Tea fermentation process self-adaptive control system and method based on parameter collaborative optimization

Abstract

The application discloses a tea fermentation process self-adaptive control system and method based on parameter collaborative optimization, and particularly relates to the technical field of tea fermentation control, and is used for solving the cross-loop positive feedback oscillation and control instability problems caused by independent controller actions of the existing hierarchical optimization strategy; a control parameter set is established by real-time collection of temperature, humidity and oxygen concentration parameters; the dynamic coupling strength between parameters is identified and key parameter groups are screened; strong / weak positive or negative collaborative factors are generated according to the real-time change direction of the parameters; the controller output weight is dynamically adjusted based on the collaborative factors; the control instability degree is calculated according to the set value tracking error after the weight adjustment, and if the control instability degree continuously rises for three periods, an entropy compensation instruction is generated; finally, the multivariable collaborative control is executed in combination with the weight and the compensation instruction; the dynamic decoupling of the multivariable control and the global stability improvement are realized, and the uniformity of tea fermentation is effectively ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tea fermentation control, and more particularly, to a tea fermentation process adaptive control system and method based on parameter collaborative optimization. BACKGROUND

[0002] In the tea fermentation process control, the coordinated regulation of multiple variables such as temperature, humidity, and oxygen concentration directly affects the fermentation quality. The existing technology adopts a hierarchical optimization strategy: the fermentation environment data is collected through a sensor network, the dynamic model of each variable is independently identified, and the corresponding controller parameters are adjusted based on a single-objective optimization algorithm. This method can maintain basic control under steady-state conditions, but in the fermentation intensive period (such as the polyphenol oxidase activity mutation stage), the independent optimization mechanism of each subsystem cannot coordinate the coupling relationship between temperature rise and oxygen demand.

[0003] The existing technology adopts a hierarchical optimization strategy, which has the following problems: when a variable controller actively adjusts the parameters due to environmental disturbances, its control action will change the dynamic characteristics of other variables, resulting in a misalignment of the identified associated model. Subsequent parameter optimization based on the misaligned model will further amplify the overall control deviation of the system, forming a positive feedback oscillation across control loops, leading to control instability, and affecting the uniformity of fermentation. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the present application provides a tea fermentation process adaptive control system and method based on parameter collaborative optimization to solve the problems raised in the background art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0006] The tea fermentation process adaptive control method based on parameter collaborative optimization comprises the following steps:

[0007] S1, real-time acquisition of temperature control parameters, humidity control parameters and oxygen concentration control parameters of the tea fermentation process, establishment of a control parameter set;

[0008] S2, identification of the dynamic coupling strength between any two control parameters in the control parameter set, and selection of a control parameter pair with a dynamic coupling strength exceeding a preset strength threshold as a key parameter group;

[0009] S3, generation of a control parameter collaborative optimization factor including a strong positive collaborative factor, a weak positive collaborative factor, or a compensatory negative collaborative factor according to the real-time change direction of the control parameters in the key parameter group:

[0010] S4, dynamic adjustment of the controller output weight corresponding to the key parameter group based on the control parameter collaborative optimization factor;

[0011] S5, based on the adjusted controller output weight, calculate the set value tracking error of the multivariable control system, calculate the control instability degree according to the change of the set value tracking error, if the control instability degree rises for three consecutive sampling periods, generate entropy reduction compensation instruction;

[0012] S6, using the adjusted controller output weight and superimposing the entropy reduction compensation instruction to execute the multivariable collaborative control.

[0013] Further, the temperature control parameter, humidity control parameter and oxygen concentration control parameter of the tea fermentation process are acquired in real time, and a control parameter set is established, including:

[0014] The temperature control parameter of the tea fermentation process is acquired by a temperature sensor;

[0015] The humidity control parameter of the tea fermentation process is acquired by a humidity sensor;

[0016] The oxygen concentration control parameter of the tea fermentation process is acquired by an oxygen concentration sensor;

[0017] The temperature control parameter, humidity control parameter and oxygen concentration control parameter acquired at the same sampling time are time-synchronized and aligned;

[0018] The control parameter set is established based on the time-synchronized and aligned temperature control parameter, humidity control parameter and oxygen concentration control parameter.

[0019] Further, the dynamic coupling strength between any two control parameters in the control parameter set is identified, and the control parameter pair with the dynamic coupling strength exceeding a preset strength threshold is selected as a key parameter group, including:

[0020] The numerical sequence of the temperature control parameter, humidity control parameter and oxygen concentration control parameter in the control parameter set within a continuous time window is extracted;

[0021] For any two control parameters, the cross-correlation coefficient of the numerical sequence of the two control parameters within the time window is calculated;

[0022] The absolute value of the cross-correlation coefficient is taken as the dynamic coupling strength;

[0023] The size relationship between the dynamic coupling strength and the preset strength threshold is compared;

[0024] When the dynamic coupling strength is greater than the preset strength threshold, the control parameter pair composed of the corresponding two control parameters is added to the key parameter group.

[0025] Further, the control parameter collaborative optimization factor including strong positive collaborative factor, weak positive collaborative factor or compensation type reverse collaborative factor is generated according to the real-time change direction of the control parameters in the key parameter group, including:

[0026] calculating the direction persistence of each control parameter in a continuous sampling period;

[0027] generating a strong positive synergy factor when the two control parameters change in the same direction and the difference in direction persistence is less than a set threshold value;

[0028] generating a weak positive synergy factor when the two control parameters change in the same direction but the difference in direction persistence exceeds the set threshold value;

[0029] generating a compensatory negative synergy factor according to the fluctuation amplitude matching degree when the two control parameters change in opposite directions.

[0030] Further, generating a compensatory negative synergy factor according to the fluctuation amplitude matching degree includes:

[0031] calculating the absolute values of the change amounts of the two control parameters in the current sampling period and the previous sampling period;

[0032] comparing the ratio of the two absolute values with a set matching threshold value;

[0033] generating a first type of compensatory negative synergy factor when the ratio is within the set matching threshold value range;

[0034] generating a second type of compensatory negative synergy factor when the ratio exceeds the set matching threshold value range.

[0035] Further, dynamically adjusting the controller output weights corresponding to the key parameter group based on the control parameter synergy optimization factor includes:

[0036] for each control parameter pair in the key parameter group:

[0037] when the control parameter synergy optimization factor is a strong positive synergy factor, increasing the controller output weights corresponding to the two control parameters in the control parameter pair;

[0038] when the control parameter synergy optimization factor is a weak positive synergy factor, maintaining the controller output weights corresponding to the two control parameters in the control parameter pair unchanged;

[0039] when the control parameter synergy optimization factor is a compensatory negative synergy factor, reducing the controller output weights corresponding to the two control parameters in the control parameter pair.

[0040] Further, based on the adjusted controller output weights, calculating the set value tracking error of the multivariable control system, and calculating the control instability degree according to the change of the set value tracking error, if the control instability degree rises for three consecutive sampling periods, generating an entropy reduction compensation instruction, including:

[0041] Based on the adjusted controller output weight, the difference between the actual value of the temperature control parameter and the temperature set value is calculated as the temperature set value tracking error, the difference between the actual value of the humidity control parameter and the humidity set value is calculated as the humidity set value tracking error, and the difference between the actual value of the oxygen concentration control parameter and the oxygen concentration set value is calculated as the oxygen concentration set value tracking error;

[0042] The absolute value of the change amount of the temperature set value tracking error in the current sampling period and the temperature set value tracking error in the previous sampling period is calculated, the absolute value of the change amount of the humidity set value tracking error in the current sampling period and the humidity set value tracking error in the previous sampling period is calculated, and the absolute value of the change amount of the oxygen concentration set value tracking error in the current sampling period and the oxygen concentration set value tracking error in the previous sampling period is calculated.

[0043] The absolute value of the change amount of the temperature set value tracking error, the absolute value of the change amount of the humidity set value tracking error, and the absolute value of the change amount of the oxygen concentration set value tracking error are added to obtain the control instability degree in the current sampling period.

[0044] It is judged whether the control instability degree in the current sampling period is greater than the control instability degree in the previous sampling period, and whether the control instability degree in the previous sampling period is greater than the control instability degree in the sampling period before the previous sampling period.

[0045] When the judgment condition is established, an entropy reduction compensation instruction is generated.

[0046] Further, the multivariable collaborative control is executed using the adjusted controller output weight and superimposing the entropy reduction compensation instruction, including:

[0047] The entropy reduction compensation instruction is distributed to the temperature control loop, the humidity control loop and the oxygen concentration control loop according to a preset proportion to obtain a compensation component of the temperature control loop, a compensation component of the humidity control loop and a compensation component of the oxygen concentration control loop.

[0048] According to the compensation component of the temperature control loop, the compensation component of the humidity control loop and the compensation component of the oxygen concentration control loop, the final control instruction of the temperature control loop, the final control instruction of the humidity control loop and the final control instruction of the oxygen concentration control loop are obtained.

[0049] According to the final control instruction of the temperature control loop, the temperature adjusting actuator is driven, according to the final control instruction of the humidity control loop, the humidity adjusting actuator is driven, and according to the final control instruction of the oxygen concentration control loop, the oxygen concentration adjusting actuator is driven.

[0050] Further, the output weight of the temperature control loop in the adjusted controller output weight is added to the compensation component of the temperature control loop to obtain the final control instruction of the temperature control loop.

[0051] The output weight of the humidity control loop in the adjusted controller output weight is added to the compensation component of the humidity control loop to obtain a final control instruction of the humidity control loop.

[0052] The output weight of the oxygen concentration control loop in the adjusted controller output weight is added to the compensation component of the oxygen concentration control loop to obtain a final control instruction of the oxygen concentration control loop.

[0053] In another aspect, the present application provides a tea fermentation process adaptive control system based on parameter collaborative optimization, comprising the following modules:

[0054] A parameter acquisition module is configured to acquire temperature control parameters, humidity control parameters and oxygen concentration control parameters of the tea fermentation process in real time, and establish a control parameter set.

[0055] A coupling screening module is configured to identify the dynamic coupling strength between any two control parameters in the control parameter set, and screen out a control parameter pair with a dynamic coupling strength exceeding a preset strength threshold as a key parameter group.

[0056] A factor generation module is configured to generate a control parameter collaborative optimization factor including a strong positive collaborative factor, a weak positive collaborative factor or a compensation type reverse collaborative factor according to the real-time change direction of the control parameters in the key parameter group.

[0057] A weight adjustment module is configured to dynamically adjust the controller output weight corresponding to the key parameter group based on the control parameter collaborative optimization factor.

[0058] An entropy reduction instruction module is configured to calculate a set value tracking error of the multivariable control system based on the adjusted controller output weight, and calculate a control instability degree according to the change of the set value tracking error, and generate an entropy reduction compensation instruction if the control instability degree rises for three consecutive sampling periods.

[0059] A collaborative execution module is configured to execute the multivariable collaborative control using the adjusted controller output weight and superimposing the entropy reduction compensation instruction.

[0060] Compared with the prior art, the present application has the following beneficial effects:

[0061] 1. Through the dynamic coupling strength identification and the synergistic optimization factor generation mechanism, the positive feedback oscillation problem between the multivariable control loops is solved, which is different from the independent identification mechanism of the existing hierarchical optimization strategy, the key parameter group is dynamically screened based on the real-time parameter interaction relationship, the coupling direction and strength of temperature-humidity-oxygen concentration are accurately quantified through the synergistic optimization factor, and the controller output weight is dynamically adjusted accordingly, the dynamic correlation between temperature rise and oxygen demand increase can be actively coordinated during the fermentation period, the model misalignment problem caused by single controller action is avoided, the cross-loop positive feedback chain is cut off from the source, and the control stability is significantly improved.

[0062] 2. By introducing the control instability degree monitoring and entropy reduction compensation mechanism, the system can realize real-time sensing of the overall instability trend of the multivariable control, when the set value tracking error continuously rises, a compensation instruction is automatically generated and superimposed to the weight adjustment result, active intervention is implemented at the initial stage of control entropy increase, the limitation of passive response to local disturbance in the prior art is broken through, the global stability of the multivariable system under dynamic working conditions is ensured while maintaining the uniformity of fermentation, and it is especially suitable for the tea fermentation scene with high coupling of temperature, humidity and oxygen demand. BRIEF DESCRIPTION OF DRAWINGS

[0063] Figure 1 The flowchart of the tea fermentation process adaptive control method based on parameter synergistic optimization of the present application;

[0064] Figure 2 The structural schematic diagram of the tea fermentation process adaptive control system based on parameter synergistic optimization of the present application. DETAILED DESCRIPTION

[0065] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0066] Embodiment 1: Figure 1 The tea fermentation process adaptive control method based on parameter synergistic optimization of the present application is given, which comprises the following steps:

[0067] S1, real-time acquisition of temperature control parameters, humidity control parameters and oxygen concentration control parameters of the tea fermentation process, establishment of a control parameter set;

[0068] S2, identifying the dynamic coupling strength between any two control parameters in the control parameter set, and screening out the control parameter pair with the dynamic coupling strength exceeding the preset strength threshold as the key parameter group;

[0069] S3, generating a control parameter synergy optimization factor including a strong positive synergy factor, a weak positive synergy factor or a compensatory negative synergy factor according to the real-time change direction of the control parameter in the key parameter group:

[0070] S4, dynamically adjusting the controller output weight corresponding to the key parameter group based on the control parameter synergy optimization factor;

[0071] S5, calculating the set value tracking error of the multivariable control system based on the adjusted controller output weight, calculating the control instability degree according to the change of the set value tracking error, and generating an entropy reduction compensation instruction if the control instability degree rises continuously for three sampling periods;

[0072] S6, performing multivariable synergy control using the adjusted controller output weight and superimposing the entropy reduction compensation instruction.

[0073] The various threshold values, reference values, proportional coefficients and range parameters involved in this embodiment, such as the intensity threshold value, the direction duration difference threshold value, the fluctuation amplitude matching degree threshold value, the weight adjustment step, the weight safety range, etc., are all initial values set based on the analysis of historical tea fermentation process data, through simulation experiment calibration or according to the general experience in the field of tea fermentation control. These parameters are not fixed and can be adjusted adaptively through limited experiments or simulations by those skilled in the art according to the specific application of tea varieties, fermentation equipment characteristics and the required control accuracy, which belongs to the conventional technical means in the field. In the following, the numerical values of some parameters will be given in the form of examples to clearly illustrate the implementation process of the present application.

[0074] S1, real-time acquisition of temperature control parameters, humidity control parameters and oxygen concentration control parameters of the tea fermentation process, establishment of a control parameter set, including:

[0075] In the control of tea fermentation process, the parameter acquisition and synchronization step is first executed. The platinum resistance temperature sensor is used to acquire the fermentation environment temperature control parameter, which is installed at the geometric center of the fermentation box, with a measurement range of 20℃ to 80℃ and a measurement accuracy of ±0.5℃. The sensor works with a fixed sampling interval, for example, data is collected every 500 milliseconds, and the collected values are temporarily stored in the data buffer in floating point form, with the temperature unit being Celsius. The capacitance type humidity sensor is used to acquire the environment humidity control parameter, which is distributedly installed at four vertical height points on the inner wall of the fermentation box, with a measurement range of 30%RH to 95%RH and an accuracy of ±2%RH. The sampling time is triggered synchronously by the hardware clock with the temperature sensor. The electrochemical oxygen sensor is used to acquire the oxygen concentration control parameter, which is placed at the outlet of the gas circulation pipeline, with a measurement range of 1%vol to 25%vol and an accuracy of ±0.5%vol. The sampling clock line of the sensor is connected to the same time sequence controller as the aforementioned sensors.

[0076] The three types of parameters obtained at the same sampling time are time-synchronized and aligned, and the specific implementation process includes: the central processor generates a reference timestamp for each sampling period, with a precision of 1 millisecond; when receiving a sensor data packet, the acquisition time field recorded in the data packet is parsed; a time window matching algorithm is used to group parameters with a time deviation within the allowed tolerance range, for example, the maximum allowed time difference is set to 10 milliseconds; sliding window compensation is implemented for overtime data, with a window width of 2 times the sampling period (i.e. 1000 milliseconds), and the nearest adjacent sampling point data is selected within the window for interpolation replacement; finally, a three-dimensional parameter vector with strictly time-aligned parameters is output, and the vector elements are arranged in the order of [temperature value, humidity value, oxygen concentration value].

[0077] Based on the time-synchronized and aligned parameters, a control parameter set is established, and the specific implementation method is: a circular storage matrix is created, the row index of the matrix corresponds to the sampling time sequence, and the column index is fixed as 3 columns to store temperature control parameters, humidity control parameters and oxygen concentration control parameters. Each matrix unit contains three data fields: parameter value (floating point type), parameter type identifier (integer encoding: 1=temperature, 2=humidity, 3=oxygen), and timestamp (millisecond Unix time). The control parameter set adopts a ring buffer structure, and the buffer depth is set to 14400 rows (corresponding to 120 minutes of data, calculated at a 500 millisecond sampling period), when the number of rows exceeds the limit due to new data, the historical data row at the earliest time point is automatically overwritten. The set provides an access interface through a global variable table, and subsequent processing extracts a sub-data set by specifying a time range (e.g. the last 300 seconds).

[0078] For boundary processing of sensor failure: when a sensor continuously samples data for 3 times beyond the range (e.g. temperature value less than 20℃ or greater than 80℃), the adjacent sensor data replacement mechanism is automatically enabled. If the humidity sensor fails, the arithmetic mean of the data from the remaining three installation points is used to replace it; if the oxygen sensor fails, a compensation model based on gas flow rate is used to calculate an approximate value, and the input parameters of the model are the fermentation tank inlet flow rate (obtained through an electromagnetic flowmeter) and the historical oxygen concentration change rate.

[0079] The sliding window compensation implementation of the time synchronization alignment link is as follows: assuming that the current sampling time is T, the available data points are searched in the window of [T-1000 ms, T+1000 ms]. If the temperature parameter is recorded at T+15 ms and the humidity parameter is recorded at T-8 ms, the time difference of 23 ms is less than the maximum allowed time difference (for example, 30 ms), and the two data points are directly used; if there is no data of the oxygen parameter in the window, the first two valid sampling points (T-500 ms and T+500 ms) are taken to calculate the estimated value of T by linear interpolation. During the control parameter set establishment process, the data type conversion rule is that all parameter values are uniformly converted into single-precision floating-point numbers for storage, the temperature value is kept to one decimal place, and the humidity and oxygen concentration are kept to integer places.

[0080] S2, identify the dynamic coupling strength between any two control parameters in the control parameter set, and select the control parameter pair with the dynamic coupling strength exceeding a preset strength threshold as a key parameter group, including the following specific implementations:

[0081] The numerical sequences of the temperature control parameter, the humidity control parameter and the oxygen concentration control parameter in the continuous time window are extracted from the control parameter set. The specific implementation process is as follows: first, the length of the time window is determined, for example, set to 300 seconds, which is determined according to the polyphenol oxidase reaction period of the tea fermentation process; all data rows within 300 seconds before the current time are intercepted in time sequence from the circular storage matrix of the control parameter set; the data of the temperature control parameter column, the humidity control parameter column and the oxygen concentration control parameter column are extracted respectively to form three independent time sequence arrays; each array element contains a parameter value and a time stamp accurate to milliseconds; each parameter sequence is subjected to data validity check, and the abnormal values (such as temperature values less than 20℃ or greater than 80℃) exceeding the sensor range are removed, and the abnormal values are replaced by the moving average of the first three valid sampling points.

[0082] For any two control parameters, the mutual information value of the two control parameter numerical sequences in the time window is calculated as the dynamic coupling strength. The specific calculation steps include: selecting two parameter types to be analyzed, for example, the temperature control parameter and the humidity control parameter; strictly aligning the numerical sequences of the two parameters according to the time stamp to ensure that each time point has a corresponding parameter value pair; respectively discretizing the numerical sequences of the two parameters according to their historical value range to divide them into several equal-width intervals; the probability distribution of each parameter value falling in each interval and the joint probability distribution of the two parameter value pairs falling in the two-dimensional joint interval are calculated; the mutual information value between the two parameter sequences is calculated according to the respective probability distribution and the joint probability distribution. This calculation process is performed for all possible parameter combinations, including temperature-humidity, temperature-oxygen concentration and humidity-oxygen concentration.

[0083] The calculated mutual information value is the dynamic coupling strength. The implementation is to store the mutual information calculation result as floating-point data, with precision reserved to three decimal places; the greater the dynamic coupling strength value, the stronger the statistical dependence between the two parameters, which can reflect the linear and nonlinear coupling relationship. This value is used to quantify the degree of mutual influence of environmental parameters such as temperature, humidity, and oxygen concentration during fermentation.

[0084] The size relationship between the dynamic coupling strength and the preset strength threshold is compared. The preset strength threshold is determined by analyzing historical fermentation data. The specific method is: collect parameter data of at least 50 batches of standard fermentation processes; calculate the average dynamic coupling strength of three groups of parameter pairs in each batch; take the 85th percentile of all batch strength values as the threshold reference value, for example, 0.65; set an adjustable range of ±0.05 to adapt to different tea varieties. The implementation process of the comparison operation is: read the preset strength threshold from the storage; compare the dynamically calculated coupling strength with the threshold value; output the comparison result Boolean value (true or false).

[0085] When the dynamic coupling strength is greater than the preset strength threshold, the control parameter pair composed of the corresponding two control parameters is added to the key parameter group. The implementation process includes: creating a data structure of the key parameter group, using a linked list to store the control parameter pair; each node stores the type identifier of the two parameters (such as "temperature-oxygen") and the corresponding dynamic coupling strength value; when the comparison result is true, a new node data is generated; insert the node into the end of the linked list; check whether the same parameter pair exists in the linked list, if it exists, update its coupling strength value. The maximum capacity of the key parameter group is set to 6 parameter pairs, and when the capacity is exceeded, the parameter pair with the smallest coupling strength is automatically eliminated.

[0086] For the continuous updating mechanism of the time window: every time a sampling period (e.g. 500 milliseconds) is completed, the time window slides forward by one sampling point; remove the data of the oldest time point in the window and add the latest collected data; recalculate the dynamic coupling strength of all parameter pairs. Boundary condition processing: when the number of valid data points in the time window is less than 30% of the total number (e.g. due to sensor failure), suspend the update of the key parameter group, maintain the last valid state and issue a warning signal. The history record of the dynamic coupling strength is saved to a separate database for subsequent fermentation quality traceability analysis.

[0087] S3, generating a control parameter synergy optimization factor including a strong positive synergy factor, a weak positive synergy factor, or a compensatory negative synergy factor according to the real-time change direction of the control parameters in the key parameter group, the specific implementation includes:

[0088] The direction persistence of each control parameter in the critical parameter group is calculated in the continuous sampling period. The specific implementation process is: the values of the current control parameter in the latest two sampling periods are extracted from the control parameter set; the difference between the value in the current sampling period and the value in the previous sampling period is calculated; when the difference is greater than zero, it is determined to be an upward direction, when the difference is less than zero, it is determined to be a downward direction, and when the difference is equal to zero, the original direction record is maintained; the direction persistence counter is set, and the initial value is zero; when the same direction change is detected continuously, the counter is increased by one each time; when the direction changes, the counter is reset to one; the final output of the direction persistence is an integer count value. For example, if the temperature control parameter rises continuously for three periods, its direction persistence is three. This calculation is performed at each sampling period update to ensure real-time performance.

[0089] When the change directions of two control parameters are the same and the difference in direction persistence is less than the set threshold, a strong positive synergistic factor is generated. The set threshold is determined by historical data analysis, and the specific method is: collecting the persistence data of parameters that change in the same direction during normal fermentation; calculating the persistence ratio distribution of all parameters that change in the same direction; taking the 25th percentile of the ratio distribution as the reference value, for example, 0.75; setting the threshold interval as the reference value ± 0.05. When implementing the comparison, the ratio of the direction persistence of the two parameters is calculated; when the ratio is greater than 0.7 and less than 1.3 (for example, parameter A persistence = 4, parameter B persistence = 5, ratio = 0.8), it is determined that the difference is less than the set threshold. The strong positive synergistic factor is stored in a specific code, for example, binary code 01.

[0090] When the change directions of two control parameters are the same but the difference in direction persistence exceeds the set threshold, a weak positive synergistic factor is generated. The implementation process is: when both parameters are in the upward or downward direction, but the ratio of the direction persistence is less than 0.7 or greater than 1.3 (for example, parameter A persistence = 2, parameter B persistence = 8, ratio = 0.25), it is determined that the difference exceeds the set threshold. The weak positive synergistic factor is stored in a specific code, for example, binary code 10. After the factor is generated, the corresponding control parameter pair identifier is associated and stored in the synergistic factor register.

[0091] When the change directions of two control parameters are opposite, a compensatory reverse synergistic factor is generated according to the fluctuation amplitude matching degree. The opposite direction determination standard is: one parameter rises and the other falls, or one parameter falls and the other rises. The fluctuation amplitude matching degree calculation includes: obtaining the values of the two control parameters in the current sampling period; obtaining the values of the two control parameters in the previous sampling period; calculating the absolute value of the difference between the current control parameter value and the value in the previous sampling period to obtain the change absolute value; calculating the ratio of the two change absolute values, taking the smaller value divided by the larger value; the ratio is the fluctuation amplitude matching degree, and its value is between 0 and 1.

[0092] The calculated fluctuation amplitude matching degree is compared with a set matching threshold. The set matching threshold is determined through experimental data analysis, specifically: analyzing the fluctuation data of 100 groups of reverse change parameter pairs; calculating the matching degree range corresponding to the best compensation effect; determining the threshold interval as 0.8 to 1.2. The comparison operation is implemented as: when the fluctuation amplitude matching degree is in the range of 0.8 to 1.0, a first type of processing path is executed; when the fluctuation amplitude matching degree is less than 0.8, a second type of processing path is executed.

[0093] When the ratio is within the set matching threshold range, a first type of compensation type reverse synergy factor is generated. The specific implementation is: when the fluctuation amplitude matching degree is between 0.8 and 1.2 (for example, the absolute value of the temperature change amount = 0.5℃, the absolute value of the humidity change amount = 0.6℃, and the ratio = 0.83), a first type of compensation type reverse synergy factor coded as binary 110 is generated. This factor indicates that the fluctuation amplitudes of the two parameters are similar and require moderate intensity compensation.

[0094] When the ratio exceeds the set matching threshold range, a second type of compensation type reverse synergy factor is generated. The specific implementation is: when the fluctuation amplitude matching degree is less than 0.8 (for example, the absolute value of the temperature change amount = 0.3℃, the absolute value of the humidity change amount = 1.2℃, and the ratio = 0.25), a second type of compensation type reverse synergy factor coded as binary 111 is generated. This factor indicates that the fluctuation amplitudes of the two parameters are significantly different and require high intensity compensation.

[0095] Synergy factor storage management is implemented as: creating a synergy factor mapping table, table entries include parameter pair identifier (such as "temperature-humidity"), synergy factor type, and generation timestamp; recalculate and overwrite old values every sampling period update; set an expiration flag, automatically marked as invalid when not updated for more than three sampling periods. Boundary condition processing: when a control parameter does not change for two consecutive periods, the direction persistence is maintained but marked as "stable state", at which time it does not participate in synergy factor generation and maintains the last valid factor value.

[0096] Abnormal processing of direction persistence calculation: when sensor failure causes data loss, the direction trend of the last three valid sampling points is used for interpolation estimation; if there is no valid data for more than five periods, the synergy factor generation for this parameter pair is suspended.

[0097] S4, based on the control parameter synergy optimization factor, dynamically adjusting the controller output weight corresponding to the key parameter group, the specific implementation includes:

[0098] For each control parameter pair in the key parameter group, a dynamic weight adjustment operation is performed. First, the current factor type is read from the control parameter co-optimization factor register. This register stores data in binary encoded fields, containing the parameter pair identifier and factor code. The controller output weight baseline value is initialized. This baseline value is determined through PID parameter tuning at system startup; for example, the initial weights for the temperature control loop are 0.35, humidity is 0.30, and oxygen concentration is 0.35. The weight values ​​are stored in floating-point variables with three decimal places.

[0099] When the co-optimization factor of the control parameters is a strong positive co-factor, a weight increase operation is performed. The specific process is as follows: The current weight values ​​of the two parameters in the control parameter pair are read from the weight storage area; the weight increment is calculated, with the increment size determined according to the factor strength level, for example, a base increment of 0.05 is used for a strong positive co-factor; the increment is applied to calculate the new weight value, which is equal to the original weight value plus the increment; upper and lower limits are imposed on the calculation results to ensure the weight value remains within a safe range of 0.1 to 0.8; the updated weight value is written back to the storage area. For example, when a strong positive co-factor appears in the temperature-oxygen parameter pair, the temperature weight increases from 0.35 to 0.40, and the oxygen weight increases from 0.35 to 0.40.

[0100] When the control parameter co-optimization factor is a weakly positive co-optimization factor, a weight maintenance operation is performed. The specific process is as follows: read the current weight values ​​of the two parameters; check if the weight values ​​are within a valid range; when the weight values ​​are valid, keep the stored values ​​unchanged; when the weight values ​​are abnormal (e.g., less than 0 or greater than 1), reset them to the initial baseline values. This operation does not change any stored data; it only sets the status flag "Weight Maintenance Activated".

[0101] When the control parameter co-optimization factor is a compensatory reverse co-optimization factor, a differentiated weight reduction operation is implemented according to its subtype. Specifically, for the first type of compensatory reverse co-optimization factor (fluctuation amplitude matching degree is in the range of 0.8 to 1.0), a linear reduction mode is adopted; for the second type of compensatory reverse co-optimization factor (fluctuation amplitude matching degree is less than 0.8), a step-wise reduction mode is adopted.

[0102] Weight adjustment boundary protection mechanism: Four checks are performed after each weight update. The checks include: whether the total weight is within the range of 0.95 to 1.05; if it exceeds this range, all loop weights are scaled proportionally; whether the rate of change of a single parameter weight exceeds the threshold of 0.15 per cycle; whether the weight value is a valid floating-point number; and whether the update operation is completed within the allowed time window (e.g., 50 milliseconds). If a check fails, a weight rollback procedure is initiated: the weight value from the previous sampling cycle is restored; an error log is recorded; and a system alarm signal is sent.

[0103] Data structure of weight storage area: create a weight matrix table, the row index corresponds to the control parameter type (1 = temperature, 2 = humidity, 3 = oxygen), and the column index corresponds to the time sequence; each cell stores the weight value, update timestamp, version number; adopt double buffering mechanism to ensure read-write consistency, active buffer area for real-time control, shadow buffer area for update operation. After each weight adjustment, automatically generate adjustment record, including parameter pair identification, original weight value, new weight value, adjustment reason (factor type), timestamp and other fields, for quality control traceability.

[0104] Timing control of weight adjustment: complete factor identification of all parameter pairs within 20 milliseconds after the start of the sampling period; complete weight calculation and update within the next 30 milliseconds; the last 10 milliseconds are used for verification and exception handling. When the key parameter group contains multiple parameter pairs, use parallel processing mechanism, and the weight adjustment of each parameter pair is executed independently.

[0105] S5, based on the adjusted controller output weight, calculate the set value tracking error of the multivariable control system, and calculate the control instability degree according to the change of the set value tracking error, if the control instability degree rises for three consecutive sampling periods, generate entropy reduction compensation instruction, the specific implementation includes:

[0106] Based on the adjusted controller output weight, calculate the difference between the actual value of the temperature control parameter and the temperature set value as the temperature set value tracking error. The specific implementation process is: read the actual value of the temperature control parameter in the current sampling period from the control parameter set; obtain the temperature set value at the current time from the process parameter database, which is dynamically generated according to the characteristics of the tea fermentation stage, for example, set to 28℃ in the early fermentation period and rise to 35℃ in the middle period; calculate the original algebraic difference between the actual value and the set value, the original difference = actual value - set value; use the adjusted controller output weight to correct the original difference to obtain the final set value tracking error used for calculation, for example, when the temperature weight is 0.4, the temperature set value tracking error = original difference x 0.4.

[0107] Based on the adjusted controller output weight, calculate the difference between the actual value of the humidity control parameter and the humidity set value as the humidity set value tracking error. The implementation is: read the current humidity control parameter actual value; query the humidity set value curve, which is generated according to the tea moisture content model; calculate the difference between the actual value and the set value; apply the controller output weight of the humidity loop for correction, for example, when the weight is 0.3, the corrected error = original difference x 0.3; the unit of humidity set value tracking error is percentage relative humidity, and the storage precision is 0.1%.

[0108] Based on the adjusted controller output weight, the difference between the actual value of the oxygen concentration control parameter and the oxygen concentration set value is calculated as the oxygen concentration set value tracking error. The implementation process is: obtaining the current oxygen concentration actual measurement value; retrieving the oxygen concentration set value, which is determined according to the fermentation oxygen demand model; calculating the difference between the actual value and the set value; multiplying the controller output weight of the oxygen concentration loop, for example, when the weight is 0.35, the final error = original difference × 0.35; the unit is volume percentage, and the precision is 0.01%.

[0109] The absolute value of the difference between the temperature set value tracking error of the current sampling period and the temperature set value tracking error of the previous sampling period is calculated. The specific operation is: extracting the temperature set value tracking errors of the current period and the previous period from the error history buffer; calculating the absolute value of the algebraic difference; the change absolute value = |current period error - previous period error|; the result is stored as an unsigned floating point number. For example, the current error is 0.5°C, the previous error is 0.3°C, and the change absolute value is 0.2°C.

[0110] The change absolute value of the humidity set value tracking error of the current sampling period and the humidity set value tracking error of the previous sampling period is calculated. The implementation is: reading the humidity set value tracking errors of the current and previous periods; calculating the absolute difference; for example, the current error is 1.2%RH, the previous error is 0.9%RH, and the change absolute value is 0.3%RH. This calculation is performed immediately after each sampling update.

[0111] The change absolute value of the oxygen concentration set value tracking error of the current sampling period and the oxygen concentration set value tracking error of the previous sampling period is calculated. Specifically: obtaining the oxygen error values of the current and previous periods; calculating the absolute difference; for example, the current error is 0.15%vol, the previous error is 0.10%vol, and the change absolute value is 0.05%vol. All change absolute value calculations use the same time base.

[0112] The change absolute value of the temperature set value tracking error, the change absolute value of the humidity set value tracking error, and the change absolute value of the oxygen concentration set value tracking error are added to obtain the control instability degree of the current sampling period. The implementation process is: creating an accumulator variable initialized to zero; reading the three change absolute values in turn; performing floating point addition operation; control instability = temperature set value tracking error change absolute value + humidity set value tracking error change absolute value + oxygen concentration set value tracking error change absolute value. This value reflects the degree of deviation of the system from the steady state as a whole.

[0113] Determine whether the control instability of the current sampling period is greater than that of the previous sampling period, and whether the control instability of the previous sampling period is greater than that of the previous sampling period. The implementation logic is: extract the last three values from the control instability history queue, denoted as E(t), E(t-1), and E(t-2); perform two comparisons: E(t)>E(t-1) and E(t-1)>E(t-2); the comparison result is stored as a Boolean value. The history queue uses a first-in-first-out structure, and the depth is maintained for more than three periods.

[0114] When the judgment condition is met, generate an entropy reduction compensation instruction. The specific generation process is: when the continuous rising condition is met, create an instruction data structure; the instruction strength is calculated according to the rising amplitude, for example, strength = (E(t)-E(t-2)) / E(t-2); set the instruction action time, the default is three sampling periods; the instruction format uses a floating-point number vector [temperature compensation, humidity compensation, oxygen compensation], and the initial value is set to a zero vector; according to the current error distribution, allocate the compensation amount, for example, when the temperature error occupies the largest proportion, the temperature compensation = strength x 0.6.

[0115] Boundary processing of control instability calculation: when the historical data is insufficient (such as the system startup stage), automatically skip the judgment step; when the absolute value of a certain change is abnormally large (for example, more than 3 times the set threshold), start the data review process: recheck the sensor data, and after confirming that there is no error, it is included in the calculation. After the entropy reduction compensation instruction is generated, record the generation reason data packet, including the control instability value of the last three periods, error distribution analysis, timestamp and other information.

[0116] Timing control mechanism: all error calculations are completed within 40 milliseconds after the start of the sampling period; the change calculation is completed within 20 milliseconds; the control instability summation is completed within 10 milliseconds; the continuous rising judgment is completed within 15 milliseconds; the last 15 milliseconds are used for instruction generation and transmission. The entire processing flow is completed within a 100-millisecond window, ensuring real-time performance.

[0117] S6, use the adjusted controller output weight and superimpose the entropy reduction compensation instruction to perform multivariable collaborative control, including:

[0118] The entropy reduction compensation instruction is distributed to the temperature control loop, the humidity control loop and the oxygen concentration control loop according to a preset ratio to obtain a compensation component of the temperature control loop, a compensation component of the humidity control loop and a compensation component of the oxygen concentration control loop. The specific implementation process is as follows: an entropy reduction compensation instruction vector is read from an instruction buffer, the vector containing a total compensation intensity value and an action duration parameter; the preset ratio is determined through fermentation process characteristic analysis, for example, temperature: humidity: oxygen concentration = 4:3:3; the compensation component of each loop = total compensation intensity x preset ratio coefficient; the compensation component is stored as a floating point number with a precision of 0.001. For example, when the total compensation intensity is 0.8, the temperature compensation component = 0.8 x 0.4 = 0.32, the humidity compensation component = 0.8 x 0.3 = 0.24, and the oxygen compensation component = 0.8 x 0.3 = 0.24. The preset ratio is dynamically optimized according to historical control effects, and the temperature ratio coefficient is automatically increased when temperature deviation frequently occurs.

[0119] The output weight of the temperature control loop in the adjusted controller output weight is added to the compensation component of the temperature control loop to obtain the final control instruction of the temperature control loop. The specific operation is as follows: the output weight value of the current temperature control loop is read from a weight storage area; the compensation component of the temperature control loop is read from a compensation component register; a floating point addition operation is performed, and the final control instruction = output weight + compensation component; the calculation result is range-constrained to ensure that it is between 0 and 1.0. For example, when the temperature output weight is 0.40 and the compensation component is 0.32, the final control instruction = 0.72. The instruction value is converted into a percentage signal for driving an actuator.

[0120] The output weight of the humidity control loop in the adjusted controller output weight is added to the compensation component of the humidity control loop to obtain the final control instruction of the humidity control loop. The implementation is as follows: the current output weight of the humidity control loop is read; the compensation component of the humidity control loop is obtained; an addition operation is performed; for example, the humidity output weight 0.30 plus the compensation component 0.24 gives 0.54; the result is stored in an instruction output queue. The calculation process is completed within 10 milliseconds to ensure real-time performance.

[0121] The output weight of the oxygen concentration control loop in the adjusted controller output weight is added to the compensation component of the oxygen concentration control loop to obtain the final control instruction of the oxygen concentration control loop. Specifically, the output weight of the oxygen concentration control loop is read; the compensation component of the oxygen concentration control loop is obtained; and the final control instruction is obtained by addition; for example, the weight 0.35 plus the compensation 0.24 gives 0.59; the generation of the final control instructions of all loops adopts a parallel processing mechanism.

[0122] The temperature control loop final control instruction drives the temperature adjustment actuator. The implementation process is: convert the final control instruction value to pulse width modulation signal duty cycle, the conversion ratio is 1% duty cycle per 0.01 instruction value; for example, the instruction value 0.72 is converted to 72% duty cycle signal; the signal is transmitted to the resistance heater through the drive circuit; the heater power is linearly adjusted with the duty cycle, and the duty cycle 0% corresponds to the off state, and 100% corresponds to full power heating. The actuator state feedback signal is collected in real time and used for closed loop verification.

[0123] The humidity control loop final control instruction drives the humidity adjustment actuator. The specific operation is: map the humidity final control instruction to the ultrasonic humidifier operating frequency; establish an instruction value-frequency table, for example, the instruction value 0.54 corresponds to a 34 kHz operating frequency; output the control signal through the frequency generator; drive the humidifier electromagnetic transducer to produce ultrasonic atomization of the corresponding frequency. The humidity adjustment amount is proportional to the actual atomized water amount.

[0124] The oxygen concentration control loop final control instruction drives the oxygen concentration adjustment actuator. The implementation is: convert the oxygen final control instruction to the electromagnetic valve opening degree instruction; opening degree = instruction value x 100%, for example, the instruction value 0.59 corresponds to a 59% valve opening; drive the proportional electromagnetic valve through the current signal; the valve opening and the oxygen supply flow are linearly related. The actuator is equipped with a position sensor to provide real-time feedback of the actual opening value.

[0125] Timing synchronization of control instruction execution: start instruction transmission 5 milliseconds before the end of the sampling period; the driving signals of the three types of actuators are issued simultaneously; set a 50 millisecond execution state monitoring window. Abnormal driving signal processing: when the actuator feedback signal deviates from the instruction value by more than 5%, start the three-level response: primary - increase the driving current; intermediate - switch to the backup driving channel; advanced - trigger system protection shutdown. All driving operations are recorded in detail logs, including instruction value, actual execution value, timestamp, device status code, etc.

[0126] Decay mechanism of compensation component: during the action of entropy reduction compensation instruction (for example, three sampling periods), the compensation component is reduced by a preset decay rate every period. The decay rate is set according to the instruction intensity, for example, when the initial intensity is 0.8, the decay rate is 0.2 / period. When the compensation component decays to 0.05 or less, it is automatically cleared. This mechanism ensures that the compensation effect exits smoothly and avoids control mutations.

[0127] Actuator drive safety protection: Maximum power limit for temperature actuators (e.g., not exceeding 90% of rated power); minimum intermittent time for humidity actuators (e.g., at least 100 milliseconds between opening and closing); minimum opening degree guarantee for oxygen valves (e.g., not less than 15%). All drive signals undergo hardware filtering to eliminate high-frequency interference. Actuator status is uploaded to the monitoring system in real time via industrial bus.

[0128] To better illustrate the implementation process, the following example is given, using vacuum anaerobic batch fermentation as an example:

[0129] During the closed-loop vacuuming stage of the fermenter (target vacuum level -95 kPa), temperature control parameters are acquired in real time using a pressure-resistant platinum resistance temperature sensor (range -20~80℃), humidity control parameters are acquired using a sealed humidity sensor (accuracy ±1.5%RH), and vacuum control parameters are monitored using an absolute pressure transmitter (range -101~0 kPa). An anti-disturbance mode is activated during time synchronization: when the vacuum pump starts or stops, causing a pressure surge exceeding 10 kPa / s, the sampling window is automatically extended to 300 milliseconds, and cubic spline interpolation compensation is used. A four-dimensional control parameter set (temperature, humidity, vacuum level, timestamp) is established, with a storage depth of 50 sampling periods.

[0130] Identifying the dynamic coupling strength of the temperature-vacuum parameter pair: Extracting a continuous 150-second data window (300 sampling points), automatically shielding the vacuum pump operating period (data points with pressure change rate > 5 kPa / s) when calculating the cross-correlation coefficient, the measured dynamic coupling strength was 0.78 (> threshold 0.7), and it was included in the key parameter group. When the temperature rise (+0.3℃ / cycle) and the vacuum decrease (pressure relief leading to +2 kPa / cycle) are detected to be in opposite directions, the fluctuation amplitude matching degree is calculated as min(|0.3|,|2|) / max(|0.3|,|2|) = 0.15 (< 0.85), generating a second type of compensating reverse cooperative factor.

[0131] The weights are reduced based on the synergy factor: the weight of the temperature loop is reduced from 0.40 to 0.33, and the weight of the vacuum loop is reduced from 0.45 to 0.38. The setpoint tracking error is calculated as follows: actual temperature 35.2℃ (setpoint 35.0℃) → error +0.2℃, actual vacuum -92.4kPa (setpoint -95.0kPa) → error +2.6kPa. The absolute values ​​of the error changes between adjacent cycles are: temperature |0.2-0.1|=0.1, vacuum |2.6-1.8|=0.8, control instability = 0.1+0.8=0.9 (increasing by 0.5→0.7→0.9 over three consecutive cycles), generating a total entropy reduction compensation command strength of 1.2.

[0132] The compensation instructions are allocated by 4:3:3 (temperature: humidity: vacuum degree): temperature compensation component = 1.2 x 0.4 = 0.48, vacuum degree compensation component = 1.2 x 0.3 = 0.36. The final control instruction: temperature loop = 0.33 + 0.48 = 0.81 (converted to heating valve opening 81%), vacuum degree loop = 0.38 + 0.36 = 0.74 (mapped to vacuum pump frequency 45Hz). In the pressure maintaining stage (pressure fluctuation <0.3kPa / min), the compensation components are frozen, and the coordinated release program is executed when pressure relief: temperature increases at 0.8℃ / min, vacuum degree recovers at a gradient of 20kPa / min.

[0133] The specific values of each key threshold and parameter described in this embodiment are determined based on the experimental design and data analysis of the system. The preset strength threshold of dynamic coupling strength is 0.65, which is determined as follows: for Qiancha No. 1 variety, 58 batches of complete process data of standardized fermentation were collected in a laminar flow fermentation tank with a volume of 5 cubic meters. For each batch, the time series data of temperature, humidity, and oxygen concentration were extracted, and the mutual information values of temperature-humidity, temperature-oxygen, and humidity-oxygen parameter pairs were calculated with a sliding window of 150 seconds, obtaining 17,400 effective sample points. After frequency statistics of the mutual information values of all samples, it was found that the distribution showed obvious skewness characteristics, and 85% of the sample points had mutual information values less than 0.652, and when the mutual information value was greater than this critical point, the probability of inducing control oscillation in the subsequent 3 control periods increased significantly by 78%. Therefore, 0.65 is established as the strength threshold for distinguishing between ordinary coupling and strong coupling, which is used for reliable screening of key parameter groups.

[0134] The setting threshold of direction duration difference is 0.7 to 1.3, which is determined by a specially designed single-factor critical experiment. The experiment fixes other conditions and only changes the ratio of the direction duration of the two same direction changing parameters, and records the adjustment time required for the system to recover from disturbance to within 2% of the set value. Experimental data shows that when the ratio is in the range of 0.73 to 1.28, the average adjustment time is the shortest and most stable; once the ratio is less than 0.69 or greater than 1.32, the standard deviation of the adjustment time will increase by 2.4 times, indicating that the control performance deteriorates. Therefore, 0.7 to 1.3 is selected as the effective threshold range of direction duration difference.

[0135] The setting matching threshold of fluctuation amplitude matching degree is 0.8 to 1.2, which is derived from a response surface analysis experiment with fermentation uniformity as the optimization target. The experiment selects temperature and humidity as the reverse change parameters, and designs 21 groups of tests in the range of the absolute value ratio of the change amount from 0.5 to 2.0. After each group of test is completed, the tea samples are subjected to color difference analysis and main biochemical component detection, and the fermentation uniformity comprehensive score is calculated. The analysis result shows that when the fluctuation amplitude matching degree is in the range of 0.82 to 1.18, the uniformity score is continuously higher than 90 points, and the quality is optimal; when the matching degree is lower than 0.78 or higher than 1.22, the score is significantly reduced. Therefore, 0.8 to 1.2 is established as the threshold interval for judging whether the fluctuation is matched.

[0136] The initial value, adjustment step and safety range of the controller output weight are determined based on the combination of control simulation and physical calibration. First, in the MATLAB / Simulink environment, a multivariable coupling model is established according to the transfer function of the fermentation tank. Through the particle swarm optimization algorithm, with the dual objectives of system overshoot less than 5% and shortest regulation time, hundreds of weight combinations are simulated and optimized, and it is found that when the initial weight distribution is temperature 0.35, humidity 0.30 and oxygen concentration 0.35, the comprehensive performance index is optimal. Regarding the adjustment step, simulation tests find that a single adjustment amount exceeding 0.08 is easy to cause overshoot, and an adjustment amount less than 0.02 is slow in response. Further step tests on the physical platform finally determine that the basic increment corresponding to the strong cooperative factor is 0.05, the linear decrement of the first type of reverse cooperative factor is 0.03, and the stepwise decrement of the second type of reverse cooperative factor is 0.07, and the system dynamic performance is best under this combination. The safety range of the weight 0.1 to 0.8 is determined by boundary test, and it is found that any loop weight less than 0.1 will lead to the loss of control of the variable, and higher than 0.8 will excessively suppress the remaining loops, causing system instability.

[0137] The above values and determination methods are detailed descriptions for the embodiments. When implementing the present application, for different tea varieties or fermentation equipment, the above parameters can be calibrated and adapted according to the same experimental logic and analysis framework through limited routine experiments.

[0138] Embodiment 2: Figure 2 The structure diagram of the tea fermentation process adaptive control system based on parameter cooperative optimization is given, and the tea fermentation process adaptive control system based on parameter cooperative optimization comprises the following modules:

[0139] The parameter acquisition module is used for acquiring the temperature control parameter, humidity control parameter and oxygen concentration control parameter of the tea fermentation process in real time, and establishing a control parameter set;

[0140] The coupling screening module is configured to identify dynamic coupling strength between any two control parameters in the control parameter set, and screen out a control parameter pair with dynamic coupling strength exceeding a preset strength threshold as a key parameter group.

[0141] The factor generation module is configured to generate a control parameter synergistic optimization factor including a strong positive synergistic factor, a weak positive synergistic factor or a compensatory negative synergistic factor according to a real-time change direction of the control parameter in the key parameter group.

[0142] The weight adjustment module is configured to dynamically adjust a controller output weight corresponding to the key parameter group based on the control parameter synergistic optimization factor.

[0143] The entropy reduction instruction module is configured to calculate a set value tracking error of the multivariable control system based on the adjusted controller output weight, calculate a control instability degree according to a change of the set value tracking error, and generate an entropy reduction compensation instruction if the control instability degree rises for three consecutive sampling periods.

[0144] The synergistic execution module is configured to execute the multivariable synergistic control using the adjusted controller output weight and superimposing the entropy reduction compensation instruction.

[0145] The calculations involved in the embodiments are all de-dimensioned numerical calculations, and preset parameters and threshold values in the calculations are set by a person skilled in the art according to actual conditions.

[0146] The above embodiments can be realized wholly or partially by software, hardware, firmware or any combination thereof. When realized by software, the above embodiments can be realized wholly or partially in the form of a computer program product.

[0147] Those skilled in the art can appreciate that the modules and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized by hardware or software depends on the specific application and the constraints of the technical solution. A person skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0148] In addition, each functional module in each embodiment of the present application can be integrated in one processing module, or each module can exist physically independently, or two or more modules can be integrated in one module.

[0149] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the division of the above-described device embodiment is only a logical function division, and there can be another division manner for actual implementation, for example, multiple devices or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different parts can be indirect couplings or communication connections through some interfaces, devices or modules, and can be in electrical, mechanical or other forms.

[0150] The above describes only the specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any modification or replacement within the technical range disclosed by the present application can be easily thought by any person skilled in the art, and should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0151] Finally: the above described is only the preferred embodiment of the present application, and is not used to limit the present application, any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. An adaptive control method for tea fermentation process based on parameter collaborative optimization, characterized in that, Includes the following steps: S1. Real-time acquisition of temperature control parameters, humidity control parameters, and oxygen concentration control parameters during the tea fermentation process, and establishment of a set of control parameters; S2. Identify the dynamic coupling strength between any two control parameters in the control parameter set, and select control parameter pairs whose dynamic coupling strength exceeds a preset strength threshold as key parameter groups. S3. Generate control parameter collaborative optimization factors, including strong positive collaborative factors, weak positive collaborative factors, or compensating reverse collaborative factors, based on the real-time change direction of control parameters in the key parameter group. Calculate the directional duration of each control parameter within a continuous sampling period; When two control parameters change in the same direction and the difference in their duration is less than a set threshold, a strong positive synergistic factor is generated. When two control parameters change in the same direction but the difference in their duration exceeds a set threshold, a weak positive synergistic factor is generated. When the two control parameters change in opposite directions, a compensating reverse coordination factor is generated based on the degree of matching of fluctuation amplitudes. S4. Dynamically adjust the controller output weights corresponding to key parameter groups based on the collaborative optimization factor of control parameters. S5. Based on the adjusted controller output weights, calculate the setpoint tracking error of the multivariable control system, calculate the control instability according to the change of the setpoint tracking error, and generate an entropy reduction compensation command if the control instability increases for three consecutive sampling periods. S6. Use the adjusted controller output weights and superimpose entropy reduction compensation instructions to perform multivariable collaborative control.

2. The adaptive control method for tea fermentation process based on parameter collaborative optimization according to claim 1, characterized in that, Real-time acquisition of temperature, humidity, and oxygen concentration control parameters during the tea fermentation process; establishment of a set of control parameters, including: Temperature control parameters for the tea fermentation process are obtained using a temperature sensor. Humidity control parameters for the tea fermentation process are obtained using a humidity sensor. Oxygen concentration control parameters for the tea fermentation process are obtained using an oxygen concentration sensor. The temperature control parameters, humidity control parameters, and oxygen concentration control parameters acquired at the same sampling time are synchronized and aligned in time. A set of control parameters is established based on the time-synchronized and aligned temperature control parameters, humidity control parameters, and oxygen concentration control parameters.

3. The adaptive control method for tea fermentation process based on parameter collaborative optimization according to claim 1, characterized in that, Identify the dynamic coupling strength between any two control parameters in the control parameter set, and select control parameter pairs whose dynamic coupling strength exceeds a preset strength threshold as key parameter groups, including: Extract the numerical sequences of temperature control parameters, humidity control parameters, and oxygen concentration control parameters within a continuous time window from the set of control parameters; For any two control parameters, calculate the cross-correlation coefficient of the numerical sequences of the two control parameters within a time window; The absolute value of the cross-correlation coefficient is used as the dynamic coupling strength; Compare the dynamic coupling strength with the preset strength threshold. When the dynamic coupling strength is greater than the preset strength threshold, the control parameter pair consisting of the two corresponding control parameters is added to the key parameter group.

4. The adaptive control method for tea fermentation process based on parameter collaborative optimization according to claim 1, characterized in that, The compensation-type inverse synergistic factor generated based on the volatility amplitude matching degree includes: Calculate the absolute value of the change of the two control parameters in the current sampling period compared to the previous sampling period; Compare the ratio of the absolute values ​​of the two changes with a set matching threshold; When the ratio is within the set matching threshold range, a first-type compensating reverse collaborative factor is generated; When the ratio exceeds the set matching threshold range, a second type of compensatory reverse collaborative factor is generated.

5. The adaptive control method for tea fermentation process based on parameter collaborative optimization according to claim 1, characterized in that, The controller output weights corresponding to key parameter groups are dynamically adjusted based on the control parameter collaborative optimization factor, including: For each pair of control parameters in the key parameter group: When the control parameter co-optimization factor is a strong positive co-optimization factor, increase the controller output weights corresponding to the two control parameters in the control parameter pair; When the control parameter co-optimization factor is a weak positive co-optimization factor, the controller output weights corresponding to the two control parameters in the control parameter pair remain unchanged. When the control parameter co-optimization factor is a compensating reverse co-optimization factor, the controller output weights corresponding to the two control parameters in the control parameter pair are reduced.

6. The adaptive control method for tea fermentation process based on parameter collaborative optimization according to claim 1, characterized in that, Based on the adjusted controller output weights, the setpoint tracking error of the multivariable control system is calculated. The control instability is then calculated based on the changes in the setpoint tracking error. If the control instability increases for three consecutive sampling periods, an entropy reduction compensation command is generated, including: Based on the adjusted controller output weights, the difference between the actual value of the temperature control parameter and the temperature setpoint is calculated as the temperature setpoint tracking error, the difference between the actual value of the humidity control parameter and the humidity setpoint is calculated as the humidity setpoint tracking error, and the difference between the actual value of the oxygen concentration control parameter and the oxygen concentration setpoint is calculated as the oxygen concentration setpoint tracking error. Calculate the absolute value of the change in temperature setpoint tracking error in the current sampling period compared to the temperature setpoint tracking error in the previous sampling period; calculate the absolute value of the change in humidity setpoint tracking error in the current sampling period compared to the humidity setpoint tracking error in the previous sampling period; calculate the absolute value of the change in oxygen concentration setpoint tracking error in the current sampling period compared to the oxygen concentration setpoint tracking error in the previous sampling period. The absolute values ​​of the changes in temperature setpoint tracking error, humidity setpoint tracking error, and oxygen concentration setpoint tracking error are added together to obtain the control instability of the current sampling period. Determine whether the control instability of the current sampling period is greater than the control instability of the previous sampling period, and whether the control instability of the previous sampling period is greater than the control instability of the sampling period before that. When the condition is met, an entropy reduction compensation instruction is generated.

7. The adaptive control method for tea fermentation process based on parameter collaborative optimization according to claim 1, characterized in that, Multivariable cooperative control is performed using the adjusted controller output weights and superimposed entropy reduction compensation instructions, including: The entropy reduction compensation command is distributed to the temperature control loop, humidity control loop and oxygen concentration control loop according to a preset ratio to obtain the compensation component of the temperature control loop, the compensation component of the humidity control loop and the compensation component of the oxygen concentration control loop. Based on the compensation components of the temperature control loop, humidity control loop, and oxygen concentration control loop, the final control commands for the temperature control loop, humidity control loop, and oxygen concentration control loop are obtained. The temperature control actuator is driven according to the final control command of the temperature control loop, the humidity control actuator is driven according to the final control command of the humidity control loop, and the oxygen concentration control actuator is driven according to the final control command of the oxygen concentration control loop.

8. The adaptive control method for tea fermentation process based on parameter collaborative optimization according to claim 7, characterized in that, The output weight of the temperature control loop in the adjusted controller output weight is added to the compensation component of the temperature control loop to obtain the final control command of the temperature control loop. The output weight of the humidity control loop in the adjusted controller output weight is added to the compensation component of the humidity control loop to obtain the final control command of the humidity control loop. The output weight of the oxygen concentration control loop in the adjusted controller output weight is added to the compensation component of the oxygen concentration control loop to obtain the final control command of the oxygen concentration control loop.

9. An adaptive control system for tea fermentation process based on parameter collaborative optimization, used to implement the adaptive control method for tea fermentation process based on parameter collaborative optimization as described in any one of claims 1-8, characterized in that, Includes the following modules: The parameter acquisition module is used to acquire temperature control parameters, humidity control parameters, and oxygen concentration control parameters in the tea fermentation process in real time, and to establish a set of control parameters. The coupling filtering module is used to identify the dynamic coupling strength between any two control parameters in the control parameter set, and filter out control parameter pairs whose dynamic coupling strength exceeds a preset strength threshold as key parameter groups. The factor generation module is used to generate control parameter collaborative optimization factors, including strong positive collaborative factors, weak positive collaborative factors, or compensatory reverse collaborative factors, based on the real-time change direction of control parameters in the key parameter group. The weight adjustment module is used to dynamically adjust the controller output weights corresponding to the key parameter groups based on the collaborative optimization factor of the control parameters. The entropy reduction instruction module is used to calculate the setpoint tracking error of a multivariable control system based on the adjusted controller output weights, calculate the control instability based on the change in the setpoint tracking error, and generate an entropy reduction compensation instruction if the control instability increases for three consecutive sampling periods. The collaborative execution module is used to perform multivariable collaborative control by using the adjusted controller output weights and superimposing entropy reduction compensation instructions.

Images