The estimation of Example 2, the internal state of the above, previously evident and which had been, but using the change amount of the equation of variation and S _t of x _t, in the modification of the second embodiment, past the stress values The internal state can be estimated by using the environment information and the model having the hidden state. In this case, the hidden state of the model is defined as an internal state.

  Here, the hidden state refers to what is assumed to be changing with time, although it cannot be observed directly like environmental information. For example, the information on whether a plant is currently in “vegetative growth” or “reproductive growth” is not information that can be measured using any measuring means, but it is important for cultivation. Is defined.

For each of the elements of the environmental information x _t, and there is at least one corresponding internal state. That is, the internal state at time t is expressed as a _t, it shall be defined as follows.

内部状態は、時刻が更新された際、１ステップ前の内部状態から一定の手続きにより更新されるとする。つまり、内部状態は、以下のような関数ｒを用いて更新される。

When the time is updated, the internal state is updated by a certain procedure from the internal state one step before. That is, the internal state is updated using a function r as follows.

この内部状態を用いて、ストレス値Ｓが次の式で表されるとする。

It is assumed that the stress value S is expressed by the following equation using this internal state.

ストレス値、および内部状態の表現は、確率分布などでも良い。その場合は、分布の表記を用いることとする。内部状態の推定方法としては、例えば、内部状態フィルタを用いた方法がある。フィルタは、１ステップ前の時点において、関数ｒを用いて現在のステップの内部状態を予測し、関数ｓを用いて観測値の推定値を計算する。その後、現在のステップで実際に得られる観測値の実測値に基づき、現在のステップの内部状態を修正する。つまり、ストレスの予測値をＳ_ｐ、実測値をＳ_ｍ、内部状態の予測値をａ_ｐ、修正後の内部状態をａ_ｒとすると、一般には、状態フィルタは以下のような関数となる。

The expression of the stress value and the internal state may be a probability distribution. In that case, distribution notation will be used. As an internal state estimation method, for example, there is a method using an internal state filter. The filter predicts the internal state of the current step using the function r and calculates the estimated value of the observed value using the function s at a time point one step before. Thereafter, the internal state of the current step is corrected based on the actual measurement value actually obtained in the current step. That is, assuming that the predicted value of stress is S _p , the measured value is S _m , the predicted value of the internal state is a _p , and the corrected internal state is a _r , the state filter generally has the following function.

このようなフィルタを用いることにより、過去のストレス情報及び環境情報と、隠れ状態を持つモデルとを用いた内部状態の推定が可能になる。

By using such a filter, it is possible to estimate the internal state using past stress information and environmental information and a model having a hidden state.

  5 and 6 are graphs showing a result of estimating the internal state of the plant PL from past stress information, control information, and a plurality of environment information (environment information 1 and environment information 2) on a certain day. Among these, FIG. 5 shows the change of stress information (measured value of stress) and control information (cultivation control value), and FIG. 6 shows the sensitivity to control and the sensitivity to environmental information 1 and 2 as the estimation result of the internal state. Showing change.

  7 and 8 are graphs showing results of estimating the internal state of the plant PL from past stress information, control information, and a plurality of environment information (environment information 1 and environment information 2) on another day. Among these, FIG. 7 shows the change of stress information (measured value of stress) and control information (cultivation control value), and FIG. 8 shows the sensitivity to control and the sensitivity to environmental information 1 and 2 as the estimation result of the internal state. Showing change.

  As described above, by estimating the internal state using the past stress information and environment information and the model having the hidden state, it is not necessary to prepare in advance information necessary for estimating the internal state by a preliminary experiment or the like. In addition, since the internal state can be estimated even if there are missing values in the objective variable and the explanatory variable, cultivation control closer to the intended yield and quality is possible even if the acquired information is missing.

  Next, another example (Example 3) of the cultivation support apparatus will be described. Note that in the third embodiment, description of portions common to the first embodiment may be omitted. In Example 3, the contents of cultivation control when items included in the growth model are used as the internal state of the plant are determined. By determining the control content of cultivation based on the items included in the growth model, it is possible to give the stress necessary for obtaining the intended result.

  Here, it is assumed that the growth model is represented by a function of one or more items. The items of the growth model are, for example, state variables and are considered to change due to stress. When a collection of items of growth models that can be changed by stress is set as a and the output of the growth model is expressed as G, the following relationship is obtained.

ａには、生育モデルの項目の一部もしくは全てが含まれるとする。生育モデルの関数ｇには、ａの他に入力があっても良い。ストレスの変化から、内部状態ａの変化を推定する。

It is assumed that a includes some or all of the items of the growth model. The function g of the growth model may have an input in addition to a. The change in the internal state a is estimated from the change in stress.

Under this condition, the internal state of the plant is estimated from the stress information and environment information acquired in step S1201 (step S1202). In this example, in step S1202, a change in an item included in the growth model is estimated as an internal state. As an estimation method, for example, it can be estimated using a regression equation. The following function f _j with stress as an explanatory variable and internal state as an objective variable is derived in advance by a preliminary experiment or the like, and stress information acquired at the same time is applied.

なお、ストレス情報に加えて、環境情報や、前の時刻に推定した内部状態などを説明変数として用いても良い。説明変数における「Ｓ，_・・・」の表記は、Ｓを説明変数として用いることは必須であるが、他に環境情報等を用いても良いことを表す。この場合、説明変数に欠損があると、回帰式への代入ができない。そのため、取得した情報の欠損を予め補っておくこと、直前の時刻の値を代わりに用いることなどが好ましい。

In addition to the stress information, environmental information, an internal state estimated at a previous time, or the like may be used as an explanatory variable. The notation “S, _... ” In an explanatory variable indicates that it is essential to use S as an explanatory variable, but other environmental information or the like may be used. In this case, if there is a missing explanatory variable, it cannot be substituted into the regression equation. For this reason, it is preferable to compensate for missing information in advance, or to use the value of the previous time instead.

  The growth state is estimated using the growth model from the stress information, the environmental information, and the internal state estimation result of the plant (step S1203). In step S1203, the growth status is estimated by applying the item included in the growth model estimated as the internal state in step S1202 to the function g of the growth model.

  And the cultivation control content is judged from the obtained growth situation estimation result (step S1204), and when it is judged that the cultivation control is necessary (step S1205), the cultivation control content is notified or transmitted as a control signal (step S1206). ).

  According to the form of the third embodiment, the deviation of the growth model in consideration of stress can be naturally incorporated into the model structure, so that the deviation of the growth model can be estimated with high accuracy. Therefore, cultivation control for obtaining the intended yield and quality can be performed more accurately.

  Moreover, as modification 1 of Example 3, the existing growth model can be used as a growth model, and the contents of cultivation control can be determined using the state variables and parameters included in the growth model as the internal state of the plant. .

  In Modification 1 of Example 3, as a specific growth model, existing growth models “TOMGRO” (JONES JW (Univ. Florida, Gainesville), DAYAN E (ARO, Hanegev), ALLEN LH (ARS, USDA) , FL), Trans ASAE (Am Soc Agric Eng) Vol. 34, No. 2, pages 663-672, published in March 1991). “TOMMGRO” is a growth model developed mainly in the United States for predicting the yield of tomatoes.

In “TOMGRO”, the temperature, the CO ₂ concentration in the air, and the amount of light are used as inputs. Within “TOMGRO”, there are a plurality of state variables, which are updated at each step, and the predicted yield value is calculated using the updated value. In the case of “TOMGRO”, the input is also handled as a part of the state variable. Moreover, there are a plurality of parameters, and these parameters are usually set to a constant value throughout the cultivation period.

  In the present embodiment, among the state variables and parameters included in “TOMGRO”, those affected by stress are defined as the internal state of the plant, and the internal state is estimated from the measured stress value.

  For example, among the state variables and parameters of TOMGRO, there are temperature (state variables) and constants (parameters) that determine the balance of the amount of photosynthetic product distributed to each part. It is not limited to. Since these default values are values that do not take stress into consideration, it is preferable to update using the stress information. As a method of updating, for example, a model having regression and hidden states can be used, but is not limited to this method.

  According to the form of the modification 1 of Example 3, the growth condition of plants, such as a tomato, can be estimated in consideration of stress using the existing growth model. That is, even when an existing growth model is used, the cultivation control considering the stress can be performed, so that a harvest result close to the intended harvest result can be obtained.

Furthermore, as a second modification of the third embodiment, a model having a hidden state may be used for the estimation of the internal state of the third embodiment. In the third embodiment, the relational expression of a ^j and S that has been clarified in advance is used for the estimation of the internal state. However, in the second modification of the third embodiment, the past stress value, the past a ^j value, The internal state can also be estimated by using a model having a hidden state.

In the second modification of the third embodiment, first, it is assumed that there is at least one corresponding internal state for each a ^j (jεM). That is, the internal state at time t is expressed as a _t, it shall be defined as follows.

内部状態は、時刻が更新された際、１ステップ前の内部状態から一定の手続きにより更新されるとする。つまり、以下のような関数ｒを用いて更新される。

When the time is updated, the internal state is updated by a certain procedure from the internal state one step before. That is, it is updated using the following function r.

この内部状態を用いて、ストレス値Ｓが次の式で表されるとする。

It is assumed that the stress value S is expressed by the following equation using this internal state.

ストレス値、および内部状態は、確率分布などで表現しても良い。その場合は、分布の表記を用いるのが好ましい。内部状態の推定方法としては、例えば、内部状態フィルタを用いた方法を用いることができる。フィルタは、1ステップ前の時点において、関数ｒを用いて現在のステップの内部状態を予測し、関数ｓを用いて観測値の推定値を計算する。

The stress value and the internal state may be expressed by a probability distribution or the like. In that case, it is preferable to use the notation of distribution. As an internal state estimation method, for example, a method using an internal state filter can be used. The filter predicts the internal state of the current step using the function r and calculates the estimated value of the observed value using the function s at a time point one step before.

Thereafter, the internal state of the current step is corrected based on the actual measurement value actually obtained in the current step. That is, assuming that the predicted value of stress is S _p , the actually measured value is S _m , the predicted value of the internal state is a _p , and the corrected internal state is a _r , the state filter generally has the following function.

このようなフィルタを用いて、植物の内部状態を推定することができる。実施例３の変形例２のように、隠れ状態を持つモデルにより内部状態を推定する場合、上述のように過去のストレス情報及び環境情報を用いることができるため、予備実験等により内部状態の推定に必要な情報を予め用意する必要がない。

Using such a filter, the internal state of the plant can be estimated. As in the second modification of the third embodiment, when the internal state is estimated by a model having a hidden state, since the past stress information and environment information can be used as described above, the internal state is estimated by a preliminary experiment or the like. It is not necessary to prepare information necessary for

  In addition, in the case of estimation using a model having a hidden state, the internal state can be estimated even if there are missing values in the objective variable and explanatory variable, so the intended yield and quality can be achieved even if the acquired information is missing. Closer cultivation control is possible. In addition, as means for using the hidden state, for example, there is machine learning or the like, but the internal state may be estimated by other methods, and the means is not particularly limited.

  As still another example (Example 4) of the cultivation support apparatus, a method of using an index other than the determination of the contents of cultivation control and the output of the growth model will be described. Note that in the fourth embodiment, description of portions common to the first embodiment may be omitted. In Example 4, when determining the cultivation control content, an index other than the output of the growth model is used.

  In step S1201 to S1204 in FIG. 4, when judging the cultivation control content from the growth state estimation result, in Example 1, the output of the growth model is used as it is as the harvest result H to be achieved, and the deviation of the growth simulation is reduced. Thus, the control content is determined. On the other hand, in Example 4, indices other than the output of a growth model can be used.

  Examples of the harvest result H to be achieved include, for example, indicators related to plants such as harvest amount, quality, and harvest date that are not included in the growth model, as well as indicators related to operations such as facility power consumption, labor costs, and profits. There is. At this time, the harvest result H is preferably represented by a function such as G, which is the output of the growth model, stress information, environmental information, and control item recording, as described below.

これらの説明変数の全てを用いても良いが、一部のみを用いてＨを表しても良く、また、これ以外の項目を説明変数として用いても良い。

Although all of these explanatory variables may be used, H may be expressed using only a part of them, and other items may be used as explanatory variables.

  According to the form of Example 4, since it is possible to use not only the output of the growth model but also other indicators for the determination of the cultivation control content, the cultivation control closer to the intended yield and quality is performed with higher accuracy. Can do.

  Moreover, in another example (Example 5) of this embodiment, the cultivation control at the time of using water stress as stress information is demonstrated. Note that in the fifth embodiment, description of portions common to the first embodiment may be omitted.

  In Example 5, the control content of cultivation can be determined using water stress as stress information. Water stress is also referred to as water stress or drought stress, and is stress caused by lack of water in plants. There are various factors that cause water stress on plants. For example, the amount of water provided is small, the electrical conductivity of water is high (also called salt stress), and the amount of water absorbed from the root is reduced because the root zone is limited.

In this example, when acquiring stress information and environmental information in step S1201 of FIG. 4, for example, the following apparatus can be used as a stress information acquisition sensor.
・ Devices that measure water stress as biological information of plants ・ Devices that measure water stress of plants indirectly by measuring soil and atmospheric conditions, composition of water provided, etc. ・ Image processing using a camera, etc. A device for indirectly measuring water stress in plants (see Fig. 1)
Water stress is stress information that can be easily obtained compared to other stress information by using these devices. Therefore, as in Example 5, by using water stress as the stress information, it is possible to easily perform cultivation control that provides a harvest result close to the intended yield and quality. Furthermore, since water is the most important element for plant cultivation, efficient cultivation control is possible by using water stress as stress information.

  Moreover, the cultivation assistance method of this invention can be performed using an example (plant cultivation assistance apparatus 100) of the above-mentioned cultivation assistance apparatus (FIGS. 1-4).

  Specifically, as a cultivation support method for supporting cultivation of plants, a step of measuring stress information of a cultivation target (step S1201 in FIG. 4), a step of measuring environmental information of a cultivation target (step S1202 in FIG. 4), A step of estimating the internal state of the cultivation target from the stress information and the environment information (step S1203 in FIG. 4), and a step of estimating the growth state of the cultivation target based on the estimation result of the internal state (step S1204 in FIG. 4). Based on the estimation result of the growth situation, the step of controlling the cultivation contents of the cultivation target (steps S1205 to S1207 in FIG. 4) is executed (see Example 1).

  Thereby, from the information on the stress of the plant to be cultivated and the surrounding environment information, the growth situation considering the stress is estimated, and the cultivation control content is determined based on the estimated growth situation, thereby appropriately controlling the cultivation control. It can be carried out. Therefore, it is possible to stably and easily produce intended high yield and high quality crops. Therefore, by using this cultivation support method, cultivation control for obtaining a harvest result close to the intended yield and quality becomes possible.

  Moreover, in the step of estimating the internal state of the cultivation target, the sensitivity of the cultivation target with respect to the environmental information can be estimated as the internal state of the cultivation target (see Example 2). Thereby, since the control content of cultivation is determined using the sensitivity with respect to the environment of a plant, it becomes possible to give an appropriate stress with respect to a plant and can obtain the harvest result closer to the intended yield and quality. Moreover, since the internal state of the cultivation object can be estimated by preliminary experiments and simple data analysis, the cultivation object can be easily controlled.

  Moreover, in the step of estimating the internal state of the cultivation target, the internal state can be estimated using a model having a hidden state (see the modified example of Example 2 and the modified example 2 of Example 3). Thereby, since the past stress information and environmental information can be used as described above, it is not necessary to prepare in advance information necessary for estimation of the internal state by a preliminary experiment or the like. In addition, in the case of estimation using a model having a hidden state, the internal state can be estimated even if there are missing values in the objective variable and explanatory variable, so the intended yield and quality can be achieved even if the acquired information is missing. Closer cultivation control is possible.

  Furthermore, in the step of estimating the internal state of the cultivation target, the internal state of the control target is represented by a state variable or a parameter (see Modification 1 of Example 3). By estimating the state variables and parameters that are affected by the stress as the internal state of the plant, the growth status of plants such as tomatoes is estimated in consideration of the stress using the existing growth model. can do. That is, even when an existing growth model is used, the cultivation control considering the stress can be performed, so that a harvest result close to the intended harvest result can be obtained.

  In addition, said cultivation assistance method is an example of the cultivation assistance method of this invention. In this example, an example of the above-described cultivation support apparatus is used to execute the cultivation support method. However, the cultivation support method of the present invention is not limited to the case where this cultivation support apparatus is used, and is executed in various forms. be able to. Moreover, an example of the cultivation assistance method of this invention can be performed by making a computer (computer 110 etc. of the plant cultivation assistance apparatus 100) read a program.

  Such a program can be appropriately provided by any program providing means (recording medium, transmission means, etc.). This program is an example of the cultivation support program of the present invention.

  The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific embodiments, and various modifications and changes can be made within the scope of the invention described in the claims. Is possible.

DESCRIPTION OF SYMBOLS 100 Plant cultivation support apparatus 110 Computer 116 Internal state estimation part 117 Growth simulation part 118 Cultivation control part 120 Sensor part 121 Stress information acquisition sensor 122 Environmental information acquisition sensor PL Plant

Japanese Patent Laid-Open No. 2007-306846 JP 2012-55207 A

Claims (10)

A cultivation support device for supporting cultivation of plants,
A stress information measuring unit for measuring stress information of a cultivation target;
An environmental information measuring unit for measuring environmental information of the cultivation target;
From the stress information and the environmental information, a state estimation unit that estimates an internal state of the cultivation target,
Based on the estimation result of the internal state, a growth state estimation unit that estimates the growth state of the cultivation target,
A cultivation support device, comprising: a control unit that controls cultivation contents of the cultivation target based on the estimation result of the growth state.   The cultivation support apparatus according to claim 1, wherein the internal state of the cultivation target is sensitivity of the cultivation target with respect to the environmental information.   The cultivation support device according to claim 1 or 2, wherein the state estimation unit estimates the internal state using a model having a hidden state.   The cultivation assistance apparatus of any one of Claim 1 thru | or 3 by which the internal state of the said cultivation object is represented by a state variable or a parameter.   The cultivation support device according to any one of claims 1 to 4, wherein the stress information is water stress information. A cultivation support method for supporting plant cultivation,
Measuring stress information of the cultivation target;
Measuring environmental information of the cultivation target;
Estimating the internal state of the cultivation target from the stress information and the environmental information;
Based on the estimation result of the internal state, estimating the growth status of the cultivation target,
And a step of controlling the cultivation contents of the cultivation object based on the estimation result of the growth state.   The cultivation support method according to claim 6, wherein in the step of estimating the internal state of the cultivation target, the sensitivity of the cultivation target with respect to the environmental information is estimated as the internal state.   The cultivation support method according to claim 6 or 7, wherein in the step of estimating the internal state of the cultivation target, the internal state is estimated using a model having a hidden state.   The cultivation support method according to any one of claims 6 to 8, wherein in the step of estimating the internal state of the cultivation target, the internal state is represented by a state variable or a parameter.   The cultivation assistance program for making a computer perform the cultivation assistance method of any one of Claims 6 thru | or 9.

Images