CN109976157B - A kind of intelligent liquid fermentation parameter control method for food - Google Patents

Abstract

The invention discloses a method for controlling parameters of intelligent liquid fermentation of food. The method includes step 1, collecting polyphenol content and color information of fermentation liquid in the oxidative fermentation process of summer and autumn tea; step 2, taking the polyphenol content and color information as fuzzy reasoning Input factors, take the temperature of oxidative fermentation of summer and autumn tea, PH of fermentation broth, dissolved oxygen value DO of fermentation broth, and rotational speed of fermentation broth as output factors of fuzzy reasoning; build a fuzzy control system; step 3, determine the input and output factors of fuzzy control system step 4, formulate fuzzy division and membership function of input and output factors; step 5, formulate probability coupling rules and fuzzy control table of input and output parameters; step 6, decoupling of fuzzy reasoning and improved Markov method ; Combine the fermentation control data provided by operators with long-term production experience to construct reasonable multi-variable fuzzy control rules to make the control of fermentation variables more accurate.

Description

A kind of intelligent liquid fermentation parameter control method for food

technical field

The invention designs the field of fermentation control, mainly a control method and system for various input and output parameters in the fermentation process.

Background technique

Fermentation technology has applications in food, medicine, chemical industry and other fields, especially in the food industry. Fermented food has a very large consumer group, thus giving birth to a huge fermented food industry; food itself is a complex system, and the production of fermented food In the process, the control of various production conditions in complex systems has always been a recognized problem both inside and outside the industry.

Fuzzy Control is derived from Fuzzy Theory, a computer digital control technology based on fuzzy set theory, fuzzy reasoning, and fuzzy variables. It belongs to the category of intelligent bionic control and is essentially a nonlinear control. , Fuzzy control has a wide range of applications in common control systems, and has the advantages of wide adaptability, no dependence on specific models, fast parameter feedback, and accurate control. However, in actual use, the input and output fuzzy control of a single variable is mostly used, which cannot meet the needs of complex parameter control in the food industry.

The traditional control of production parameters in the fermentation process is to directly use PID for steady-state adjustment. However, this method requires repeated adjustment of PID parameters. In actual use, there are disadvantages such as long debugging time and untimely parameter control.

For example, the application number "CN201510777283.5" discloses a fuzzy control system for a non-negative pressure water supply unit and its fuzzy control method. The fuzzy control algorithm is used to improve the steady-state accuracy of the control system. The non-linearity, hysteresis and timely change of the pressure water supply unit keep the outlet water pressure of the water supply unit constant, which can ensure the water quality and safety.

For example, the application number "CN200910085548.X" discloses a fuzzy control method and a fuzzy controller, including a detection input input from a detection device, and mapping the deviation value between the detection input and the set value to the input universe to obtain Fuzzy input quantity; finally get the fuzzy output quantity, convert it into output quantity, and transmit it to the control mechanism, so that the chlorination of the water plant can achieve the best control effect.

However, the above-mentioned published patents have relatively single input and output variables in the fuzzy control system, and do not provide a multi-variable complex fuzzy control system, which cannot be applied to the complex process parameter control of the food liquid fermentation process.

SUMMARY OF THE INVENTION

The present invention is to propose a multi-variable fermentation control method and system based on fuzzy control, which can replace the current fermentation parameter control method in food liquid fermentation production, realize a liquid food fermentation control based on fuzzy theory, and solve the problem of current food fermentation control. The problem that the controlled object cannot be accurately controlled due to the characteristics of nonlinearity, strong coupling, time-varying and lag in the process.

The technical scheme of the method of the present invention is: an intelligent liquid fermentation parameter control method for food, which is characterized by comprising the following steps:

Step 1, collecting the polyphenol content and color information of the fermentation broth in the oxidative fermentation process of summer and autumn tea;

Step 2, take the polyphenol content and color information as input factors of fuzzy reasoning, and take the temperature of oxidative fermentation of summer and autumn tea, pH of fermentation broth, dissolved oxygen value DO of fermentation broth, and rotational speed of fermentation broth as output factors of fuzzy reasoning; Control System;

Step 3, determine the change domain of the input and output factors of the fuzzy control system;

Step 4, formulate fuzzy division and membership function of input and output factors;

Step 5, formulate the input and output parameter probability coupling rules and fuzzy control table;

Step 6: Decoupling of fuzzy reasoning and improved Markov method;

In step 7, the four output factors in the oxidative fermentation process of summer and autumn tea are used as control parameters, and are input into the actuator to adjust the heating device in the actuator, the acid-base pump for changing the pH value of the fermentation broth, and the The air pump for the dissolved oxygen value of the fermentation liquid and the stirring motor for changing the stirring speed of the fermentation liquid.

Further, the fuzzy control system is an improved fuzzy control system with dual inputs and four outputs.

Further, in step 3, the change domain of the input and output factors is: the content of polyphenols varies from 1.4 to 2.1, the Euclidean distance of the color ranges from 0 to 15.6, the fermentation temperature: 25-70 °C; the pH of the fermentation broth: 6.0 ~8.0; dissolved oxygen value DO of fermentation broth: 0 to 100; rotational speed of fermentation broth stirring: 20 to 300.

Further, the membership function is selected as: triangular piecewise membership function.

Further, the division rules of the fuzzy sets of polyphenol content are as follows:

Very low VL∈(1.4,1.5); low LOW∈(1.5,1.6)∪(1.6,1.7); somewhat low RL∈(1.6,1.7)∪(1.7,1.8); suitable MED∈(1.7,1.8)∪ (1.8, 1.9); somewhat high RH ∈ (1.8, 1.9) ∪ (1.9, 2.0); high H ∈ (1.9, 2.0) ∪ (2.0, 2.1); very high VH ∈ (2.0, 2.1);

The fuzzy set partition rules for the color value of the fermentation broth are as follows:

Suitable MED∈(0,2); more suitable RM∈(0,2)∪(2,4); slightly higher LH∈(2,4)∪(4,6); slightly higher RH∈(4,6) ∪(6,8); high H∈(6,8)∪(8,10); very high VH∈(8,12)∪(12,15);

The temperature fuzzy set division rules for fermentation are as follows:

Low temperature MINT∈(25,34); lower temperature LT∈(25,34)∪(34,43); medium temperature MT∈(34,43)∪(43,52); higher temperature PT∈(43,52)∪ (52,61); high temperature MAXT∈(52,61)∪(61,70);

The fuzzy set partition rules for pH of fermentation broth are as follows:

Acid MINpH ∈ (6.0, 6.5); somewhat acidic LpH ∈ (6.0, 6.5) ∪ (6.5, 7.0); neutral MpH ∈ (6.5, 7.0) ∪ (7.0, 7.5); somewhat alkaline PpH ∈ (7.0, 7.5) ∪(7.5,8.0); base MAXpH∈(7.5,8.0);

The fuzzy set division rules of the dissolved oxygen value DO of the fermentation broth are as follows:

Low MINDO∈(0,25); Low LDO∈(0,25)∪(25,50); Medium MDO∈(25,50)∪(50,70); High PDO∈(50,75)∪ (75,100); High MAXDO ∈ (75,100);

The fuzzy set division rules of the stirring speed of the fermentation broth are as follows:

Very low MINS∈(20,67); low LS∈(20,67)∪(67,114); low RS∈(67,114)∪(114,161); medium speed MS∈(114,161)∪(161,208); high PS ∈(161,208)∪(208,255); High HS∈(208,255)∪(255,300); Very High MAXS∈(255,300);

Further, the specific process of decoupling fuzzy reasoning and improved Markov method is as follows:

This fuzzy inference and improved Markov method decoupling model is composed as follows:

For the collected polyphenol content and color information of the fermentation broth, as well as the fermentation temperature, pH of the fermentation broth, dissolved oxygen value DO of the fermentation broth, and stirring speed of the fermentation broth that need to be output to the control mechanism, calculate the weight of the model for each parameter combination, and calculate the weights of the models with multiple parameter combinations. A model is used as an observation sequence, and the forward factor α _t (i) is used to initialize the forward factor, α ₁ (i)=π _i b _i (Y ₁ ), where 1≤i≤N, Y1 is the initial moment in the time series The probability of ; use the recursive method to continuously calculate the weight, and gradually recurse α _t+1 (j)=[∑α _t (i)α _ij ]b _j (Y _t+1 ) from front to back, where 1≤t≤T -1, 1≤j≤N, α _t is the probability of the observation sequence at time t, b _j is the probability of the observation sequence given the Markov model;

The fermentation control model is regarded as an observation sequence, O=O ₁ O ₂ ,...,O _T observation model is λ=(A, B, π), calculate P(O|λ), and calculate the value of each parameter. How well the data fit the observation sequence as a given model;

Choose a certain Markov model λ _i ={A _i ,B _i ,π _i },i=1,2,...,C, where A _i , B _i , π _i are all model parameters; for A given fermentation model observation sequence O=O ₁ ,O ₂ ,..., _OT and the model parameters of the hidden Markov model λ _i ,i=1,2,...,C, where O _T is a factor The observed state of O at time T;

For a particular state sequence Q=q ₁ ,q ₂ ,...,q _T , the probability of producing the observation sequence O=O ₁ O ₂ ,...,O _T is:

where b _qT is the probability that the probability model observes the sequence O at t=T. For a given model parameter λ, the probability of generating the state sequence Q=q ₁ , q ₂ ,...,q _T is: P(Q|λ ) = π _q1 a _q1q2 a _q2q3 ··· a _qT-1qT (4.11)

Among them, α _qT-1qT is the function parameter. In order to calculate the probability that the model produces the observation sequence O=O ₁ O ₂ ,...,O _T , each hidden state sequence must be taken into account, and the calculation of each of them produces the observation sequence O =O ₁ O ₂ ,...,O _T probability, and then summed, therefore, the obtained probability is:

It can be known from equation (4.12) that the probability of the observation sequence O=O ₁ O ₂ ,..., _OT is equal to the sum of the probabilities of all hidden state sequences that may generate this observation sequence. The algorithm based on the recursive idea of the forward method calculates P(O|λ), which reduces the time complexity of the algorithm to N ² T, where N is the dimension of the observation sequence. The decoupling value of each fermentation control parameter in the coupled state can be quickly solved.

The fermentation system includes a fermentation tank, a control actuator, a control system, and a sensor system; the fermentation tank is combined with a sensor, and the sensor system includes a temperature sensor, a pH sensor, and a visible/near-infrared spectrum sensor; the control actuator includes a stirring motor, a temperature control Module, acid-base pump; control system includes main control computer, PLC;

The sensor is connected with the PLC, and the PLC collects the data of the sensor; the PLC is connected with the main control computer, the main control computer sends instructions to the PLC, and the PLC is connected with the control actuator to directly control the action of the actuator; the main control computer, After receiving the sensor data, the fuzzy control system is processed to obtain control parameters, which are then transmitted to the PLC; the actuator is directly connected to the fermenter to control the production conditions in the fermenter.

Further, the construction method of the fuzzy control system includes:

Obtain two or more key state values of the fermentation broth and use them as the input value X _k of the fuzzy control system; obtain the key production control parameters in the fermentation process according to production experience and experiments, and use them as the output value of the fuzzy control system Y _k ;

Formulate a fuzzy control rule table (FuzzyControlTable), fuzzify the input value _Xk of the fuzzy control system, and then de-fuzzy it through the fuzzy membership function to obtain the output value _Yk accurate value that can be used for production adjustment;

The present invention designs a fuzzy control model for liquid fermentation of food, including:

The parameter input part is used to receive the fermentation key state parameters measured by the detection device, including the content, color, viscosity, turbidity, etc. of some key components, and bring the above input into the fuzzy control rule table and map it to the input universe , get the fuzzy value of the input parameter;

The parameter processing part performs fuzzy reasoning and decision-making on the fuzzy value of the input value, and obtains the fuzzy value corresponding to the output value;

The parameter output part de-fuzzifies the fuzzy control parameters obtained by the above parameter processing part to obtain the control output, and presents the control output to the execution and action mechanism for actual control, mainly including temperature control of the fermenter, stirring Speed control, ventilation volume control, pH control;

It can be seen from the above technical solutions that the present invention designs an improved fuzzy control method, which uses multiple fermentation states as input quantities and multiple control parameters as output quantities. For the precise control of fermentation parameters and fermentation quality, a closed-loop multi-variable fuzzy control method and system is constructed to achieve the purpose of saving energy and reducing emissions and improving product quality.

The beneficial effects of the present invention are: (1) based on the improved fuzzy control multi-variable input and output control system, combined with the Markov model for defuzzification and decoupling of output parameters, which can accurately control complex food liquid fermentation conditions in real time Simultaneous control of various fermentation variables to achieve the purpose of energy saving and emission reduction;

(2) Combining the fermentation control data provided by operators with long-term production experience, construct reasonable multi-variable fuzzy control rules to make the control of fermentation variables more accurate;

Description of drawings

Fig. 1 is the control flow chart of tea extract fermentation fuzzy system

Fig. 2 is the fuzzy control rule surface of the fermentation temperature parameters of the tea extract

Figure 3 is the flow chart of the intelligent liquid fermentation control system for food

Detailed ways

The fermentation system includes a fermentation tank, a control actuator, a control system, and a sensor system; the fermentation tank is combined with a sensor, and the sensor system includes a temperature sensor, a pH sensor, and a visible/near-infrared spectrum sensor; the control actuator includes a stirring motor, a temperature control Module, acid-base pump; control system includes main control computer, PLC;

The sensor is connected with the PLC, and the PLC collects the data of the sensor; the PLC is connected with the main control computer, the main control computer sends instructions to the PLC, and the PLC is connected with the control actuator to directly control the action of the actuator; the main control computer, After receiving the sensor data, the fuzzy control system is processed to obtain control parameters, which are then transmitted to the PLC; the actuator is directly connected to the fermenter to control the production conditions in the fermenter. This example designs a fuzzy control system for tea extract fermentation based on fuzzy control. The main steps include:

Step 1, collecting the polyphenol content and color information of the fermentation broth in the oxidative fermentation process of summer and autumn tea;

Step 2, take the polyphenol content and color information as input factors of fuzzy reasoning, and take the temperature of oxidative fermentation of summer and autumn tea, pH of fermentation broth, dissolved oxygen value DO of fermentation broth, and rotational speed of fermentation broth as output factors of fuzzy reasoning; Control System;

Step 3, determine the change domain of the input and output factors of the fuzzy control system;

Step 4, formulate fuzzy division and membership function of input and output factors;

Step 5, formulate the input and output parameter probability coupling rules and fuzzy control table;

Step 6: Decoupling of fuzzy reasoning and Markov method;

In step 7, the four output factors in the oxidative fermentation process of summer and autumn tea are used as control parameters, and are input into the actuator to adjust the heating device in the actuator, the acid-base pump for changing the pH value of the fermentation broth, and the The air pump for the dissolved oxygen value of the fermentation liquid and the stirring motor for changing the stirring speed of the fermentation liquid. Further explanation: In Figure 1, 1 is the two key parameters of the tea extract received by the system, polyphenol content and color value; 2 is the improved fuzzy control system and the system construction process; 3 is the parameter probability coupling step of the improved fuzzy control system ; 4 is the fuzzy calculation reasoning; 5 is the decoupling step of the Markov method of the output parameters; 6 is the system output control parameters; 7 is the control object—the extract of summer and autumn tea; 8 is the fermentation parameters (temperature, pH, dissolved oxygen, 9 is the control parameters temperature, pH, dissolved oxygen and rotational speed that improve the actual output of the fuzzy control system;

The main steps in establishing this improved fuzzy control system are:

Step 1: Determine the change domain of input and output factors;

Step 2: Formulate fuzzy division and membership function of input and output factors;

Step 3: Formulate parameter probability coupling rules and fuzzy control table;

Step 4: Decoupling of fuzzy reasoning and Markov method.

The specific steps are as follows:

Step 1: Determine the change domain of input and output factors;

The fuzzy inference system detects the polyphenol content and color of the fermentation broth in the process of oxidative fermentation of summer and autumn tea every 2 minutes. From the accumulated experimental data, it can be seen that the polyphenol content varies from 1.4 to 2.1, and the Euclidean distance of the color ranges from 0 to 0. between 15.6. The system includes four output factors, the ranges of which are temperature: 25-70℃; pH: 6.0-8.0; DO: 0-100; speed: 20-300.

Change domain of input and output variables

Step 2: Formulate fuzzy division and membership function of input and output factors;

The choice of membership function has a great influence on the performance of fuzzy inference system. The triangular membership function is selected in this system, and the division rules of the input parameter polyphenol fuzzy set are as follows:

Very low VL ∈ (1.4, 1.5); low LOW ∈ (1.5, 1.6) ∪ (1.6, 1.7); somewhat low RL ∈ (1.6, 1.7) ∪ (1.7, 1.8); suitable MED ∈ (1.7, 1.8) ∪ (1.8, 1.9); somewhat high RH ∈ (1.8, 1.9) ∪ (1.9, 2.0); high H ∈ (1.9, 2.0) ∪ (2.0, 2.1); very high VH ∈ (2.0, 2.1);

The fuzzy partition table of input and output factors is as follows

Fuzzy Set Partitioning of Polyphenols

The fuzzy set partition rules of the input parameter color Euclidean distance are as follows, respectively

Suitable MED∈(0,2); more suitable RM∈(0,2)∪(2,4); slightly higher LH∈(2,4)∪(4,6); slightly higher RH∈(4,6) ∪(6,8); high H∈(6,8)∪(8,10); very high VH∈(8,12)∪(12,15);

The fuzzy set rule table is as follows:

Fuzzy Set Partitioning of Color Euclidean Distance

The fuzzy set partition rules for the output parameter temperature are as follows:

Low temperature MINT∈(25,34); lower temperature LT∈(25,34)∪(34,43); medium temperature MT∈(34,43)∪(43,52); higher temperature PT∈(43,52)∪ (52, 61); high temperature MAXT∈(52, 61)∪(61, 70);

The fuzzy set partition rule table for the output parameter temperature:

Fuzzy set partitioning of temperature

The fuzzy set partition rule for the output parameter pH is as follows

Acid MINpH ∈ (6.0, 6.5); somewhat acid LpH ∈ (6.0, 6.5) ∪ (6.5, 7.0); neutral MpH ∈ (6.5, 7.0) ∪ (7.0, 7.5); somewhat alkaline PpH ∈ (7.0, 7.5) ∪(7.5, 8.0); base MAXpH∈(7.5, 8.0);

The Partitioning of Fuzzy Sets of pH

The fuzzy set partition rules for the output value DO are as follows:

Low MINDO ∈ (0, 25); lower LDO ∈ (0, 25) ∪ (25, 50); medium MDO ∈ (25, 50) ∪ (50, 70); higher PDO ∈ (50, 75) ∪ (75, 100); high MAXDO ∈ (75, 100);

Fuzzy Set Partitioning of DO Values

Very low MINS ∈ (20, 67); low LS ∈ (20, 67) ∪ (67, 114); low RS ∈ (67, 114) ∪ (114, 161); medium speed MS ∈ (114, 161) ∪(161, 208); high PS ∈ (161, 208)∪(208, 255); high HS ∈ (208, 255)∪(255, 300); extremely high MAXS ∈ (255, 300). ;

Fuzzy set partitioning of rotational speed

Step 3: Formulate parameter probability coupling rules and fuzzy control table;

The fuzzy decision-making system designed in this system is a dual-input, four-output fuzzy inference system, which is decomposed into four dual-input, single-output fuzzy inference systems, corresponding to the four output factors temperature, pH, DO, Rotating speed. Polyphenol content was divided into 7 linguistic variables, color was divided into 6 linguistic variables, and output factors were divided into 5 and 7 linguistic variables. Under the guidance of experienced production experts, combined with long-term production experience, the fuzzy control rules are reasonably formulated.

The fuzzy control rules for temperature are as follows:

Some fuzzy control logic statements are as follows:

1.If(duofen is VL)and(color is MED)then(temp is MINT) (1)

2.If(duofen is VL)and(color is RM)then(temp is MINT) (1)

3.If(duofen is VL)and(color is LH)then(temp is LT) (1)

4.If(duofen is VL)and(color is RH)then(temp is MT) (1)

5.If(duofen is VL)and(color is H)then(temp is PT) (1)

6.If(duofen is VL)and(color is VH)then(temp is PT) (1)

7.If(duofen is LOW)and(color is MED)then(temp is MINT) (1)

8.If(duofen is LOW)and(color is RM)then(temp is MINT)(1)

9.If(duofen is LOW)and(color is LH)then(temp is LT) (1)

10.If(duofen is LOW)and(color is RH)then(temp is MT) (1)

11.If(duofen is LOW)and(color is H)then(temp is PT) (1)

12.If(duofen is LOW)and(color is VH)then(temp is PT) (1)

13.If(duofen is RL)and(color is MED)then(temp is MINT) (1)

14.If(duofen is RL)and(color is RM)then(temp is MINT) (1)

15.If(duofen is RL)and(color is LH)then(temp is LT)(1)

...

Fuzzy policy control table for temperature:

Temperature fuzzy strategy control table

The fuzzy strategy control table of pH

Temperature fuzzy strategy control table

The Fuzzy Policy Control Table of DO

The Fuzzy Policy Control Table of DO

Fuzzy strategy control table of rotational speed

Fuzzy strategy control table of rotational speed

Step 4: Decoupling of fuzzy reasoning and Markov method.

This fuzzy inference and Markov method decoupling model consists of the following:

For the collected polyphenol content and color information of the fermentation broth, as well as the fermentation temperature, pH of the fermentation broth, dissolved oxygen value DO of the fermentation broth, and stirring speed of the fermentation broth that need to be output to the control mechanism, calculate the weight of the model for each parameter combination, and calculate the weights of the models with multiple parameter combinations. A model is used as an observation sequence, and the forward factor α _t (i) is used to initialize the forward factor, α ₁ (i)=π _i b _i (Y ₁ ), where 1≤i≤N, Y1 is the initial moment in the time series The probability of ; use the recursive method to continuously calculate the weight, and gradually recurse α _t+1 (j)=[∑α _t (i)α _ij ]b _j (Y _t+1 ) from front to back, where 1≤t≤T -1, 1≤j≤N, α _t is the probability of the observation sequence at time t, b _j is the probability of the observation sequence given the Markov model;

The fermentation control model is regarded as an observation sequence, O=O ₁ O ₂ ,...,O _T observation model is λ=(A, B, π), calculate P(O|λ), and calculate the value of each parameter. How well the data fit a sequence of observations as a given model.

Select a certain Markov model λ _i ={A _i ,B _i ,π _i },i=1,2,...,C, where A _i , B _i , π _i are all model parameters. For a given fermentation model observation sequence O=O ₁ ,O ₂ ,..., _OT and the model parameters of the hidden Markov model λ _i ,i=1,2,...,C, where O _T is The observed state of factor O at time T;

For a particular state sequence Q=q ₁ ,q ₂ ,...,q _T , the probability of producing the observation sequence O=O ₁ O ₂ ,...,O _T is:

where b _qT is the probability that the probability model observes the sequence O at t=T. For a given model parameter λ, the probability of generating the state sequence Q=q ₁ , q ₂ ,...,q _T is: P(Q|λ ) = π _q1 a _q1q2 a _q2q3 ··· a _qT-1qT (4.11)

Among them, α _qT-1qT is the function parameter. In order to calculate the probability that the model produces the observation sequence O=O ₁ O ₂ ,...,O _T , each hidden state sequence must be taken into account, and the calculation of each of them produces the observation sequence O =O ₁ O ₂ ,...,O _T probability, and then summed, therefore, the obtained probability is:

It can be known from equation (4.12) that the probability of the observation sequence O=O ₁ O ₂ ,..., _OT is equal to the sum of the probabilities of all hidden state sequences that may generate this observation sequence. The algorithm based on the recursive idea of the forward method calculates P(O|λ) reduces the time complexity of the algorithm to N ² T, and can quickly solve the decoupling value of each fermentation control parameter in the coupled state. Finally, a fuzzy control surface is formed, as shown in Figure 2, the fuzzy control rule surface of the fermentation temperature parameters of tea extract.

By adopting the fuzzy control method, the fuzzy controller in this embodiment obtains the fuzzy control quantity according to the fuzzy theory according to the detection 2 input quantities input by the detection device, and controls the controlled object according to the 4 control output quantities after defuzzification, which overcomes the problem of the current situation. There are no unified mathematical model and inaccurate control caused by nonlinear, strong coupling, time-varying and hysteresis characteristics in the control process, which makes the control process more accurate and achieves better control effects.

In the description of this specification, reference to the terms "one embodiment," "some embodiments," "exemplary embodiment," "example," "specific example," or "some examples," or the like, is meant to incorporate the embodiment. A particular feature, structure, material, or characteristic described by an example or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, The scope of the invention is defined by the claims and their equivalents.

Claims (5)

1. An intelligent food liquid fermentation parameter control method is characterized by comprising the following steps:

step 1, collecting polyphenol content and color information of fermentation liquor in the oxidative fermentation process of summer and autumn tea;

step 2, taking the polyphenol content and the color information as input factors of fuzzy reasoning, and taking the temperature of oxidative fermentation of the summer and autumn tea, the pH value of fermentation liquor, the dissolved oxygen value DO of the fermentation liquor and the stirring speed of the fermentation liquor as output factors of the fuzzy reasoning; constructing a fuzzy control system;

step 3, determining a change discourse domain of input and output factors of the fuzzy control system;

step 4, formulating fuzzy division and membership function of input and output factors;

step 5, making a probability coupling rule and a fuzzy control table of input and output parameters;

step 6, decoupling and combining fuzzy reasoning and improved Markov methods;

the specific process of decoupling and combining the fuzzy inference and the improved Markov method comprises the following steps:

the decoupling model of the fuzzy reasoning and improved Markov method is formed as follows:

calculating the weight of a model of each parameter combination for the collected polyphenol content and color information of the fermentation liquor, the fermentation temperature, pH of the fermentation liquor, DO of the dissolved oxygen value of the fermentation liquor and the stirring speed of the fermentation liquor which need to be output to a control mechanism, taking a plurality of models as an observation sequence, and adopting a forward factor alpha_t(i) For initialization of the forward factor, α₁(i)＝π_ib_i(Y₁) Wherein i is more than or equal to 1 and less than or equal to N,y1 is the probability of the initial time in the time sequence; continuously calculating the weight by using a recursive method, and gradually recursing alpha from front to back_t+1(j)＝[∑α_t(i)α_ij]b_j(Y_t+1) T is more than or equal to 1 and less than or equal to T-1, j is more than or equal to 1 and less than or equal to N, alpha_tTo observe the probability of a sequence at time t, b_jProbability of observing a sequence for a given Markov model;

the fermentation control model is regarded as an observation sequence, O ═ O₁O₂,...,O_TThe observation model is lambda (A, B, pi), P (O | lambda) is calculated, and the data of each parameter is used as the matching degree of a given model and the observation sequence;

selecting a certain Markov model lambda_i＝{A_i,B_i,π_iH, i-1, 2, C, wherein a_i，B_i，π_iAre all parameters of the model; observation of the sequence O ═ O for a given fermentation model₁,O₂,···,O_TAnd a model parameter lambda of the hidden Markov model_iI-1, 2, C, wherein O_TThe observed state of the factor O at the time T;

q for a particular state sequence₁,q₂,...,q_TGenerating an observation sequence O ═ O₁O₂,...,O_TThe probability of (c) is:

wherein b is_qTFor the probability of the probabilistic model observing the sequence O at T ═ T, a state sequence Q ═ Q is generated for given model parameters λ₁,q₂,...,q_TThe probability of (c) is: p (Q | λ) ═ pi_q1a_q1q2a_q2q3···a_qT-1qT(4.11)

Wherein alpha is_qT-1qTFor the function parameters, an observation sequence O ═ O is generated for the calculation model₁O₂,...,O_TMust hide each type of the probabilityThe state sequences are all taken into account and are calculated to each produce the observed sequence O ═ O₁O₂,...,O_TThen summed, so the probability is:

from formula (4.12), the observation sequence O ═ O₁O₂,...,O_TIs equal to the sum of the probabilities of all the hidden state sequences that may generate this observation sequence, P (O | λ) is calculated based on the algorithm of the forward method recursive idea, so that the temporal complexity of the algorithm is reduced to N²T and N are dimensions of the observed sequence; solving the decoupling value of each fermentation control parameter in the coupling state;

and 7, taking four output factors in the oxidative fermentation process of the summer and autumn tea as control parameters, inputting the control parameters into an execution mechanism, and respectively adjusting a heating device, an acid-base pump for changing the pH value of the fermentation liquor, an air pump for changing the dissolved oxygen value of the fermentation liquor and a stirring motor for changing the stirring speed of the fermentation liquor in the execution mechanism.

2. The intelligent food liquid fermentation parameter control method according to claim 1, wherein the fuzzy control system is a dual-input, four-output improved fuzzy control system.

3. The method for controlling the intelligent liquid fermentation parameters of food according to claim 1, wherein in step 3, the argument of the variation of the input and output factors is as follows: the change of polyphenol content is 1.4-2.1, the Euclidean distance of color is 0-15.6, and the fermentation temperature is as follows: 25-70 ℃; pH of the fermentation liquor: 6.0-8.0; and (3) dissolved oxygen value DO of fermentation liquor: 0 to 100 parts; the stirring speed of the fermentation liquor is as follows: 20 to 300.

4. The method of claim 1, wherein the membership function is selected as: triangular segmented membership functions.

5. The method as claimed in claim 3, wherein the fuzzy set of polyphenol content is divided into the following rules, respectively

Very low VL e (1.4, 1.5); LOW ∈ (1.5,1.6) U (1.6, 1.7); somewhat lower RL ∈ (1.6,1.7) U (1.7, 1.8); suitable MED e (1.7,1.8) U (1.8, 1.9); a high RH ∈ (1.8,1.9) U (1.9, 2.0); high H ∈ (1.9,2.0) U (2.0, 2.1); very high VH e (2.0, 2.1);

the fuzzy set division rule of the color value of the fermentation liquor is as follows respectively

Suitable MED e (0, 2); suitable RM ∈ (0,2) U (2,4) is compared; slightly higher LH ∈ (2,4) U (4, 6); a high RH ∈ (4,6) and U (6, 8); high H ∈ (6,8) U (8, 10); very high VH ∈ (8,12) < u (12, 15);

the temperature fuzzy set partitioning rule of fermentation is as follows:

low temperature MINT ∈ (25, 34); a lower temperature LT ∈ (25,34) U (34, 43); medium temperature MT ∈ (34,43) — (43, 52); higher temperature PT ∈ (43,52) — (52, 61); high temperature MAXT ∈ (52,61) U (61, 70);

the fuzzy set division rule of the pH of the fermentation liquor is as follows:

acid MINpH epsilon (6.0, 6.5); the acid LpH ∈ (6.0,6.5) U (6.5, 7.0); neutral MpH ∈ (6.5,7.0) U (7.0, 7.5); the dotted base PpH ∈ (7.0,7.5 ∈ U (7.5, 8.0); the base MAXpH is in the range of (7.5, 8.0);

the fuzzy set division rule of the fermentation liquor dissolved oxygen value DO is as follows:

low MINDO ∈ (0, 25); lower LDO ∈ (0,25) U (25, 50); medium MDO ∈ (25,50) U (50, 70); higher PDO ∈ (50,75) U (75,100); high MAXDO ∈ (75,100);

the fuzzy set division rule of the fermentation liquor stirring speed is as follows:

extremely low MINS ∈ (20, 67); low LS ∈ (20,67) U (67,114); lower RS ∈ (67,114) U (114,161); medium speed MS ∈ (114,161) — U (161,208); higher PS ∈ (161,208) — U (208,255); high HS ∈ (208,255) u (255,300); very high MAXS ∈ (255,300).

Images