A method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse

Abstract

The present invention relates to the field of process industry production and processing technology, and in particular to a method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse. By establishing a state space model for each fermentation stage in a continuous state variable fermentation process, establishing a comprehensive benefit optimization index for each fermentation stage, and calculating the target feed substrate concentration for each fermentation stage when the comprehensive benefit optimization index is minimized, and in the process of solving the feed substrate concentration for each fermentation stage after the initial fermentation stage, utilizing the similarity of each stage of cell growth in the fermentation tank, and utilizing the target feed substrate concentration of the previous fermentation stage and the parameter matrix of the long-term comprehensive benefit, the comprehensive benefit optimization index for the current fermentation stage is solved. The present invention optimizes the control effect of the continuous fermentation process, improves the control performance of the fermentation process, thereby reducing the fermentation cost, improving the fermentation efficiency and fermentation quality, and achieving maximum economic benefits.

Description

Technical Field

[0001] The present invention relates to the technical field of process industry production and processing, and in particular to a method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse. Background Art

[0002] Fermentation engineering is the process of culturing cells and producing metabolites in large quantities in fermenters under optimal fermentation conditions. It is widely used in fields such as food fermentation and biopharmaceuticals. Fermentation engineering typically identifies general conditions for bacterial cell growth and metabolism during the shake flask stage. Optimal conditions and methods for controlling the fermentation process are then determined through small-scale fermentation tank experiments. Large-scale testing is required before large-scale industrial production. Due to the numerous reactions and activities occurring within the fermenter, the fermentation process is subject to significant uncertainty. Subtle differences such as the water used, the length of time the strain has been stored, and the size of the fermenter can lead to variations in microbial metabolism. The same strain and culture medium can produce varying results in different production batches. Therefore, fermentation process control is a crucial component of the fermentation process. Industry typically controls the fermentation process by maintaining certain parameters, such as pH, dissolved oxygen concentration, temperature, and residual substrate.

[0003] Continuous fermentation is a common fermentation method characterized by relatively stable operating conditions. In continuous fermentation, fresh culture substrate and outflowing culture fluid are added at the same rate, maintaining a constant volume within the fermenter. The stability of these operating conditions helps maintain stable microbial growth and metabolism, thereby achieving stable productivity and product quality. While continuous fermentation initially relied on manual experience, current research is now using feed substrate concentration as a control variable, based on mathematical models of cell culture operating conditions and production objectives. This linear control model is then used to design stable linear controllers, addressing the inherent nonlinear characteristics of fermentation models.

[0004] However, because actual cell growth involves complex chemical and biological reactions and the distribution characteristics of cell populations, its dynamic evolution is complex. The mathematical model for cell culture differs at different stages of cell growth and is difficult to accurately describe. Rebuilding and solving the fermentation model at each stage consumes excessive computing resources, resulting in excessive fermentation costs and reduced fermentation efficiency and quality. Summary of the Invention

[0005] Therefore, the technical problem to be solved by the present invention is to overcome the problem in the prior art that each fermentation process requires the fermentation model to be rebuilt and solved, resulting in reduced fermentation efficiency and fermentation quality.

[0006] To solve the above technical problems, the present invention provides a method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse, comprising:

[0007] Establish a state space model for the initial fermentation stage of the continuous fermentation process and design a state feedback controller. Based on the optimal steady state of the cell state variables in the initial fermentation stage, establish the comprehensive efficiency optimization index of the initial fermentation stage.

[0008] Design a long-term comprehensive evaluation index of the feed substrate concentration in the initial fermentation stage and obtain a parameter matrix of the long-term comprehensive benefits of the initial fermentation stage;

[0009] Solve the feed substrate concentration in the initial fermentation stage when the comprehensive benefit optimization index of the initial fermentation stage is minimized, and use it as the target feed substrate concentration in the initial fermentation stage;

[0010] For each fermentation stage after the initial fermentation stage, a state space model of the current fermentation stage is established and a state feedback controller is designed; based on the optimal steady state of the cell state variables in the current fermentation stage, the comprehensive benefit optimization index of the current fermentation stage is established;

[0011] Based on the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage, the comprehensive benefit optimization index of the current fermentation stage is solved using the target feed substrate concentration and the parameter matrix of the long-term comprehensive benefit of the previous fermentation stage. The parameter matrix of the long-term comprehensive benefit of the current fermentation stage and the target feed substrate concentration when the comprehensive benefit optimization index of the current fermentation stage is minimized are obtained, including:

[0012] Design a long-term comprehensive evaluation index for feed substrate concentration in the current fermentation stage;

[0013] Initialize the parameter matrix of the long-term comprehensive benefits of the current fermentation stage and the feed substrate concentration, and use the target feed substrate concentration of the previous fermentation stage and the parameter matrix of the long-term comprehensive benefits to calculate the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage;

[0014] Based on the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage, the parameter matrix of the long-term comprehensive benefit of the current fermentation stage is estimated;

[0015] Optimizing the feed substrate concentration in the current fermentation stage based on the estimated value of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage;

[0016] Based on the estimated value of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage and the optimized feed substrate concentration of the current fermentation stage, return to recalculate the comprehensive benefit optimization index deviation between the current fermentation stage and the previous fermentation stage, and re-estimate the parameter matrix of the long-term comprehensive benefits of the current fermentation stage, and then optimize the feed substrate concentration of the current fermentation stage again until the estimated value of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage no longer changes, and obtain the parameter matrix of the long-term comprehensive benefits of the current fermentation stage and the target feed substrate concentration.

[0017] Preferably, a state space model of the initial fermentation stage in the continuous fermentation process is established and a state feedback controller is designed; based on the optimal steady state of the cell state variables in the initial fermentation stage, comprehensive efficiency optimization indicators of the initial fermentation stage are established, including:

[0018] The state space model of the initial fermentation stage is:

[0019] x a (k+1)=A a x a (k)+B a u a (k)

[0020] Among them, A a and B a is the state space model parameter of the initial fermentation stage, x a (k) and x a (k+1) represents the cell state variables at the initial fermentation stage at time k and time k+1, respectively, u a (k) represents the feed substrate concentration at the initial fermentation stage at time k;

[0021] Based on the state space model of the initial fermentation stage, the state feedback controller is designed as:

[0022]

[0023] Among them, K a is the control amount in the initial fermentation stage, x 0,a Represents the optimal steady state of cell state variables during the initial fermentation phase.

[0024] Preferably, based on the optimal steady state of the cell state variables in the initial fermentation stage, the comprehensive benefit optimization index of the initial fermentation stage is established as:

[0025]

[0026] Among them, c a (k) represents the comprehensive benefit optimization index of the initial fermentation stage at time k, It represents the difference between the measured value of the cell state in the initial fermentation stage and the optimal steady-state value at time k. E and F are both weight parameter matrices.

[0027] Preferably, a long-term comprehensive evaluation index of the feed substrate concentration in the initial fermentation stage is designed to obtain a parameter matrix of the long-term comprehensive benefits of the initial fermentation stage, including:

[0028] Design a long-term comprehensive evaluation index for the feed substrate concentration in the initial fermentation stage. The formula is:

[0029]

[0030] Among them, J a (k) and J a (k+1) are the long-term comprehensive evaluation indicators of the feed substrate concentration in the initial fermentation stage at time k and time k+1, γ is the attenuation factor, is the parameter matrix of the long-term comprehensive benefits in the initial fermentation stage, and R is the weight parameter matrix.

[0031] Preferably, when the state space model parameters of the initial fermentation stage are unknown, solving the feed substrate concentration in the initial fermentation stage when the comprehensive benefit optimization index of the initial fermentation stage is minimized as the target feed substrate concentration in the initial fermentation stage includes:

[0032] First, the parameter matrix of the long-term comprehensive benefits in the initial fermentation stage is estimated, including:

[0033] The long-term comprehensive evaluation index of the feed substrate concentration in the initial fermentation stage is expressed as:

[0034]

[0035] in, is x a (k),u a (k), x 0,a The relevant parameter vector, V a Yes and W a The parameter matrix corresponding to the sub-matrix parameters;

[0036] Establish V a The relationship with the comprehensive benefit optimization index in the initial fermentation stage is as follows:

[0037] c a (k) = z a (k)V a

[0038] Among them, z a (k) = h a (k)-γh a (k+1);

[0039] The system identification method is used to solve the minimum comprehensive benefit optimization index V in the initial fermentation stage. a The estimated value of W a estimated value of;

[0040] The feed substrate concentration in the initial fermentation stage is optimized based on the parameter matrix of the long-term comprehensive benefits of the initial fermentation stage. The formula is:

[0041]

[0042] in, is the estimated value of the parameter matrix of the long-term comprehensive benefits of the initial fermentation stage at time k, The parameter matrix W of the long-term comprehensive benefits in the initial fermentation stage is a The sub-matrix of They are estimated value of;

[0043] Then the control quantity K of the initial fermentation stage at time k+1 is a (k+1) is

[0044] Based on the optimized feed substrate concentration in the initial fermentation stage, the parameter matrix for re-estimating the long-term comprehensive benefits of the initial fermentation stage is returned, and the feed substrate concentration in the initial fermentation stage is optimized again until the estimated value of the parameter matrix for the long-term comprehensive benefits of the initial fermentation stage no longer changes, thereby obtaining the parameter matrix for the long-term comprehensive benefits of the initial fermentation stage and the target feed substrate concentration.

[0045] Preferably, when the state space model parameters of the initial fermentation stage are known, the parameter matrix of the long-term comprehensive benefits of the initial fermentation stage is directly obtained; and the feed substrate concentration of the initial fermentation stage is solved when the comprehensive benefit optimization index of the initial fermentation stage is minimized, as the target feed substrate concentration of the initial fermentation stage, and the methods include the minimum principle, dynamic programming method, numerical calculation method and gradient method.

[0046] Preferably, a long-term comprehensive evaluation index of the feed substrate concentration in the current fermentation stage is designed, including:

[0047]

[0048] Among them, J b (k) and J b (k+1) are the long-term comprehensive evaluation indicators of the feed substrate concentration in the current fermentation stage at time k and time k+1, respectively, c b (k) is the comprehensive benefit optimization index of the current fermentation stage at time k, γ is the attenuation factor, xb (k) is the cell state variable of the current fermentation stage at time k, u b (k) is the feed substrate concentration of the current fermentation stage at time k, is the optimal steady state of the cell state variables in the current fermentation stage; is the parameter matrix of the long-term comprehensive benefits of the current fermentation stage, A b and B b are the state space model parameters of the current fermentation stage, and E, F, and R are all weight parameter matrices.

[0049] Preferably, the parameter matrix of the long-term comprehensive benefit of the current fermentation stage and the feed substrate concentration are initialized, and the target feed substrate concentration of the previous fermentation stage and the parameter matrix of the long-term comprehensive benefit are used to calculate the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage, including:

[0050] Assuming that the initial time of the current fermentation stage is τ, when k = τ, let u b (k)=u a* , and initialize

[0051] Among them, u b (k) represents the initial feed substrate concentration of the current fermentation stage at time k, u a* represents the target feed substrate concentration of the previous fermentation stage, The estimated value of the parameter matrix representing the long-term comprehensive benefits of the current fermentation stage at time k;

[0052] When k>τ, based on the parameter matrix of the long-term comprehensive benefits of the previous fermentation stage, the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage is calculated, including:

[0053] The comprehensive benefit optimization index of the current fermentation stage at time k is in, is the difference between the measured value of the cell state in the current fermentation stage and the optimal steady-state value at time k, E and F are both weight parameter matrices, u b (k) is the feed substrate concentration of the current fermentation stage at time k;

[0054] The comprehensive benefit optimization index of the previous fermentation stage at time k is Among them, J a' (k) and J a' (k+1) are the long-term comprehensive evaluation indicators of the feed substrate concentration in the previous fermentation stage at time k and time k+1, γ is the attenuation factor, x a'(k) is the cell state variable of the previous fermentation stage at time k, u a' (k) is the feed substrate concentration of the previous fermentation stage at time k, x 0,a ' is the optimal steady state of the cell state variables in the previous fermentation stage, The parameter matrix of the long-term comprehensive benefits calculated for the previous fermentation stage; yes The parameter matrix estimates corresponding to the submatrix parameters;

[0055] The comprehensive profit deviation between the current fermentation stage and the previous fermentation stage at time k is δ(k) = c b (k)-c a' (k);

[0056] The comprehensive profit deviation between the current fermentation stage and the previous fermentation stage at time k-1 is δ(k-1)=c b (k-1)-c a (k-1).

[0057] Preferably, based on the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage, the parameter matrix of the long-term comprehensive benefit of the current fermentation stage is estimated, and the formula includes:

[0058]

[0059] in, and are the comprehensive benefit deviation vectors at time k-1 and time k, U(k) is the difference gain in the current fermentation stage, and are the estimated values ​​of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage at time k-1 and time k, The parameter matrix of the long-term comprehensive benefits calculated for the previous fermentation stage.

[0060] Preferably, the feed substrate concentration in the current fermentation stage is optimized based on the estimated value of the parameter matrix of the long-term comprehensive benefits in the current fermentation stage, and the formula is:

[0061]

[0062] Among them, u b (k+1) is the feed substrate concentration of the current fermentation stage at time k+1, The parameter matrix W is the long-term comprehensive benefit of the current fermentation stage. b The sub-matrix of They are Estimated value.

[0063] Then the control quantity K of the current fermentation stage at time k+1 isb (k+1) is

[0064] The above technical solution of the present invention has the following beneficial effects compared with the prior art:

[0065] The present invention discloses a method for optimizing and controlling a continuous fermentation process at different stages based on parameter reuse. The method establishes a state space model for each fermentation stage in a continuous state variable fermentation process, establishes a comprehensive benefit optimization index for each fermentation stage, and calculates the feed substrate concentration for each fermentation stage when the comprehensive benefit optimization index is minimized, so that the cell culture state in the fermenter can be stabilized at the optimal steady-state value in each fermentation stage. In solving the feed substrate concentration for each fermentation stage after the initial fermentation stage, the present invention utilizes the similarity of each cell growth stage in the fermenter and uses the target feed substrate concentration of the previous fermentation stage and the parameter matrix of the long-term comprehensive benefit to solve the comprehensive benefit optimization index for the current fermentation stage. This method fully utilizes the known cell growth state and empirical culture parameter information, improves the algorithm's rapid startup performance, improves the solution efficiency, and enables the cell culture state in the fermenter to stabilize at the optimal steady-state value more quickly. Therefore, the present invention optimizes the control effect of the continuous fermentation process, improves the control performance of the fermentation process, thereby reducing fermentation costs, improving fermentation efficiency and fermentation quality, and maximizing economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings, wherein:

[0067] Figure 1 It is a diagram of a continuous fermentation tank system;

[0068] Figure 2 It is a flow chart of a method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse;

[0069] Figure 3 Schematic diagram of the optimized process of fermentation stage a and fermentation stage b in Example 2;

[0070] Figure 4 This is a comparison chart of control performance when the parameter reuse method is used in the fermentation process of utilis yeast. DETAILED DESCRIPTION

[0071] The present invention will be further described below with reference to the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. However, the embodiments are not intended to limit the present invention.

[0072] Example 1

[0073] Figure 1 This is a production process diagram of a continuous fermentation tank control system. The present invention provides a method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse. Figure 2 As shown, the steps include:

[0074] S1. Establish a state space model for the initial fermentation stage of the continuous fermentation process and design a state feedback controller. Based on the optimal steady state of the cell state variables in the initial fermentation stage, establish the comprehensive efficiency optimization indicators for the initial fermentation stage, including:

[0075] Establish a state-space model for the initial fermentation stage of a continuous fermentation process and design a state feedback controller, including:

[0076] According to the cell culture dynamics, with t as the sampling period, the state space model of the continuous fermentation process at the sampling time k in the initial fermentation stage is established as follows:

[0077] x a (k+1)=A a x a (k)+B a u a (k)

[0078] Among them, A a and B a is the state space model parameter of the initial fermentation stage, x a (k) and x a (k+1) represents the cell state variables at the initial fermentation stage at time k and time k+1, respectively, u a (k) represents the feed substrate concentration at the initial fermentation stage at time k.

[0079] Based on the state space model of the initial fermentation stage, the state feedback controller is designed as:

[0080]

[0081] Among them, K a is the control amount in the initial fermentation stage, x 0,a Represents the optimal steady state of cell state variables during the initial fermentation phase.

[0082] Specifically, x a (k) = [X a ,S a ,P a ] T , where X a 、S a 、P aThey represent the biomass concentration, substrate concentration, and product concentration at time k, respectively. The corresponding x 0,a =[X 0,a ,S 0,a ,P 0,a ] T , where X 0,a 、S 0,a 、P 0,a They represent the optimal steady-state biomass concentration, substrate concentration, and product concentration at time k, respectively.

[0083] There are many optimization objectives to consider during fermentation, and they must be set based on actual production needs. During continuous fermentation, the optimal steady-state of all state variables must be comprehensively considered. This means that biomass concentration, substrate concentration, and product concentration must all be stable at their corresponding optimal steady-state values ​​to achieve stable productivity and product quality.

[0084] Based on the optimal steady state of cell state variables in the initial fermentation stage, the comprehensive benefit optimization index of the initial fermentation stage is established as follows:

[0085]

[0086] Among them, C a (k) represents the comprehensive benefit optimization index of the initial fermentation stage at time k, It represents the difference between the measured value of the cell state in the initial fermentation stage and the optimal steady-state value at time k. E and F are both weight parameter matrices.

[0087] In order to maximize the benefits of the initial fermentation stage, it is necessary to find the comprehensive benefit optimization index c of the initial fermentation stage. a (k) Minimum target feed substrate concentration, i.e.

[0088] S2. Design a long-term comprehensive evaluation index for the feed substrate concentration in the initial fermentation stage to obtain a parameter matrix for the long-term comprehensive benefits of the initial fermentation stage, including:

[0089] Design a long-term comprehensive evaluation index for the feed substrate concentration in the initial fermentation stage. The formula is:

[0090] J a (k) = c a (k)+γJ a (k+1)

[0091] Among them, J a (k) and J a (k+1) are the long-term comprehensive evaluation indicators of the feed substrate concentration in the initial fermentation stage at time k and time k+1, respectively, and γ is the attenuation factor with a value of 0≤γ≤1.

[0092] When the feed substrate concentration u in the initial fermentation stage is a When (k) is the optimal value, J a (k+1) can be approximately regarded as The relevant functions, namely Then, in the initial fermentation stage, the long-term comprehensive evaluation index of feed substrate concentration can be approximately expressed as:

[0093]

[0094] in, is the parameter matrix of the long-term comprehensive benefits in the initial fermentation stage, and R is the weight parameter matrix; is x a (k),u a (k), x 0,a The relevant parameter vector, V a Yes and W a The parameter matrix corresponding to the sub-matrix parameters.

[0095] S3. Calculate the feed substrate concentration in the initial fermentation stage when the comprehensive benefit optimization index in the initial fermentation stage is minimized, and use it as the target feed substrate concentration in the initial fermentation stage.

[0096] When the state space model parameter A in the initial fermentation stage a and B a When the parameter matrix of the long-term comprehensive benefit of the initial fermentation stage is known, the parameter matrix of the long-term comprehensive benefit of the initial fermentation stage can be directly obtained. In addition, this embodiment adopts the minimum principle to solve the feed substrate concentration of the initial fermentation stage when the comprehensive benefit optimization index of the initial fermentation stage is minimized, that is, by solving the Riccati differential equation of the matrix, the quadratic objective equation Solve the optimization problem.

[0097] Preferably, the solution method for the known state space model parameters is not limited to analytical methods such as the minimum principle and dynamic programming method, but also includes numerical calculation method and gradient method.

[0098] When the state space model parameter A in the initial fermentation stage a and B aWhen it is unknown, the dynamic programming method is used to solve the comprehensive benefit optimization problem. This method requires the model structure to be known, and is also applicable to the case where the model parameters are known, and has low requirements for model parameters. The solution idea is: using a trial-and-error method, randomly select appropriate feed substrate concentrations during the fermentation process to repeatedly predict and optimize long-term comprehensive benefits. At the initial moment, that is, k = 0, under the constraint of the feed substrate concentration range in the actual fermentation process, a suitable initial feed substrate concentration is randomly selected; a short period of fermentation and sampling is carried out at this concentration, and the collected cell growth status data is combined with the comprehensive benefit optimization index, and the long-term comprehensive evaluation index of the feed substrate concentration is used to predict the long-term comprehensive benefits of the current feed substrate concentration. If the model parameters are unknown, the long-term comprehensive benefits are first estimated, that is, the parameter matrix W related to the long-term comprehensive benefits is estimated a ; and perform secondary optimization on the long-term comprehensive evaluation index of feed substrate concentration to obtain a better feed substrate concentration; then perform long-term comprehensive benefit prediction and optimization on the optimized feed substrate concentration, and repeat this process to obtain the target feed substrate concentration. The specific steps include:

[0099] S301. Estimating a parameter matrix of the long-term comprehensive benefits of the initial fermentation stage, including:

[0100] The long-term comprehensive evaluation index of the feed substrate concentration in the initial fermentation stage is expressed as:

[0101]

[0102] Build V a The relationship with the comprehensive benefit optimization index in the initial fermentation stage is as follows:

[0103] c a (k) = z a (k)V a

[0104] Among them, z a (k) = h a (k)-γh a (k+1);

[0105] Since x a (k),u a (k), x 0,a , E, F and other parameters are known, and the system identification method is used to solve the minimum comprehensive benefit optimization index V in the initial fermentation stage. a The estimated value of W a The estimated value of , the solution methods include but are not limited to recursive least squares method, maximum likelihood method, stochastic approximation method, multiple innovation identification method, auxiliary model identification, etc.

[0106] S302, the long-term comprehensive evaluation index J of the feed substrate concentration in the initial fermentation stage can be used a (k) for u a (k) The analytical method with zero partial derivatives is used to optimize the feed substrate concentration. The formula is:

[0107]

[0108] in, is the estimated value of the parameter matrix of the long-term comprehensive benefits of the initial fermentation stage at time k, The parameter matrix W of the long-term comprehensive benefits in the initial fermentation stage is a The sub-matrix of They are estimated value of;

[0109] Then the control quantity K of the initial fermentation stage at time k+1 is a (k+1) is

[0110] S303. Let k=k+1.

[0111] S304. Based on the optimized feed substrate concentration of the initial fermentation stage, return to S301 to re-estimate the parameter matrix of the long-term comprehensive benefits of the initial fermentation stage, and optimize the feed substrate concentration of the initial fermentation stage again until the estimated value of the parameter matrix of the long-term comprehensive benefits of the initial fermentation stage no longer changes, thereby obtaining the parameter matrix of the long-term comprehensive benefits of the initial fermentation stage and the target feed substrate concentration.

[0112] S4. For each fermentation stage after the initial fermentation stage, establish a state space model for the current fermentation stage and design a state feedback controller; based on the optimal steady state of the cell state variables in the current fermentation stage, establish comprehensive efficiency optimization indicators for the current fermentation stage, specifically including:

[0113] If during the fermentation process, at time k = τ, the fermentation process transitions from the previous fermentation stage to the current fermentation stage, the model parameters change. At this time, the optimal feed substrate concentration u a* It is not the optimal solution in the current fermentation stage.

[0114] According to the cell culture dynamics, with t as the sampling period, the state space model of the continuous fermentation process at the sampling time k in the current fermentation stage is re-established:

[0115] x b (k+1)=A b x b (k)+B b u b (k)

[0116] Among them, A b and B b is the state space model parameter of the current fermentation stage, x b (k) and x b (k+1) represents the cell state variables of the current fermentation stage at time k and time k+1, respectively, u b (x) represents the feed substrate concentration of the current fermentation stage at time k.

[0117] Based on the state space model of the current fermentation stage, the state feedback controller is designed as:

[0118]

[0119] Among them, K b is the control amount of the current fermentation stage, x 0,b Represents the optimal steady state of cell state variables in the current fermentation stage.

[0120] Specifically, x b (k) = [X b ,S b ,P b ] T , where X b 、S b 、P b They represent the biomass concentration, substrate concentration, and product concentration of the current fermentation stage at time k, respectively. The corresponding x 0,b =[X 0,b ,S 0,b ,P 0,b ] T , where X 0,b 、S 0,b 、P 0,b They represent the optimal steady-state biomass concentration, substrate concentration, and product concentration in the current fermentation stage at time k, respectively.

[0121] Based on the optimal steady state of the cell state variables in the current fermentation stage, the comprehensive benefit optimization index of the current fermentation stage is established as follows:

[0122]

[0123] Among them, c b (k) represents the comprehensive benefit optimization index of the current fermentation stage at time k, It represents the difference between the measured value of the cell state in the current fermentation stage at time k and the optimal steady-state value. E and F are both weight parameter matrices.

[0124] In order to maximize the benefits of the current fermentation stage, it is necessary to find the comprehensive benefit optimization index c of the current fermentation stage.b (k) Minimum target feed substrate concentration, i.e.

[0125] S5. Based on the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage, the comprehensive benefit optimization index of the current fermentation stage is solved using the target feed substrate concentration and the parameter matrix of the long-term comprehensive benefit of the previous fermentation stage, and the parameter matrix of the long-term comprehensive benefit of the current fermentation stage and the target feed substrate concentration are obtained when the comprehensive benefit optimization index of the current fermentation stage is minimized, specifically including:

[0126] S501. Design a long-term comprehensive evaluation index for the feed substrate concentration in the current fermentation stage. The formula is:

[0127]

[0128] in, It is the parameter matrix of the long-term comprehensive benefits of the current fermentation stage.

[0129] Since the model parameter A in the current fermentation stage b and B b If unknown, the estimation of the long-term comprehensive benefits of the current fermentation stage is converted into the parameter matrix W b If the optimal feed substrate concentration is determined by re-traversing the solution process, it will take a long time and the fermentation cost will be high. Therefore, in the control process of the current fermentation stage, this embodiment reuses the empirical control data of the previous fermentation stage, that is, the target feed substrate concentration u of the previous fermentation stage. a* Parameter matrix of long-term comprehensive benefits By constructing and solving optimization problems based on parameter reuse, we fully utilize the known cell growth status and empirical culture environment data, and the similarities between the two fermentation stages to improve the algorithm's rapid startup performance and the optimization effect, so that each cell culture state can quickly stabilize at the optimal steady-state value, thereby reducing fermentation costs and maximizing economic benefits.

[0130] S502. Initialize the parameter matrix of the long-term comprehensive benefits of the current fermentation stage and the feed substrate concentration, and calculate the comprehensive benefit optimization index deviation between the current fermentation stage and the previous fermentation stage using the target feed substrate concentration of the previous fermentation stage and the parameter matrix of the long-term comprehensive benefits.

[0131] When k = τ, let u b (k)=u a* , and initialize

[0132] Among them, u b (k) represents the initial feed substrate concentration of the current fermentation stage at time k, u a*represents the target feed substrate concentration of the previous fermentation stage, The estimated value of the parameter matrix representing the long-term comprehensive benefits of the current fermentation stage at time k.

[0133] When k>τ, based on the parameter matrix of the long-term comprehensive benefits of the previous fermentation stage, the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage is calculated, including:

[0134] The comprehensive benefit optimization index of the current fermentation stage at time k is

[0135] The comprehensive benefit optimization index of the previous fermentation stage at time k is

[0136] Among them, J a' (k) and J a' (k+1) are the long-term comprehensive evaluation indicators of the feed substrate concentration in the previous fermentation stage at time k and time k+1, γ is the attenuation factor, x a' (k) is the cell state variable of the previous fermentation stage at time k, u a' (k) is the feed substrate concentration of the previous fermentation stage at time k, x 0,a' is the optimal steady state of the cell state variables in the previous fermentation stage, The parameter matrix of the long-term comprehensive benefits calculated for the previous fermentation stage; yes The parameter matrix estimates corresponding to the submatrix parameters;

[0137] The comprehensive profit deviation between the current fermentation stage and the previous fermentation stage at time k is δ(k) = c b (k)-c a' (k);

[0138] The comprehensive profit deviation between the current fermentation stage and the previous fermentation stage at time k-1 is δ(k-1)=c b (k-1)-c a (k-1).

[0139] S503. Estimate a parameter matrix of the long-term comprehensive benefits of the current fermentation stage based on the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage.

[0140] At time k, the comprehensive benefit c known at time k-1 b (k-1), c a (k-1) and The estimated value of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage at time k-1 by the design difference gain U(k) Correction is made by using the similarity between the previous fermentation stage and the current fermentation stage to make the estimated value of the parameter matrix of the long-term comprehensive benefit of the current fermentation stage Approximating the true value W b , the formula includes:

[0141]

[0142] in, and are the comprehensive benefit deviation vectors at time k-1 and time k, U(k) is the difference gain in the current fermentation stage, and are the estimated values ​​of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage at time k-1 and time k, The parameter matrix of the long-term comprehensive benefits calculated for the previous fermentation stage.

[0143] S504. Optimize the feed substrate concentration in the current fermentation stage based on the estimated value of the parameter matrix of the long-term comprehensive benefits in the current fermentation stage.

[0144] The algorithms for optimizing the feed substrate concentration in the current fermentation stage include but are not limited to the interior point method, exterior point method, sequential quadratic programming method, genetic algorithm, particle swarm algorithm, etc. However, the optimization process is relatively complex. Therefore, this embodiment adopts the long-term comprehensive evaluation index J of the feed substrate concentration in the current fermentation stage. b (k) for u b (k) The analytical method with zero partial derivative is used to optimize the feed substrate concentration in the current fermentation stage. The formula is:

[0145]

[0146] Among them, u b (k+1) is the feed substrate concentration of the current fermentation stage at time k+1, The parameter matrix W is the long-term comprehensive benefit of the current fermentation stage. b The sub-matrix of They are Estimated value.

[0147] Then the control quantity K of the current fermentation stage at time k+1 is b (k+1) is

[0148] S505. Let k=k+1.

[0149] S506. Based on the estimated value of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage and the optimized feed substrate concentration of the current fermentation stage, return to S502 to recalculate the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage, and re-estimate the parameter matrix of the long-term comprehensive benefits of the current fermentation stage, and then optimize the feed substrate concentration of the current fermentation stage again until the estimated value of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage no longer changes, and obtain the parameter matrix of the long-term comprehensive benefits of the current fermentation stage and the target feed substrate concentration.

[0150] The present invention aims at the situation where the amount of initial cell culture status data at each stage of cell growth in a fermentation tank during continuous fermentation is small, especially under the condition that the model parameters are unknown or inaccurately known. It is suitable for continuous fermentation processes with known process model order and process requirements and measurable status.

[0151] In summary, the present invention discloses a method for optimizing and controlling a continuous fermentation process at different stages based on parameter reuse. By establishing a state space model for each fermentation stage in a continuous state variable fermentation process and establishing a comprehensive benefit optimization index for each fermentation stage, the feed substrate concentration of each fermentation stage is calculated to minimize the comprehensive benefit optimization index, so that the cell culture state in the fermenter can be stabilized at the optimal steady-state value in each fermentation stage. In the process of solving the feed substrate concentration of each fermentation stage after the initial fermentation stage, the present invention utilizes the similarity of each stage of cell growth in the fermenter, and uses the target feed substrate concentration of the previous fermentation stage and the parameter matrix of the long-term comprehensive benefit to solve the comprehensive benefit optimization index of the current fermentation stage. This fully utilizes the known cell growth state and empirical culture parameter information, improves the algorithm's rapid startup performance, improves the efficiency of the solution, and enables the cell culture state in the fermenter to stabilize at the optimal steady-state value more quickly. Therefore, the present invention optimizes the control effect of the continuous fermentation process, improves the control performance of the fermentation process, thereby reducing fermentation costs, improving fermentation efficiency and fermentation quality, and maximizing economic benefits.

[0152] Example 2

[0153] To verify the effect of the present invention, this example applies a method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse to a fermentation process using Torulopsis utilis.

[0154] First, the state space model of the initial fermentation stage in the continuous fermentation process, i.e., fermentation stage a, is established and the state feedback controller is designed:

[0155] x a (k+1)=A a x a (k)+B a u a(k)

[0156] Set the state space model parameters of fermentation stage a

[0157] Based on the state space model of fermentation stage a, the state feedback controller is designed as:

[0158]

[0159] The optimal steady-state x of the cell state variables in fermentation stage a 0,a =[X 0,a ,S 0,a ,P 0,a ] T =[7.038,2.404,25] T .

[0160] Based on the optimal steady state of cell state variables in fermentation stage a, the comprehensive benefit optimization index of fermentation stage a is established as:

[0161]

[0162] set up F=1.

[0163] At k=0, set the concentration parameters x of the initial cell culture state a (0) = [0, 8.5, 0] T , and randomly select the appropriate initial feed substrate concentration u a (0) = 5.1884. Solve the target feed substrate concentration u of fermentation stage a when the comprehensive benefit optimization index of fermentation stage a is minimized a* Parameter matrix of long-term comprehensive benefits

[0164] When the fermentation process enters the next fermentation stage b, the state space model of the continuous fermentation process of fermentation stage b with respect to sampling time k is re-established:

[0165] x b (k+1)=A b x b (k)+B b u b (k)

[0166] set up

[0167] Based on the state space model of fermentation stage b, the state feedback controller is designed as:

[0168]

[0169] The optimal steady-state x of the cell state variables in fermentation stage b 0,b =[X 0,b ,S 0,b ,P 0,b ] T =[7.038,2.404,25] T .

[0170] Based on the optimal steady state of the cell state variables in fermentation stage b, the comprehensive benefit optimization index of fermentation stage b is established as:

[0171]

[0172] set up F=1.

[0173] In the fermentation stage b, the comprehensive benefit optimization problem based on parameter reuse is solved, and the parameter matrix estimation value of the fermentation stage a is reused. and the optimal feed substrate concentration u a* To optimize the solution of the feed substrate concentration of fermentation stage b. At the beginning of fermentation stage b, that is, when k = τ, set the initial parameter matrix of fermentation stage b Let u b (τ)=u a* , at this time, the concentration parameters of the cell culture state are x b (τ)=[0,8.5,0] T .

[0174] Based on the deviation of the comprehensive benefit optimization index between fermentation stage b and fermentation stage a, the comprehensive benefit optimization index of fermentation stage b was solved using the target feed substrate concentration of fermentation stage a and the parameter matrix of long-term comprehensive benefits. The parameter matrix and target feed substrate concentration of long-term comprehensive benefits of fermentation stage b were obtained when the comprehensive benefit optimization index of fermentation stage b was minimized.

[0175] Figure 3 It is a schematic diagram of the optimized process of fermentation stage a and fermentation stage b.

[0176] The target feed substrate concentration of fermentation stage b is calculated using the solution method in S3 and the solution method based on parameter reuse in S5. Figure 4 The following chart compares the control performance of two methods for calculating the target feed substrate concentration in fermentation stage b. The time k required to achieve the optimal feed substrate concentration is 392 and 100, respectively. The parameter-reuse-based comprehensive benefit optimization method can quickly determine the optimal feed substrate concentration while improving control performance, allowing each cell culture state to quickly stabilize at the optimal steady-state value.

[0177] According to the above description, the method of the present invention realizes the optimization and control of the fermentation process of utilis yeast. It can be seen that the comprehensive benefit optimization problem solving based on parameter reuse makes full use of the known cell growth status and empirical culture parameter information. When the amount of cell culture status data is small at the beginning of a new fermentation stage, the similarity between the two is utilized to improve the optimization effect, enhance the rapid startup performance of the algorithm, and improve the control performance, thereby reducing the fermentation cost and maximizing the economic benefit.

[0178] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0179] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0180] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0181] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0182] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse, characterized in that: include: Based on the cell state variables and feed substrate concentration, a state space model of the initial fermentation stage in the continuous fermentation process was established and a state feedback controller was designed. Establish comprehensive benefit optimization indicators for the initial fermentation stage: c a (k) is the comprehensive benefit optimization index of the initial fermentation stage at time k, is the difference between the measured value of the cell state in the initial fermentation stage and the optimal steady-state value at time k, x a (k) is the cell state variable of the initial fermentation stage at time k, x 0,a is the optimal steady state of the cell state variables in the initial fermentation stage; u a (k) is the feed substrate concentration at the initial fermentation stage at time k; Design long-term comprehensive evaluation indicators of feed substrate concentration in the initial fermentation stage: J a (k) and J a (k+1) is the long-term comprehensive evaluation index of the feed substrate concentration in the initial fermentation stage at time k and time k+1, and the parameter matrix of the long-term comprehensive benefit of the initial fermentation stage is γ is the attenuation factor, E, F and R are weight parameter matrices, A a and B a are the state space model parameters of the initial fermentation stage; Solve the feed substrate concentration that minimizes the comprehensive benefit optimization index in the initial fermentation stage, as the target feed substrate concentration for the initial fermentation stage; For each fermentation stage after the initial fermentation stage, a state space model of the current fermentation stage is established and a state feedback controller is designed; based on the optimal steady state of the cell state variables in the current fermentation stage, the comprehensive benefit optimization index of the current fermentation stage is established; Design a long-term comprehensive evaluation index for feed substrate concentration in the current fermentation stage; Initialize the parameter matrix of the long-term comprehensive benefits of the current fermentation stage and the feed substrate concentration, and use the target feed substrate concentration of the previous fermentation stage and the parameter matrix of the long-term comprehensive benefits to calculate the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage; Based on the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage, the parameter matrix of the long-term comprehensive benefit of the current fermentation stage is estimated; Optimizing the feed substrate concentration in the current fermentation stage based on the estimated value of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage; Based on the estimated value of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage and the optimized feed substrate concentration of the current fermentation stage, return to recalculate the comprehensive benefit optimization index deviation between the current fermentation stage and the previous fermentation stage, and re-estimate the parameter matrix of the long-term comprehensive benefits of the current fermentation stage, and then optimize the feed substrate concentration of the current fermentation stage again until the estimated value of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage no longer changes, and obtain the parameter matrix of the long-term comprehensive benefits of the current fermentation stage and the target feed substrate concentration when the comprehensive benefit optimization index is minimized.

2. The method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse according to claim 1, characterized in that: Establish a state-space model for the initial fermentation stage of a continuous fermentation process and design a state feedback controller, including: The state space model of the initial fermentation stage is: x a (k+1)=A a x a (k)+B a u a (k) Among them, A a and B a is the state space model parameter of the initial fermentation stage, x a (k) and x a (k+1) represents the cell state variables at the initial fermentation stage at time k and time k+1, respectively, u a (k) represents the feed substrate concentration at the initial fermentation stage at time k; Based on the state space model of the initial fermentation stage, the state feedback controller is designed as: u a (k)=K a (x a (k)-A a-1 x 0,a ) Among them, K a is the control amount in the initial fermentation stage, x 0,a Represents the optimal steady state of cell state variables during the initial fermentation phase.

3. The method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse according to claim 1, characterized in that: When the state space model parameters of the initial fermentation stage are unknown, the feed substrate concentration at which the comprehensive benefit optimization index of the initial fermentation stage is minimized is solved as the target feed substrate concentration of the initial fermentation stage, including: First, the parameter matrix of the long-term comprehensive benefits in the initial fermentation stage is estimated, including: The long-term comprehensive evaluation index of the feed substrate concentration in the initial fermentation stage is expressed as: in, is x a (k),u a (k), x 0,a The relevant parameter vector, V a Yes and W a The parameter matrix corresponding to the sub-matrix parameters; Establish V a The relationship with the comprehensive benefit optimization index in the initial fermentation stage is as follows: c a (k)=z a (k)V a Among them, z a (k) = h a (k)-γh a (k+1); The system identification method is used to solve the minimum comprehensive benefit optimization index V in the initial fermentation stage. a The estimated value of W a estimated value of; The feed substrate concentration in the initial fermentation stage is optimized based on the parameter matrix of the long-term comprehensive benefits of the initial fermentation stage. The formula is: in, is the estimated value of the parameter matrix of the long-term comprehensive benefits of the initial fermentation stage at time k, The parameter matrix W of the long-term comprehensive benefits in the initial fermentation stage is a The sub-matrix of They are estimated value of; Then the control quantity K of the initial fermentation stage at time k+1 is a (k+1) is Based on the optimized feed substrate concentration in the initial fermentation stage, the parameter matrix for re-estimating the long-term comprehensive benefits of the initial fermentation stage is returned, and the feed substrate concentration in the initial fermentation stage is optimized again until the estimated value of the parameter matrix for the long-term comprehensive benefits of the initial fermentation stage no longer changes, thereby obtaining the parameter matrix for the long-term comprehensive benefits of the initial fermentation stage and the target feed substrate concentration.

4. The method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse according to claim 1, characterized in that: When the state space model parameters of the initial fermentation stage are known, the parameter matrix of the long-term comprehensive benefits of the initial fermentation stage is directly obtained; The feed substrate concentration when the comprehensive benefit optimization index of the initial fermentation stage is minimized is solved as the target feed substrate concentration of the initial fermentation stage. The methods include minimum principle, dynamic programming method, numerical calculation method and gradient method.

5. The method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse according to claim 1, characterized in that: Design long-term comprehensive evaluation indicators for feed substrate concentration in the current fermentation stage, including: Among them, J b (k) and J b (k+1) are the long-term comprehensive evaluation indicators of the feed substrate concentration in the current fermentation stage at time k and time k+1, respectively, c b (k) is the comprehensive benefit optimization index of the current fermentation stage at time k, γ is the attenuation factor, x b (k) is the cell state variable of the current fermentation stage at time k, u b (k) is the feed substrate concentration of the current fermentation stage at time k, is the optimal steady state of the cell state variables in the current fermentation stage; is the parameter matrix of the long-term comprehensive benefits of the current fermentation stage, A b and B b are the state space model parameters of the current fermentation stage, and E, F, and R are all weight parameter matrices.

6. The method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse according to claim 5, characterized in that Initialize the parameter matrix of the long-term comprehensive benefit of the current fermentation stage and the feed substrate concentration. Use the target feed substrate concentration of the previous fermentation stage and the parameter matrix of the long-term comprehensive benefit to calculate the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage, including: Assuming that the initial time of the current fermentation stage is τ, when k = τ, let u b (k)=u a* , And initialize Among them, u b (k) represents the initial feed substrate concentration of the current fermentation stage at time k, u a* represents the target feed substrate concentration of the previous fermentation stage, The estimated value of the parameter matrix representing the long-term comprehensive benefits of the current fermentation stage at time k; When k>τ, based on the parameter matrix of the long-term comprehensive benefits of the previous fermentation stage, the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage is calculated, including: The comprehensive benefit optimization index of the current fermentation stage at time k is in, is the difference between the measured value of the cell state in the current fermentation stage and the optimal steady-state value at time k, E and F are both weight parameter matrices, u b (k) is the feed substrate concentration of the current fermentation stage at time k; The comprehensive benefit optimization index of the previous fermentation stage at time k is Among them, J a ′(k) and J a ′(k+1) is the long-term comprehensive evaluation index of the feed substrate concentration in the previous fermentation stage at time k and time k+1, γ is the attenuation factor, x a ′(k) is the cell state variable of the previous fermentation stage at time k, u a ′(k) is the feed substrate concentration of the previous fermentation stage at time k, x 0,a ′ is the optimal steady state of the cell state variables in the previous fermentation stage, The parameter matrix of the long-term comprehensive benefits calculated for the previous fermentation stage; yes The parameter matrix estimates corresponding to the submatrix parameters; The comprehensive profit deviation between the current fermentation stage and the previous fermentation stage at time k is δ(k) = c b (k)-c a ′(k); The comprehensive profit deviation between the current fermentation stage and the previous fermentation stage at time k-1 is δ(k-1)=c b (k-1)-c a (k-1).

7. The method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse according to claim 6, characterized in that: Based on the deviation of the comprehensive benefit optimization index between the current fermentation stage and the previous fermentation stage, the parameter matrix of the long-term comprehensive benefit of the current fermentation stage is estimated. The formula includes: in, and are the comprehensive benefit deviation vectors at time k-1 and time k, U(k) is the difference gain in the current fermentation stage, and are the estimated values ​​of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage at time k-1 and time k, The parameter matrix of the long-term comprehensive benefits calculated for the previous fermentation stage.

8. The method for optimizing and controlling continuous fermentation processes at different stages based on parameter reuse according to claim 7, characterized in that: Based on the estimated value of the parameter matrix of the long-term comprehensive benefits of the current fermentation stage, the feed substrate concentration of the current fermentation stage is optimized. The formula is: Among them, u b (k+1) is the feed substrate concentration of the current fermentation stage at time k+1, The parameter matrix W is the long-term comprehensive benefit of the current fermentation stage. b The sub-matrix of They are estimated value; Then the control quantity K of the current fermentation stage at time k+1 is b (k+1) is

Images