Control variable optimization method, bioresource production method and bioresource production system

Abstract

To provide a control variable optimization method capable of determining more optimal culture conditions by a predictive model using machine learning, and a bioresource production method and a bioresource production system using the control variable optimization method.SOLUTION: A bioresource production system S according to another aspect of the present invention comprises a culture system B that produces bioresources, and a control variable optimization system A that optimizes control variables acquired from the culture system. The control variable optimization system separates the control variables into initial variables and variable variables, creates a predictive model that is adapted to each of the initial variables and the variable variables, and optimizes the control variables by combining the two.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a control variable optimization method, a bioresource production method, and a bioresource production system. [Background technology]

[0002] Bio-production systems play an important role in fermentation technology, which uses microorganisms to produce useful proteins and antibiotics, cell culture technology, which cultivates animal and plant cells to produce pharmaceuticals, and algae culture technology, which cultivates microalgae and produces useful organic matter through photosynthesis from atmospheric carbon dioxide.

[0003] Known techniques related to biological production systems include, for example, mathematical models (Non-Patent Document 1 below), experimental design (Non-Patent Document 2 below), and dimensionality reduction (Non-Patent Document 3 below).

[0004] In the mathematical model shown in Non-Patent Document 1 below, a dynamic mathematical model is constructed consisting of antibody production amount, viable cell number, apoptotic cell number, dead cell number, culture volume, and pH. The parameters of this mathematical model are estimated based on the results of culture tests with different pH values, and the optimal control value of pH is finally predicted using the mathematical model with the estimated parameters.

[0005] In addition, the experimental design method shown in Non-Patent Document 2 below discloses a technique in which several medium components and fermentation parameters are optimized using a greedy method or the like, important factors are screened, and then the optimization is performed using an experimental design method or the like.

[0006] In addition, in the dimension reduction described in Non-Patent Document 3 below, a dimension reduction method that combines Boruta, a dimension reduction method, and variable importance VIP in PLS is applied to near-infrared spectral data, and lactate and glucose concentrations are predicted from the dimension-reduced near-infrared spectral data using a machine learning model. [Prior art documents] [Non-patent literature]

[0007] [Non-Patent Document 1] Optimization of a pH-shift control strategy for producing monoclonal antibodies in Chinese hamster ovary cell cultures using a pH-dependent dynamic model, T. Hogiri et al., J. Biosci.Bioeng., 2018, 125(2) [Non-Patent Document 2] Optimization of heterologous production of Bacillus ligniniphilus L1 laccase in Escherichia coli through statistical design of experiments, I.Mutanda et al., Microbiol.Res. 2023, 274 [Non-Patent Document 3] Transfer learning and wavelength selection method in NIR spectroscopy to predict glucose and lactate concentrations in culture media using VIP-Boruta, H. Kaneko et al., Anal. Sci. Adv., 2021, 2, 470-479. Summary of the Invention [Problem to be solved by the invention]

[0008] However, there are issues such as the difficulty of accurately predicting a culture system using a mathematical model because the mathematical model cannot fully represent the complexities of a real culture system, and the difficulty of optimizing control variables with different qualities when using only one machine learning model.

[0009] In addition, while greedy optimization methods are effective when each control variable is independent and does not interact with each other, there is a problem in that the control variables in cultivation often interact with each other, and therefore greedy optimization methods are likely to only obtain local solutions.

[0010] In addition, experimental design can optimize even when there is interaction, but when there are many types of control variables, or when trying to consider changes to each control variable from moment to moment, the number of combinations becomes enormous, and even experimental design cannot fully optimize (optimization requires a huge number of experiments, but a characteristic of bioproduction is that there is a limit to the number of cultures that can be performed at one time, and each culture can be expensive and take a long time). In addition, when there are many control variables, humans cannot fully understand and interpret the experimental results.

[0011] For the reasons mentioned above, experimental design methods are not able to deal with optimization problems with many control variables, and currently the number of control variables is often limited or constant control is used.

[0012] In view of the above problems, the present invention aims to provide a control variable optimization method that can determine more optimal culture conditions by using a predictive model based on machine learning, and a bioresource production method and a bioresource production system using the same. [Means for solving the problem]

[0013] A control variable optimization method according to one aspect of the present invention that solves the above-mentioned problems is characterized in that it separates the control variables into initial variables and variable variables, creates a prediction model that is adaptive to each of the initial variables and the variable variables, and optimizes the control variables by combining the two.

[0014] In addition, in this respect, although not limited thereto, it is preferable to calculate the priority of the initial variables and the variable variables from the gradient of the control variable with respect to the objective variable, and to determine the order of optimization of the initial variables and the variable variables.

[0015] In addition, in this respect, although not limited thereto, it is preferable to optimize the control variables by repeatedly using two models adapted to the initial variables and the variable variables, respectively, and recursively performing inverse analysis until the optimal control values ​​of the initial variables and the variable variables converge.

[0016] In addition, in this respect, although not limited thereto, it is preferable that the control variables are optimized by assigning medium temperature and pH to variable variables and medium components and inoculation amount to initial variables, and combining two models that are adapted to the initial variables and variable variables, respectively.

[0017] In addition, in this respect, although not limited thereto, it is preferable to perform dimensionality reduction by extracting coefficients that minimize the approximation error by projecting the enormous number of explanatory variables used in each model onto a Gompertz curve.

[0018] In addition, in this respect, although not limited thereto, it is preferable that, when the optimal control values ​​of the initial variables and the variable variables are calculated, the distribution of the predicted values ​​of the control variables and the relative position of the optimal control values ​​with respect to the learning data are presented.

[0019] In addition, in this respect, although not limited thereto, it is preferable to have the feature of separating the variable variables into time-dependent variables so that all the variable variables can be treated equally to the initial variables.

[0020] In addition, a bioresource production method according to another aspect of the present invention involves separating control variables into initial variables and variable variables, creating predictive models that are adapted to each of the initial variables and the variable variables, and combining these two to optimize the control variables, thereby carrying out cultivation using the control variables optimized by a control variable optimization method.

[0021] In addition, a bioresource production system according to another aspect of the present invention includes a culture system for bioresource production, and a control variable optimization system for optimizing control variables obtained from the culture system. The control variable optimization system separates the control variables into initial variables and variable variables, creates predictive models adapted to each of the initial variables and the variable variables, and optimizes the control variables by combining these two. Effect of the Invention

[0022] As described above, the present invention provides a control variable optimization method that can determine more optimal culture conditions by a predictive model using machine learning, as well as a bioresource production method and a bioresource production system using the same. [Brief description of the drawings]

[0023] [Figure 1] FIG. 1 is a functional block diagram of the entire bioresource production system according to an embodiment. [Diagram 2] FIG. 2 is a diagram showing the flow of data processing in the bioresource production method according to the embodiment. [Diagram 3] FIG. 1 is a diagram showing an image regarding determination of the priority of control variables in a bioresource production method according to an embodiment. [Figure 4] FIG. 1 is a diagram showing an image of a case where dimensional reduction is performed in a bioresource production method according to an embodiment. [Diagram 5] FIG. 1 is a diagram showing an image of determining the validity of an optimal value in a bioresource production method according to an embodiment. [Figure 6] FIG. 1 is a diagram showing an outline of the results of actual culture in a bioresource production method according to an embodiment. [Figure 7] FIG. 1 is a diagram showing an overview of the application of a Gompertz curve when actually culturing is performed in a bioresource production method according to an embodiment. [Figure 8] FIG. 2 is a flowchart showing calculation of optimal control values ​​for the bioresource production method according to the embodiment. [Figure 9]FIG. 2 is a diagram showing functional blocks during real-time control of the entire bioresource production system according to the embodiment. [Figure 10] FIG. 1 is a flowchart showing calculation of optimal control values ​​during real-time control of a bioresource production system according to an embodiment. BEST MODE FOR CARRYING OUT THEINVENTION

[0024] Hereinafter, the embodiments of the present invention will be described in detail with reference to the drawings. However, the present invention can be embodied in many different forms, and is not limited to the specific descriptions given in the following embodiments and examples.

[0025] First, FIG. 1 shows the overall functional blocks of a biological resource production system S according to this embodiment (hereinafter referred to as "this system").

[0026] As shown in the figure, the present system S comprises a culture system B and a control variable optimization system A that adjusts the control variables for the culture system B. As shown in the figure, the present system S cultivates, i.e. produces, a bioresource using the culture system B, while the control variables used in the culture system B are adjusted by the control variable optimization system A.

[0027] As described above, the culture system B in the present system S is a system capable of culturing a bioresource. Specifically, the culture system B in the present system S includes at least an incubator B1, a culture control device B2 that controls the incubator B1, various sensors B3, and a sensor control device B4 that controls the sensors.

[0028] The incubator B1 in the present system S is a device including a culture vessel for culturing a bioresource, and by introducing a specific target bioresource into the incubator B1, the bioresource can be cultured and grown, i.e., produced. The specific structure of the incubator B1 is not limited as long as it has the above functions, but includes, for example, a culture vessel such as a jar fermenter, a temperature control device for controlling the temperature of the jar fermenter, a rotation device for rotating the culture vessel, an acid or base supply device for adjusting pH, an oxygen supply device for supplying oxygen, carbon dioxide, etc., a carbon dioxide supply device, and in some cases, an observation device for observing the inside of the culture vessel, but is not limited thereto. In addition to introducing the bioresource itself to be grown into the culture vessel of the incubator B1 of the present system S as described above, culture liquid, medium, etc. are also added as necessary.

[0029] In addition, in this system S, "bioresources" refers to resources produced from living organisms, and more specifically refers to plants, microorganisms, cells, algae, bacteria, viruses, etc. themselves, or useful products produced by their activities, but is not limited to these.

[0030] Specifically, this system S can be applied to microbial manufacturing, pharmaceutical production using cells, algae manufacturing, and agriculture. For example, in the process of producing substances using microorganisms in "microbial manufacturing," there are initial variables such as medium components, inoculation amount, metabolic switching time, feed start time, culture solution volume at the start, autoclave time and temperature, feed solution, feed solution concentration, and pH adjustment solution, as well as variable variables such as Temp, pH, Agit, DO, aeration volume, feed rate, temperature dead zone, pressure, and oxygen concentration. Therefore, the microbial manufacturing process can be controlled using the initial variables and variable variables.

[0031] In "cell-based pharmaceutical production", for example, in antibody (protein) production using CHO cells (animal cells), there are initial variables such as medium components, seeding amount, metabolic switching time, feed start time, culture medium volume at start, autoclave time and temperature, feed solution, feed solution concentration, pH adjustment solution, and variable variables such as Temp, pH, Agit, DO, aeration volume, feed rate, temperature dead zone, pressure, oxygen concentration, etc. Therefore, the cell-based pharmaceutical production process can be controlled using the initial variables and variable variables.

[0032] Furthermore, in the process of producing substances using algae in "algae manufacturing," there are initial variables such as medium components, inoculation amount, metabolic switching time, feed start time, culture solution volume at the start, autoclave time and temperature, feed solution, feed solution concentration, and pH adjustment solution, as well as variable variables such as temp, pH, agit, DO, aeration volume, feed rate, temperature dead zone, pressure, oxygen concentration, CO2 supply, and light intensity. Therefore, the process of algae manufacturing can be controlled using the initial variables and variable variables.

[0033] In "agriculture", for example, in organic and natural farming, there are initial variables such as chemical items (pH, EC, nitrogen ion concentration, phosphate ion concentration, potassium ion concentration), physical items (water content, maximum water holding capacity), and biological items (total bacterial count, total carbon amount, total nitrogen amount, total phosphorus amount, total potassium amount, C / N ratio, C / P ratio), and variable variables such as the amount of carbon input and nitrogen input from organic materials each year. Therefore, in organic and natural farming, control is possible using the initial variables and variable variables.

[0034] In addition, in this system S, various sensors B3 are connected to the culture vessel. Here, the "sensor" is a device used to grasp the situation in the culture vessel and its vicinity, and examples thereof include a pH sensor for detecting pH, a temperature sensor for detecting temperature, a dissolved oxygen amount sensor for detecting dissolved oxygen concentration (DO), a culture condition detection sensor for detecting the culture condition such as an oxygen sensor for detecting oxygen concentration, and a culture status detection sensor for detecting the culture status such as a carbon dioxide sensor for detecting carbon dioxide concentration, an infrared sensor for detecting infrared rays, an RGB sensor for detecting color, and an electric potential sensor for detecting electric potential, but are not limited thereto. By using these various sensors, it is possible to detect the culture environment, obtain the production status of bioresources, etc. In addition, by controlling the culture control device B2, etc. after detecting these, the culture conditions in the bioresource production can be adjusted.

[0035] In addition, at least some of the various sensors B3 of the present system S are connected to the culture control device B2. Specifically, the culture control device B2 is connected to culture condition detection sensors such as a pH sensor, a temperature sensor for detecting temperature, a dissolved oxygen amount sensor for detecting dissolved oxygen concentration (DO), and an oxygen sensor for detecting oxygen concentration, and the culture control device B2 can change the culture conditions detected by these sensors, such as pH, temperature, dissolved oxygen concentration, and oxygen concentration. This allows the present system S to realize and maintain optimal culture conditions. As described later, data measured by some of the various sensors B3 is collected in the culture control device B2, and sensor data information (such as Jar Number, version number, and serial number) is added to the data, which is then sent to the control variable optimization system A.

[0036] In addition, at least some of the various sensors B3 of the system S are connected to a sensor control device B4. Specifically, the sensor control device B4 is connected to culture status detection sensors that detect the culture status, such as a carbon dioxide sensor that detects carbon dioxide concentration, an infrared sensor that detects infrared rays, an RGB sensor that detects color, and an electric potential sensor that detects electric potential. By being connected to the above culture status detection sensors, the sensor control device B4 can perform necessary control on these sensors and transmit acquired data to devices outside the culture container. As a result, not only can the culture status be confirmed, but it can also be used as data for machine learning, which will be described later. As will be described later, data measured by some of the various sensors B3 is collected by the sensor control device B4, sensor data information (Jar Number, version number, serial number, etc.) is added, and the data is sent to the control variable optimization system A.

[0037] In addition, the present system S includes a control variable optimization system A connected to the culture system B. The control variable optimization system A can optimize the control variables contained therein based on the outputs from the culture condition detection sensor and the culture status detection sensor. Details of this will be described later.

[0038] In addition, in this system S, the culture system B and the control variable optimization system A are connected so as to enable data transmission and reception, but they may be directly connected by wiring such as a cable, or may be connected via a LAN (Local Area Network) or a WAN (Wide Area Network). In the example shown in the figure, the culture control device B2 and the sensor control device B4 are connected to the LAN via Wifi or the like, and are further connected to the control variable optimization system A via a WAN. A specific example of the connection between the LAN, WAN, etc. and various devices is a communication device such as a modem.

[0039] Furthermore, the present system S may further include a control computer C. By providing the control computer C and connecting it to the above-mentioned culture system B, control variable optimization system A, etc., it becomes possible to control these as a whole.

[0040] In this system S, the control computer C can be literally realized, and the control variable optimization system A can be realized by using a so-called computer. The computer is a device that can perform predetermined processing based on input data and output data, and is preferably configured to include, for example, a central processing unit (CPU), which is a component of a general information processing device, a non-volatile recording medium such as a hard disk or flash memory, a volatile recording medium such as a memory, input devices such as a keyboard or mouse, a communication device such as a modem, and a bus connecting these. In addition, the computer is also preferably equipped with a display device that displays the results of data processing.

[0041] Furthermore, as described above, the control variable optimization system A of the present system S can optimize the control variables contained therein based on the outputs from the culture condition detection sensor and the culture status detection sensor, which can be realized by executing a so-called computer program. That is, according to the present system S, a control variable optimization method can be realized by having a computer execute a program for optimizing the control variables (a control variable optimization program (hereinafter referred to as "the present program")), and the present system S can realize an optimized bioresource production method by using the optimized control variables.

[0042] This diagram shows functional blocks for explaining the flow (processing flow) of control optimization realized by this program. In other words, the control variable optimization system A in this system S functions as each block by executing this program.

[0043] First, as is clear from the above description, values ​​acquired by various sensors B3 in culture system B are uploaded and accumulated as various data via a LAN, WAN, etc. in a cloud DB in control variable optimization system A. The cloud DB can be realized by recording data in a predetermined area of ​​a non-volatile recording medium such as a computer hard disk.

[0044] In addition, the data uploaded to the cloud DB that is to be processed is collected (data collection), converted to a standardized data format, and prepared for further processing. This converted data is then written back to the DB.

[0045] In addition, it is preferable to perform pre-processing for efficient specific processing on the data acquired by the various sensors B3 in addition to data conversion for unifying the above-mentioned formats. For example, since the data acquired by the various sensors B3 contains noise, it is preferable to smooth the data by, for example, moving average processing to remove the noise from the viewpoint of further improving the optimization accuracy. It is also preferable to perform processing to convert the measured values ​​of each sensor into different numerical values. For example, in the case of an RGB sensor, conversion to Lab values ​​is performed, and in the case of a near-infrared sensor, absorbance conversion is performed. The data that has undergone this pre-processing is written back to the DB.

[0046] Then, in the control variable optimization system A of the present system S, machine learning is performed using the data that has been preprocessed as described above, a prediction model is constructed, and a process (control variable optimization process) is performed to calculate optimal control values. The specific contents will be clear from the description below, but the data obtained by the various sensors B3 above are part of the control variables, i.e., variable variables, and the control variables including other variables (initial variables) are optimized.

[0047] Furthermore, the data (control variables) calculated by the above-mentioned control variable optimization process is subjected to a process of determining whether or not they are appropriate as optimal values ​​(validity determination process). The validity is determined based on the relative position with respect to the learning data, and whether or not the optimal control value is in a high distribution of the objective variable. In this process, if the data (control variables) are determined to be valid, they are written back to the cloud DB, and if they are determined to be invalid, they are returned to the above-mentioned data collection step or the control variable optimization process step and processed again.

[0048] Then, data (control variables) that are judged to be valid in the above process are sent to the above culture system B, more specifically, the culture control device B2, and the culture conditions of the incubator B1 are adjusted by the culture control device B2 to become the above control variables, thereby optimizing the production of bioresources.

[0049] Here, we will explain the details of the control variable optimization method (hereinafter referred to as "this method") specifically performed by the control variable optimization system A. This method is characterized by separating the control variables into initial variables and variable variables, creating a prediction model adapted to each of the initial variables and the variable variables, and optimizing the control variables by combining these two. Note that Figure 2 shows the data processing flow of this method.

[0050] As shown in the figure, it is preferable to use data obtained by experimental design in this method. Specifically, the control variables that are variously changed by the above-mentioned system S, and the data obtained as a result of changing the control variables are used.

[0051] As described above, in this method, the control variables are separated into initial variables and variable variables. Here, the term "control variables" refers to conditions (variables) necessary for optimizing the culture, the term "variable variables" refers to variables that are observable, change from moment to moment, and can be controlled, and the term "initial variables" refers to variables that are observable and do not change after being set only once as an initial value for the culture. In this method, the optimization process is performed including the initial variables and variable variables.

[0052] Specific examples of variable variables include, but are not limited to, medium temperature, pH, Agit, dissolved oxygen concentration (DO), flow rate, feed rate, temperature dead zone, pressure, oxygen concentration, etc. Examples of initial variables include, but are not limited to, medium components, inoculation amount, metabolic switching time, feed start time, culture solution volume at start, autoclave time and temperature, feed solution, feed solution concentration, pH adjustment solution, etc.

[0053] In addition, the control variables are separated into the variable variables and the initial variables as described above, but the explanatory variables of the prediction model include state variables in addition to the control variables, and the explanatory variables of the prediction model are separated into "variable variables", "initial variables", and "state variables". However, "state variables" are observable but cannot be controlled when culturing, and are not subject to optimization in the prediction model. Therefore, the optimization process described below is performed only on the variable variables and initial variables among the explanatory variables.

[0054] Specific examples of state variables include, but are not limited to, exhaust CO2, exhaust O2, RGB, electric potential, visible light, near infrared, fluorescence, UV, viscosity, vibration, odor, respiratory quotient, ventilation volume, operating time of the heater / cooler pump, operating time of the pH adjustment solution pump, feed weight, kLA, glucose, enzyme sensor, etc.

[0055] In the above, after extracting the variable variables and the initial variables from the control variables, the method determines which variable variables have higher priority. Here, an image of the determination of the priority of the variable variables and the initial variables is shown in FIG. 3. Specifically, this figure shows a method of calculating the influence of the control variables on the objective variable. Here, the "objective variable" is a variable that serves as an index for finding an optimal value, and if the fluctuation of the predicted value of this objective variable is large, the priority is determined to be high, and if it is small, the priority is determined to be low. As shown in this figure, within the set range of the control variables, values ​​that the control variables can take are generated, and the data is input into the machine learning model to calculate how the objective variable will fluctuate according to the value of the control variables. By calculating the value obtained by subtracting the minimum value from the maximum value of the calculated fluctuation (= the influence on the objective variable (gradient of the control variable with respect to the objective variable)), it is possible to estimate which variable has the greater influence in control. When performing sequential optimization, it is better to optimize variables with greater influence first, so in the present invention, it is determined that the variable with the greater influence calculated by the above method has the higher priority, and that variable is optimized first.

[0056] After the priority order of the control variables is determined by the above process, a predetermined process is performed on the variable variables or initial variables that are determined to have a high priority order. However, the same process is performed on the variable variables and initial variables.

[0057] First, if it is determined that the priority of the initial variable is high, machine learning is performed based on the prediction model of the initial variable to calculate the optimal value of the initial variable. Then, machine learning is performed based on the prediction model of the variable variable to calculate the optimal value of the variable variable. On the other hand, if it is determined that the priority of the variable variable is high, machine learning is performed based on the prediction model of the variable variable to calculate the optimal value of the variable variable. Then, machine learning is performed based on the prediction model of the initial variable to calculate the optimal value of the initial variable. That is, the point of this method is that prediction models are created in a form corresponding to each of the initial variable and the variable variable, and the control variable can be optimized by combining them.

[0058] In machine learning prediction models based on a prediction model of variable variables, a data format consisting of variable variables, initial variables, state variables, and objective variables is used as training data, and the values ​​that change the variable variables, the values ​​that fix the initial variables and state variables, and the objective variable is the resultant value. Machine learning is performed using the training data collected by this prediction model. After the machine learning is completed, the control value of the variable variables that maximizes the objective variable is obtained by reverse analysis of this prediction model.

[0059] The prediction model in machine learning based on the prediction model of the initial variables is similar to the above. Specifically, a data format consisting of variable variables, initial variables, state variables, and objective variables is used as the learning data, and the value that changes the initial variables, the value that fixes the variable variables and state variables, and the objective variable are the resulting values. Machine learning is performed using the learning data collected by this prediction model. After the machine learning is completed, the control value of the initial variables that maximizes the objective variable is obtained by reverse analysis of this prediction model.

[0060] Incidentally, the learning data for the prediction model for the initial variables in this method has state variables with variable variables equal to (variable variables + number of sensors) x incubation time (for example, if the incubation time is 1000 min, each state variable will have 1000 variables). Therefore, if a prediction model is created without dimensionality reduction, it is expected that the prediction accuracy of the prediction model will deteriorate due to problems of multicollinearity and the curse of dimensionality. Therefore, in this method, it is also useful to reduce the dimension by approximating the time series data of the sensor to a Gompertz curve y = a x exp (-b x exp (-c x x)) + d. In the above example, by approximating a Golpentz curve for a certain state variable X, it is possible to reduce the 1000 variables to four variables, the coefficients a, b, c, and d that determine the Golpentz curve. It is preferable to use logarithmic functions, exponential functions, Golpentz curves, etc., in accordance with bio-knowledge as the approximation function. An image of this case is shown in Figure 4. In other words, this process reduces the dimensions of the huge number of explanatory variables used in each model by projecting them onto the Gorpenz curve and extracting coefficients that minimize the approximation error.

[0061] In the above explanation of this method, the variable variables and the initial variables are treated as different variables, but in some cases, it is possible to make the variable variables the initial variables. Specifically, the variable variables can be divided into variables for the culture time, such as the variable variable X0 at 0 min, the variable variable X10 at 10 min, the variable variable X20 at 20 min, and the variable variable Xn at n min, so that they can be treated as initial variables. In this case, it is possible to omit the determination of the priority of the control variables and the machine learning of the variable variables. In other words, it is also preferable to separate the variable variables into variables for each time so that all the variable variables can be treated equally to the initial variables.

[0062] When the machine learning is completed, a determination is made as to whether or not the optimal value obtained as a result of this prediction model has converged. If it has converged, the process proceeds to the next step, but if it has not converged, the process returns to the above priority determination step, and machine learning is performed again to calculate the optimal value. Regarding "whether or not it has converged," various determination methods can be adopted, but it is preferable to determine whether or not it has converged by, for example, whether or not the difference between the solutions obtained by successive calculations gradually becomes smaller and is smaller than a predetermined amount. In other words, by repeatedly using two models that are adapted to the initial variables and the variable variables, respectively, and recursively performing inverse analysis until the optimal control values ​​of the initial variables and the variable variables converge, it is possible to obtain more accurate results by optimizing the control variables.

[0063] When the optimal value has converged, it is then judged whether this optimal value is appropriate as the final control value, and if it is not appropriate, the process returns to the experimental design method and increases the amount of collected data, continuing until the final control value is extracted in a manner that prevents the optimal value from becoming a particular solution due to the amount and bias of the learning data.On the other hand, if it is judged that the optimal value is appropriate as the optimal control value, the obtained result is confirmed as the optimal control value and proposed, and the process ends.

[0064] In this method, the variable variables are controlled over time, and the initial variables are optimized for multiple variables, resulting in complex, multidimensional optimal conditions. For this reason, it is difficult for humans to judge whether the presented control values ​​are likely to actually lead to good cultivation results. Therefore, by reducing the dimensions of the multidimensional control variables of the learning data and the optimal control values ​​and projecting the learning data and the optimal control values ​​into the reduced-dimensional space, it becomes possible to confirm the relative position of the optimal control values ​​and quantitatively visualize the effect of the control variables on the objective variables. This is shown in Figure 5.

[0065] In this case, a prediction model is created using the learning data in the projected space as input, the contour lines of the predicted values ​​are calculated, and the contour lines are added to the above diagram, making it possible to visualize the distribution of the predicted values, and to confirm whether the optimal control value is in a high distribution of the objective variable.Then, by confirming the relative position of this optimal control value and whether the optimal control value is in a high distribution of the objective variable, the validity of the obtained optimal control value can be judged.In other words, when the optimal control values ​​of the initial variables and the variable variables are calculated, it is preferable to present the distribution of the predicted values ​​of the control variables and the relative position of the optimal control value with respect to the learning data.

[0066] As described above, the present embodiment can provide a control variable optimization method capable of determining more optimal culture conditions by a prediction model using machine learning, as well as a bioresource production method and a bioresource production system using the same.

[0067] To explain the effects of the present invention more specifically, in this embodiment, optimal culture conditions are proposed by a predictive model using machine learning, making it possible to improve the productivity of microorganisms by nearly 10% compared to culture conditions using conventional optimization methods.

[0068] The results of applying this technology to the actual cultivation of a certain microorganism are shown in Figure 6. This result was obtained by creating a predictive model for the cultivation of a certain microorganism, calculating the optimal conditions for a certain objective variable, and culturing the microorganism according to the model.

[0069] In addition, there are a total of seven control variables for this cultivation. Conventionally, the aim is to maximize the objective variable by selecting the value of each variable, whereas in this method, of the seven control variables, four are assigned as initial variables and three as variable variables.

[0070] The differences between the conventional and proposed methods in terms of variable and initial variables are shown in (a) and (b), respectively. As can be seen from these figures, the variable variables in the proposed method change during the culture time, whereas in the conventional method they must be kept constant. Also, in the proposed method, it is possible to extract optimal values ​​for the initial variables that take into account the changes in the variable variables, and it can be seen that values ​​completely different from those in the conventional method are selected.

[0071] Finally, the difference in productivity is shown in Figure 6(c). Compared to the conventional method (left bar graph), the objective variable value is higher when using this method (right bar graph), and we can see that productivity is nearly 10% higher.

[0072] In addition, in actual cultivation, the cultivation conditions are constantly changing depending on the weather, the inoculation, and the state of the medium. However, in this method, the variable variables are controlled in response to these changes, making it possible to maintain improved high productivity in cultivation.

[0073] In addition, because culture conditions have traditionally been examined based on the experience of experienced engineers, it has been difficult to visualize the effect of control variables on the objective variable.In contrast, as shown in Figure 5 above, this method reduces dimensions, making it possible to plot a graph with the predicted value of the calculated objective variable on the vertical axis and the control variable on the horizontal axis, visualizing the effect of the control variable on the objective variable, which has the advantage of being useful as information when constructing culture equipment.

[0074] In addition, as shown in Figure 4 above, projection onto the Gompertz curve enables dimensionality reduction and improves prediction accuracy by about 35% compared to when the Gompertz curve is not used. Figure 7 shows the results of applying this method to a certain microbial culture.

[0075] For a certain objective variable of a certain microbial culture, we created a prediction model in which the explanatory variables were not reduced in dimension (3 culture condition variables + one wavelength of the visible light sensor * 852 points) and in which the dimensions were reduced by Gompertz curve approximation (a) (3 culture condition variables + 4 Gompertz curve variables), and compared the prediction accuracy of each (RMSE standardized by the objective variable) and performed a corresponding t-test to test whether there was a significant difference. The results are shown in Figure (b) (violin plot). In this figure, the width of the light colors (gray) represents the data density, the dashed line represents the average RMSE with dimensionality reduction, and the solid line represents the average RMSE with dimensionality reduction. In addition, "****" in the figure indicates that the corresponding t-test was p<0.0001.

[0076] As shown in this figure, the prediction accuracy (RMSE) was significantly higher when dimension reduction was performed, with the average result being about 35% ((= RMSE without dimension reduction - RMSE with dimension reduction) / RMSE with dimension reduction).

[0077] Incidentally, an example of a flowchart for calculating optimal control values ​​is shown in Fig. 8. In this calculation, learning data for variable variables and learning data for initial variables are created based on the converted data. In the learning data for variable variables, the variable variables are treated as control variables, and the initial variables and sensor data are treated as state variables. On the other hand, in the learning data for initial variables, the initial variables are treated as control variables, and the variable variables and sensor data are treated as state variables.

[0078] Here, the training dataset is converted into a dimensionally reduced dataset for each dataset by dimensionality reduction using PCA or approximation to a Gompertz curve, and feature selection using correlation coefficients, clustering, regularization, etc.

[0079] Next, the dimension-reduced dataset is used as training data to create a predictive model using machine learning such as neural networks and random forests. Next, the controlled variables that have a large effect on the culture are identified by calculating the gradient of the objective variable against the controlled variables using domain knowledge of the culture or a model, and the priority order for optimization of the initial variables and variable variables is determined. Below, an example is explained using a case where the variable variable has a large influence.

[0080] Optimization begins with calculating the optimal control values ​​for the variable variables with high priority using the prediction model created first (=back analysis). The data used for back analysis is one of the acquired data that was not used for learning and is used as batch data for simulation.

[0081] Next, the optimal control value of the variable variable calculated as the state variable of the prediction model with the initial variable as the control variable is reflected, and the optimal control value of the initial variable is calculated. Furthermore, the optimal control value of the variable variable calculated is reflected in the prediction model with the variable variable as the control variable, and the optimal control value of the variable variable is calculated again. Thereafter, optimization of the variable variable and the initial variable is performed recursively until the optimal control value converges, and the finally converged optimal control values ​​of the variable variable and the initial variable are obtained.

[0082] The multidimensional control variables of the training data and the optimal control value are reduced in dimension using PCA or other methods, and the training data and the optimal control value are projected onto the reduced-dimensional space to confirm the relative position of the optimal control value. A prediction model is then created using the training data in the projected space as input, and the contour lines of the predicted values ​​are calculated. The distribution of the predicted values ​​is calculated by adding the contour lines to the above diagram, and the contour lines are added to the above diagram to obtain a diagram depicting the relative positions of the optimal control values ​​and the distribution of the predicted values, and it is then possible to determine whether the optimal control value is appropriate.

[0083] Then, by having a human input the optimal control value that is judged to be appropriate into the culture system, specifically the culture device, before starting culture, culture can be achieved according to the optimal control value predicted by the AI.

[0084] By the way, this method can also be considered to calculate the optimal control value while controlling the culture in real time while culturing. Figure 9 shows a functional block diagram for calculating the optimal control value during real-time control. In this block diagram, only the control variable optimization system A is different, and the rest is the same as in Figure 1 above.

[0085] In this example, the real-time data converted data is input to the constructed prediction model, and the optimal control value of the variable variable is calculated by performing reverse analysis. The calculated optimal control value of the variable variable is fed back to the culture control device B2 in real time, thereby realizing real-time culture control by AI.

[0086] FIG. 10 shows a flow chart of the calculation of the optimal control value in this case, that is, in the case of real-time control. In this case, previously acquired data is used. That is, before the start of the culture, one of the acquired data that was not used for learning is used as data for reverse analysis to calculate the optimal control value of the initial variable. After this calculation, the optimal control value of the initial variable is calculated and input to the culture control device B2.

[0087] In addition, the calculated optimal control values ​​of the initial variables are reflected as state variables in the real-time converted data.

[0088] After the start of cultivation, the real-time converted data is used as input to calculate optimal control values ​​for the variable variables in real time.

[0089] The calculated optimal control values ​​of the variable variables are then fed back in real time to the culture control device B2, thereby achieving real-time culture control by AI. [Industrial Applicability]

[0090] INDUSTRIAL APPLICABILITY The present invention has industrial applicability as a control variable optimization method, a bioresource production method, and a bioresource production system. [Explanation of symbols]

[0091] S ···Bioresource production system A···Control variable optimal system B...Culture system B1...Incubator B2...Culture control device B3...Various sensors B4 Sensor control device C···Control computer

Claims

1. A control variable optimization method characterized by separating control variables into initial variables and variable variables, creating a predictive model that is adapted to each of the initial variables and the variable variables, and optimizing the control variables by combining the two.

2. 2. The method for optimizing a controlled variable according to claim 1, further comprising the step of calculating the priority of the initial variables and the variable variables from a gradient of the controlled variable with respect to the objective variable, and determining an order of optimization of the initial variables and the variable variables.

3. 2. The control variable optimization method according to claim 1, characterized in that the control variables are optimized by repeatedly using two models adapted to the initial variables and the variable variables, respectively, and recursively performing inverse analysis until optimal control values ​​of the initial variables and the variable variables converge.

4. 2. The method for optimizing control variables according to claim 1, characterized in that the control variables are optimized by combining two models that are adapted to the initial variables and the variable variables, respectively, by assigning medium temperature and pH to the variable variables and medium components and inoculation amount to the initial variables.

5. 2. The control variable optimization method according to claim 1, wherein the explanatory variables used in each model are projected onto a Gompertz curve to extract coefficients that minimize an approximation error, thereby reducing the dimension of the explanatory variables.

6. 2. The control variable optimization method according to claim 1, further comprising the step of presenting a distribution of predicted values ​​of the control variables and a relative position of the optimal control value with respect to the learning data when the optimal control values ​​of the initial variables and the variable variables are calculated.

7. 2. The method for optimizing control variables according to claim 1, wherein the variable variables are separated into time-dependent variables so that all the variable variables can be treated equally to the initial variables.

8. A bioresource production method for cultivating and cultivating organisms based on the optimized control variables according to claim 1.

9. A culture system for producing bioresources; A control variable optimization system that optimizes a control variable obtained from the culture system; The controlled variable optimization system is a bioresource production system in which the controlled variables are separated into initial variables and variable variables, a predictive model is created that is adapted to both the initial variables and the variable variables, and the controlled variables are optimized by combining these two.

Images