A high-activity selenium yeast intelligent production control system based on gradient feed monitoring

Abstract

The present application relates to the technical field of program control, and discloses a high-activity selenium yeast intelligent production control system based on gradient feeding monitoring. The system comprises a data acquisition module, a control parameter acquisition module, a gradient calculation module, an intelligent control module and a feeding control module. The system acquires process production data, extracts key control parameters and weights, calculates real-time gradient values of each parameter and forms a six-dimensional gradient vector, generates a control gradient index in combination with parameter weights and time sequence attenuation weights, and then forms an addition control decision. According to the decision and process state dynamic data, the feeding opportunity is identified and an active addition instruction is generated, so that the programmed control and adjustment of raw materials and additives are realized, and the control precision and production automation level are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of program control, in particular to a high-activity selenium yeast intelligent production control system based on gradient feeding monitoring. BACKGROUND

[0002] Modern industrial batch production process is a complex multivariable, nonlinear dynamic system, which requires accurate control and coordinated management of multiple key process parameters such as temperature, pressure, pH, flow, concentration, etc. Batch production has the characteristics of intermittence and periodicity, and the change law and control requirements of process parameters in each production cycle are significantly different.

[0003] At present, the batch production process mainly uses traditional program control system based on PID controller. This kind of control system adjusts various process parameters in production process through preset control program and fixed control algorithm. The traditional program control system usually adopts single feedback control strategy, calculates control output signal according to the deviation between real-time data monitored by sensor and set value, and adjusts the running state of corresponding actuator, such as heater, stirrer, feeding device and other equipment. The control program is generally controlled by time sequence or process stage, and each stage adopts preset control parameter combination.

[0004] However, batch production process has significant nonlinear, time-varying and strong coupling characteristics. In actual production, different batches of process often show obvious difference, which is mainly caused by small fluctuations of raw material quality, changes of environmental conditions, differences of equipment state and inconsistencies of operation conditions, etc. Even under the same process parameter setting, the reaction kinetics curve, material conversion rate and product generation law of different batches may have great difference, which leads to the difficulty in ensuring the stability of final product quality and production efficiency.

[0005] In addition, the traditional program control system lacks deep understanding of process mechanism and prediction ability. The feeding control strategy is usually based on empirical formula or preset time program, which cannot be dynamically adjusted according to real-time process state and material demand. This open-loop feeding control method often leads to waste of raw materials, and may cause uneven local concentration distribution or imbalance of proportioning, affecting the stability of process and product quality. Improper control of feeding time and feeding rate will cause fluctuation of process parameters, further aggravating the difference between batches.

[0006] The existing monitoring system mainly relies on single-point measurement and discrete sampling, lacking the ability of continuous monitoring of the gradient change of the process. The conventional monitoring method cannot timely capture the transient change and local anomaly of the process parameters, resulting in time lag in the response of the control system. The gradient monitoring technology can realize real-time tracking of the change trend of the process parameters, providing more abundant and accurate information feedback for intelligent control.

[0007] Therefore, a high-activity selenium yeast intelligent production control system based on gradient feeding monitoring is provided. SUMMARY

[0008] The purpose of the present application is to provide a high-activity selenium yeast intelligent production control system based on gradient feeding monitoring, which comprises a data acquisition module for acquiring process production data; a control parameter acquisition module for extracting key control parameters and weights of the production data; a gradient calculation module for obtaining real-time gradient values of each parameter in the process production data and obtaining a six-dimensional gradient vector; analyzing the six-dimensional gradient vector to obtain process state dynamic data; an intelligent control module for fusing the six-dimensional gradient vector based on the parameter importance weight and the timing decay weight of the key control parameters to generate a control gradient index; determining the type and amount of additives according to the control gradient index to generate an addition control decision; a feeding control module for identifying the best addition opportunity window based on the addition control decision and the process state dynamic data to generate an active addition control instruction; and adding the reaction raw materials and functional additives according to the active addition control instruction.

[0009] To achieve the above purpose, the present application provides the following technical scheme:

[0010] A high-activity selenium yeast intelligent production control system based on gradient feeding monitoring, comprising:

[0011] A data acquisition module for real-time acquisition of process production data;

[0012] A control parameter acquisition module for extracting key control parameters of the process production data through a control parameter identification model;

[0013] A gradient calculation module for obtaining real-time gradient values of each parameter in the process production data and obtaining a six-dimensional gradient vector; constructing a multi-dimensional gradient vector field model to analyze the six-dimensional gradient vector to obtain process state dynamic data;

[0014] An intelligent control module for weighted fusion of the six-dimensional gradient vector based on the parameter importance weight and the timing decay weight of the key control parameters to generate a control gradient index; determining the type and amount of additives according to the control gradient index to generate an addition control decision;

[0015] The feed control module identifies an optimal addition time window based on the addition control decision and the process state dynamic data, and generates active addition control instructions; and the reaction raw materials and functional additives are added according to the active addition control instructions.

[0016] Preferably, the process production data includes solid concentration, reactant concentration, target product concentration, additive concentration, gas solubility and pH value;

[0017] The control parameter identification model includes: a feature extraction layer that pre-processes and vectorizes the process production data, extracts time domain feature parameters and frequency domain feature parameters; a weight calculation layer that uses an attention mechanism to evaluate the importance of the time domain feature parameters and the frequency domain feature parameters, and calculates the influence weight coefficients of each parameter; a parameter screening layer that obtains key control parameters based on weight threshold and correlation analysis; and a dynamic adjustment layer that updates the influence weight coefficients according to historical key control parameter feedback.

[0018] Preferably, the six-dimensional gradient vector acquisition process includes: sampling the parameters in the process production data in a continuous time sequence, calculating the change rates of the parameters in a continuous time window, obtaining the temperature change gradient, the pressure change gradient, the flow change gradient, the concentration change gradient, the pH change gradient and the dissolved oxygen change gradient, and constructing a six-dimensional gradient vector;

[0019] The multi-dimensional gradient vector field model includes: a vector decomposition layer that decomposes the six-dimensional gradient vector, extracts gradient components, including heat transfer state components, fluid dynamics state components, material transport state components, chemical reaction state components, ion balance state components and oxidation-reduction state components; a field strength calculation layer that calculates the corresponding field strength distribution based on the gradient components, obtains multi-dimensional field strengths, including temperature field strength, pressure field strength, flow field strength, concentration field strength, pH field strength and dissolved oxygen field strength; a state fusion layer that uses tensor operation to weight and fuse the multi-dimensional field strengths, obtains fused field strength features; and a prediction output layer that predicts process state dynamic data based on the fused field strength features, including stability evaluation, trend prediction and abnormal warning.

[0020] Preferably, the parameter importance weight acquisition process includes:

[0021] The historical process production data is acquired, the contribution of each parameter to the quality is calculated, and the parameter importance weight is allocated according to the contribution size; and the time sequence decay weight acquisition process includes: setting a time decay strategy, and allocating the weight according to the time distance;

[0022] The control gradient index acquisition process comprises: weighted calculation, normalization of parameter importance weight and time sequence attenuation weight, combination of real-time gradient value, and acquisition of weighted gradient control component; multi-dimensional fusion operation, analysis of the weighted gradient control component, and acquisition of control gradient value; index standardization processing, processing of the control gradient value by using a hyperbolic tangent function, and generation of a standardized control gradient index.

[0023] Preferably, the generation process of the addition control decision comprises:

[0024] Type mapping judgment, establishment of a corresponding relationship between the control gradient index and the type of the additive, selection of a buffering additive when the index value is less than a first threshold value, selection of a nutrient additive when the index value is greater than or equal to the first threshold value and less than a second threshold value, selection of a regulating additive when the index value is greater than or equal to the second threshold value and less than a third threshold value, selection of a catalytic additive when the index value is greater than or equal to the third threshold value and less than a fourth threshold value, and selection of an inhibitory additive when the index value is greater than or equal to the fourth threshold value;

[0025] Dose precision calculation, optimization control of the dosing amount of each type of additive based on the control gradient index and in combination with a proportional-integral-derivative control algorithm, generation of an addition control decision, and inclusion of additive type code, precise dosing amount value, execution priority ranking, and estimated cost information.

[0026] Preferably, the optimal addition time window identification process comprises:

[0027] Based on the process state dynamic data, a stability index is calculated, and the fluctuation degree of each parameter is analyzed to determine the stability level;

[0028] According to the action mechanism and response characteristics of different additives, the dosing time window is predicted;

[0029] A time conflict detection mechanism is established among multiple additives to obtain a conflict detection result; when the predicted dosing time interval of two and / or multiple additives is less than a preset time, the time conflict is determined, a priority queue management mechanism is used to solve the conflict according to the priority ranking in the addition control decision, the high-priority additive is executed preferentially, and the low-priority additive is executed later;

[0030] Considering the stability index, the stability level, the dosing time window, and the conflict detection result, a dynamic programming algorithm is used to obtain the optimal addition time window of each additive.

[0031] Preferably, the dosing control program process comprises:

[0032] The active addition control instruction is analyzed, and the adding parameter and execution sequence are extracted; the corresponding metering pump, valve and pipeline system are dispatched according to the adding requirement; the closed-loop control mode is adopted to monitor the adding flow and cumulative amount in real time, so that the adding precision is ensured; the process parameter change is continuously monitored during the adding process, and the adding speed and mode are adjusted in time; after the adding is completed, the system is reset and the state is confirmed, and the adding execution log is recorded.

[0033] Compared with the prior art, the beneficial effects of the present application are:

[0034] 1、The present application comprehensively analyzes the gradient trend of six-dimensional parameter (temperature, pressure, flow, concentration, pH and dissolved oxygen) changes by constructing a multi-dimensional gradient vector field model, and extracts various state information such as heat transfer, fluid dynamics and chemical reaction of the process, so as to comprehensively perceive the dynamic evolution of the process state. By predicting the process stability, trend change and abnormal fluctuation, real-time perception and trend prediction of key process state are realized. By fusing tensor calculation and state field strength modeling, not only the understanding ability of complex biological reaction process is improved, but also the timeliness and accuracy of control decision are significantly improved, so that the whole high-activity selenium yeast production process is changed from "static control" to "dynamic intelligent adjustment", so that the reaction efficiency is optimized and the fluctuation risk is reduced while the product quality stability is ensured.

[0035] 2、The present application introduces a double fusion based on "parameter importance weight" and "time sequence decay weight", and the six-dimensional gradient information monitored in real time is weighted to form a control gradient index. The control gradient index not only reflects the actual influence degree of each parameter on the process, but also considers the decay trend of the parameter in the time dimension, so as to realize scientific evaluation and control basis of parameter change. On this basis, the system can intelligently match the type of additive and accurately calculate the additive amount according to the standardized control gradient index, realize the whole process closed-loop control from "quantitative perception" to "precise execution", significantly improve the material utilization efficiency and product consistency, and provide strong control support for high-standard yeast production.

[0036] 3、The present application constructs a dynamic planning algorithm based on stability index and time window prediction, and fuses the action mechanism and response characteristics of the additive to dynamically identify the optimal adding time window. At the same time, a multi-additive time conflict detection mechanism is set, and a priority queue management strategy is adopted to effectively avoid the addition conflict in a short time. This way not only ensures the maximum play of various additives, but also avoids the waste of resources and the decline of product quality caused by addition sequence conflict, and improves the cooperativity of the whole production process. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1A high-activity selenium yeast intelligent production control system structure schematic diagram based on gradient feeding monitoring is provided for the present application.

[0038] Figure 2 A high-activity selenium yeast intelligent production control flow schematic diagram based on gradient feeding monitoring is provided for the present application.

[0039] Figure 3 An intelligent production control flow schematic diagram is provided for the present application embodiment. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a 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.

[0041] Embodiment one:

[0042] The present application provides a high-activity selenium yeast intelligent production control system based on gradient feeding monitoring, and the technical solutions are as follows, for specific reference Figure 1 and Figure 2 :

[0043] The data acquisition module is used for real-time acquisition of process production data.

[0044] The control parameter acquisition module extracts the key control parameters of the process production data through a control parameter identification model.

[0045] The gradient calculation module is used for obtaining the real-time gradient values of each parameter in the process production data, obtaining a six-dimensional gradient vector, constructing a multi-dimensional gradient vector field model to analyze the six-dimensional gradient vector, and obtaining process state dynamic data.

[0046] The intelligent control module performs weighted fusion on the six-dimensional gradient vector based on the parameter importance weight and the time sequence attenuation weight of the key control parameters, generates a control gradient index, determines the type and amount of the additive according to the control gradient index, and generates an addition control decision.

[0047] The feeding control module identifies the best addition opportunity window based on the addition control decision and the process state dynamic data, generates a proactive addition control instruction, and adds the reaction raw materials and functional additives according to the proactive addition control instruction.

[0048] Further, the process production data includes solid concentration, reactant concentration, target product concentration, additive concentration, gas solubility, and pH value.

[0049] The control parameter identification model comprises: a feature extraction layer, which pre-processes and feature vectorizes the process production data, extracts time domain characteristic parameters and frequency domain characteristic parameters; a weight calculation layer, which uses an attention mechanism to evaluate the importance of the time domain characteristic parameters and the frequency domain characteristic parameters, and calculates the influence weight coefficients of the parameters; a parameter screening layer, which obtains key control parameters based on a weight threshold and correlation analysis; and a dynamic adjustment layer, which updates the influence weight coefficients according to historical key control parameter feedback.

[0050] In the embodiment, the application comprehensively reflects the dynamic change state in the production process of high-activity selenium yeast by introducing multi-dimensional process production data. By constructing a control parameter identification model comprising a feature extraction layer, a weight calculation layer, a parameter screening layer and a dynamic adjustment layer, deep mining and feature extraction of production data are realized. Through adaptive adjustment capability, the parameter weight and screening strategy can be optimized according to historical feedback information, so as to improve the perception ability and response speed of the entire control system to the change of production state, ensure that the feeding decision is more accurate and timely, effectively improve the yield and quality of high-activity selenium yeast, reduce resource waste and production fluctuation risk, and has significant industrial application value and intelligent control advantage.

[0051] The six-dimensional gradient vector acquisition process comprises: collecting six core process parameters of temperature, pressure, feed flow, substance concentration, pH and dissolved oxygen concentration in the reactor in real time through a sensor array, continuously sampling each parameter in time series, the sampling frequency is ten times per second, calculating the change rate of each parameter in a continuous time window, forming temperature change gradient, pressure change gradient, flow change gradient, concentration change gradient, pH change gradient and dissolved oxygen change gradient, and combining to form a six-dimensional gradient vector;

[0052] The multi-dimensional gradient vector field model comprises: a vector decomposition layer, which decomposes the six-dimensional gradient vector to extract a temperature gradient component to reflect a heat transfer state, a pressure gradient component to reflect a fluid dynamics state, a flow gradient component to reflect a material conveying state, a concentration gradient component to reflect a chemical reaction state, a pH gradient component to reflect an ion balance state, and a dissolved oxygen gradient component to reflect an oxidation-reduction state; a field strength calculation layer, which calculates corresponding field strength distributions based on the gradient components, wherein a temperature field strength is obtained by multiplying the temperature gradient by a heat conduction coefficient, a pressure field strength is obtained by multiplying the pressure gradient by a flow resistance coefficient, a flow field strength is obtained by multiplying the flow gradient by a pipeline characteristic coefficient, a concentration field strength is obtained by multiplying the concentration gradient by a diffusion coefficient, a pH field strength is obtained by multiplying the pH gradient by a buffer capacity coefficient, and a dissolved oxygen field strength is obtained by multiplying the dissolved oxygen gradient by a mass transfer coefficient; six field strength components are fused by tensor operation, and the weight distribution is determined according to the influence degree of each parameter on the process; a prediction output layer, which predicts the process state at the next moment based on the fused field strength and a historical state transition mode, generates stability evaluation, trend prediction and abnormal early warning.

[0053] In the embodiment, by constructing the six-dimensional gradient vector and introducing the multi-dimensional gradient vector field model, high-frequency and continuous dynamic monitoring of the changes of various key process parameters in the high-activity selenium yeast production process is realized, and the change trends of core variables such as temperature, pressure, flow, concentration, pH and dissolved oxygen in the production process can be accurately captured. By functionally decomposing the six-dimensional gradient vector according to process attributes, not only the physical analysis ability of the changes of each parameter is improved, but also the heat transfer state, the fluid dynamics state, the material conveying state, the chemical reaction state, the ion balance state and the oxidation-reduction state are systematically described. The field strength calculation layer further converts each gradient component into a corresponding process field strength value, and combines the tensor weighted fusion technology to extract the comprehensive dynamic change characteristics reflecting the overall process operation. Finally, through the state space model in the prediction output layer, the evolution trend of the process state can be perceived in advance, and stability evaluation, trend prediction and abnormal early warning information are output, thereby providing strong data support for intelligent control and precise feeding. The deep dynamic perception and prediction capability effectively improves the controllability and stability of the process, enhances the adaptability of the system to complex changing conditions, and reduces the frequency of manual intervention and abnormal response delay.

[0054] Further, the parameter importance weight acquisition process comprises:

[0055] The historical process production data is acquired, the historical process data is collected to establish a parameter effect correlation database, the contribution degree of each parameter to the final product quality is calculated by using a principal component analysis method, and the importance weight is allocated according to the contribution degree size; for example, the temperature parameter weight accounts for 25%, the pressure parameter weight accounts for 15%, the flow parameter weight accounts for 20%, the concentration parameter weight accounts for 25%, the acid-base degree parameter weight accounts for 10%, and the dissolved oxygen parameter weight accounts for 5%; the time sequence attenuation weight acquisition process comprises: setting a time attenuation strategy, and performing weight allocation on the historical data according to the time distance; for example, the current time weight is full score, the previous one minute data weight is 95% of full score, the previous five minutes data weight is 80% of full score, the previous ten minutes data weight is 60% of full score, and the previous thirty minutes data weight is 30% of full score, so as to ensure that the recent data play a leading role in the control decision;

[0056] The control gradient index acquisition process comprises:

[0057] The weight normalization processing is performed, the parameter importance weight is multiplied by the time sequence attenuation weight, and the comprehensive weight of each parameter at different time is obtained;

[0058] The gradient value weighting calculation is performed, the real-time gradient value of each parameter is multiplied by the corresponding comprehensive weight, and six weighted gradient components are obtained;

[0059] The multi-dimensional fusion operation is performed, the six weighted gradient components are linearly combined and summed, and a single control gradient value is formed;

[0060] The index standardization processing is performed, the control gradient value is mapped into a standard interval of zero to one by using a hyperbolic tangent function, and a standardized control gradient index is generated for subsequent decision judgment.

[0061] In this embodiment, by introducing the dual mechanism of parameter importance weight and timing decay weight, the precise dynamic weighting processing of process control data is realized. By constructing the parameter effect correlation database between historical process data and product quality, and combining with the principal component analysis method, the actual contribution of each core parameter to the final product quality can be quantified scientifically, so as to allocate the parameter importance weight with basis, so that the control strategy is more targeted and scientific. At the same time, the timing decay mechanism effectively reflects the importance of data timeliness, ensuring that the current and recent working condition changes have a dominant effect on control judgment, significantly improving the timeliness and sensitivity of control response. On this basis, by weighting and fusing the real-time gradient value and the comprehensive weight, and using multi-dimensional linear combination, a unified control gradient index is formed, and then through standardization processing, it is mapped to a unified evaluation interval, which not only enhances the stability and comparability of the gradient index, but also provides a clear and intuitive reference for subsequent intelligent control and feed decision-making. The method improves the overall system's comprehensive perception and intelligent judgment ability of complex process state, which helps to realize intelligent production control with higher quality, higher efficiency and lower fluctuation.

[0062] Further, the generation process of the addition control decision includes:

[0063] Type mapping judgment, establishing the correspondence between the control gradient index and the type of additive, when the index value is less than the first threshold value, selecting the buffer type additive, when the index value is greater than or equal to the first threshold value and less than the second threshold value, selecting the nutrient type additive, when the index value is greater than or equal to the second threshold value and less than the third threshold value, selecting the adjustment type additive, when the index value is greater than or equal to the third threshold value and less than the fourth threshold value, selecting the catalytic type additive, and when the index value is greater than or equal to the fourth threshold value, selecting the inhibition type additive;

[0064] Dose accurate calculation, based on the proportional-integral-derivative control algorithm to calculate the dosing amount of each type of additive, the proportional control part determines the basic dosing amount according to the current deviation, the integral control part compensates and adjusts according to the cumulative deviation, and the differential control part makes predictive adjustment according to the deviation trend, and the total dosing amount is obtained by weighted summation of the three parts;

[0065] Constraint optimization, setting upper and lower limit constraints of the addition amount to avoid excessive addition, setting addition rate constraints to prevent system impact, setting cost constraints to control economy, and finding the optimal dosing scheme under the condition of meeting all constraints through a multi-objective optimization algorithm;

[0066] Decision information output, generating complete addition control decisions including additive type code, accurate dosing amount value, execution priority ranking and estimated cost information.

[0067] In this embodiment, by accurately analyzing and mapping the control gradient index, the correspondence between the index interval and different types of additives is established, the intelligent recognition and automatic classification of additive types are realized, and the pertinence and scientificity of the feeding decision are effectively improved. By introducing the proportional-integral-derivative (PID) control algorithm for dynamic calculation of the dosage, not only the current process state deviation can be responded in time, but also the historical trend and future change trend can be considered, so that the precise control of the additive dosage is realized. At the same time, the system sets multiple constraint conditions including the upper and lower limits of the addition amount, the addition rate and the cost, and adopts a multi-objective optimization algorithm to ensure the optimal balance between process stability, economy and execution feasibility, effectively preventing system fluctuations and resource waste caused by excessive or rapid addition. The finally generated addition control decision contains multi-dimensional information such as type code, specific dosage, execution priority and cost estimation, which facilitates quick response and automatic execution of the system, greatly improves the decision efficiency, dosage accuracy and overall economic benefit of the intelligent control system, and ensures the stability of the high-activity selenium yeast production process and high-quality output.

[0068] Further, the optimal addition time window identification process comprises:

[0069] A stability index is calculated based on the dynamic data of the process state, and the stability level is determined by analyzing the fluctuation degree of each parameter; the stability level of the system is determined by analyzing the fluctuation degree of each parameter, and when the stability index is greater than 0.8, it is determined as a high stability state suitable for immediate addition, when the index is between 0.6 and 0.8, it is determined as a medium stability state requiring cautious addition, and when the index is less than 0.6, it is determined as a low stability state for temporary addition;

[0070] According to the action mechanism and response characteristics of different additives, the addition time window is predicted; for example, a thirty-minute prediction window is set for buffer additives to consider their sustained stabilizing effect, a sixty-minute prediction window is set for nutritional additives to consider their metabolic absorption process, a fifteen-minute prediction window is set for regulating additives to consider their rapid regulating effect, a forty-five-minute prediction window is set for catalytic additives to consider their reaction promotion period, and a twenty-minute prediction window is set for inhibitory additives to consider their inhibition response time;

[0071] A time conflict detection mechanism is established among multiple additives to obtain a conflict detection result; when the predicted addition time interval of two and / or multiple additives is less than the preset time, it is determined as a time conflict, and a priority queue management mechanism is used to solve the conflict according to the priority sorting in the addition control decision, and the high-priority additives are executed first, and the low-priority additives are executed later;

[0072] Considering the stability index, stability level, addition time window and conflict detection result, a dynamic programming algorithm is used to obtain the optimal addition time window of each additive.

[0073] In this embodiment, by introducing the stability index calculation and multi-dimensional process parameter fluctuation analysis, the stability level of the current state of the system can be dynamically evaluated, thereby providing a scientific risk assessment basis for the addition of additives. The action mechanism and response characteristics of different types of additives are fully considered, and by setting different prediction time windows, the use of each type of additive is more in line with its actual effect period, improving the accuracy and timeliness of the feeding strategy. At the same time, the system constructs an additive time conflict detection and priority management mechanism, avoiding the action interference or resource competition between different types of additives, effectively guaranteeing the continuity and coordination of the process system operation. Finally, through dynamic programming algorithm, multiple factors such as stability level, prediction window and time conflict are considered to realize the intelligent identification and reasonable arrangement of the best addition time of various additives, significantly improving the automation level and execution reliability of the addition control, and providing a strong support for the continuous optimization and quality guarantee of the high-activity selenium yeast production process.

[0074] Further, the addition control program process comprises:

[0075] The active addition control instruction is analyzed to extract the addition parameters and execution sequence; the corresponding metering pump, valve and pipeline system are dispatched according to the addition requirements; a closed-loop control mode is adopted to monitor the addition flow and cumulative amount in real time, ensuring the addition accuracy; the process parameter changes are continuously monitored during the addition process, and the addition rate and mode are adjusted in time; after the addition is completed, the system is reset and the state is confirmed, and the addition execution log is recorded.

[0076] Specifically, the active addition control instruction is structured and analyzed to extract the additive type identification code, target addition amount value, set addition rate parameter and execution time window range, verify the instruction integrity and parameter rationality, and generate a standardized execution parameter set; the additive type identification code is used to query the equipment configuration database to automatically select the corresponding metering pump number, control valve position and conveying pipeline path, establish a complete addition channel connection, execute the equipment self-checking program to confirm the normal equipment state, and perform pipeline pre-flushing and equipment pre-heating preparation; the closed-loop feedback control system is started to realize accurate addition, the target flow value of the flow controller is set, the actual flow data in the pipeline is monitored in real time, the deviation between the target value and the actual value is calculated, the speed of the metering pump is adjusted through the control algorithm to realize flow tracking, and the added amount is accumulated and calculated; when the accumulated amount reaches 98% of the target addition amount, the addition is stopped to ensure accuracy; the real-time changes of six-dimensional process parameters in the reactor are monitored at a frequency of once per second during the entire addition process, the instantaneous change rate of each parameter is calculated, and when the change rate of any parameter exceeds the preset safety threshold, an abnormal alarm is triggered immediately, and a self-adaptive control strategy is started to dynamically adjust the addition rate to reduce the impact on the system; after the addition task is completed, a standardized equipment reset operation sequence is executed, the related control valves are closed to stop material flow, the metering pump is stopped to prevent continuous conveying, and detailed execution logs including time stamp, additive code, actual addition amount, execution time and completion status are recorded to update the equipment operation record and maintenance plan arrangement.

[0077] In the embodiment, by structured analysis of the active addition control instruction, the standardization of the extraction of the addition task parameters and the reasonable scheduling of the execution sequence can be realized, and the clarity of the control logic and the consistency of the execution are significantly improved. Combined with the automatic matching function of the equipment configuration database, the system can intelligently select and connect the metering pump, valve and conveying pipeline, automatically build a complete addition channel and perform pre-processing preparation, reduce human intervention, and improve the automation level. Through the closed-loop feedback control mechanism, the addition flow and cumulative amount are accurately adjusted in real time, the accuracy of the additive amount is effectively guaranteed, and the addition is actively stopped when approaching the target value to prevent over-addition. At the same time, the system monitors the dynamic changes of the multi-dimensional process parameters in the reactor in real time at a high frequency, can identify abnormal fluctuations in time and start the self-adaptive control strategy to dynamically adjust the addition rate, reduce the impact of the addition operation on the system stability. The standardized reset process after the task is completed and the detailed execution log record not only contribute to process traceability and operation audit, but also provide decision basis for subsequent equipment maintenance and operation optimization, and overall, the precision, safety and intelligent level of the addition control are greatly improved.

[0078] The application provides a high-activity selenium yeast intelligent production control system based on gradient feeding monitoring, and specifically refers to Figure 3, through multi-dimensional data acquisition, deep feature extraction and dynamic perception mechanism, precise dynamic control and intelligent management of the production process are realized. The system real-time collects process production data including solid concentration, reactant concentration, target product concentration, additive concentration, gas solubility and pH, and extracts key parameters and their weights through a control parameter identification model, combines a six-dimensional gradient vector with a multi-dimensional gradient vector field model to monitor and predict the process state dynamically. This forward-looking perception capability not only can identify the trend of change in the production process in advance, but also can generate stability evaluation and abnormal warning, providing a reliable basis for intelligent feeding. The intelligent control module generates a scientific and reasonable control gradient index through the weighted fusion of parameter importance and time sequence decay weight, accurately determines the type and dosage of additives, and optimizes the best addition time window, ensuring the efficiency and pertinence of the feeding decision. In addition, the closed-loop feedback and adaptive adjustment mechanism further improves the addition accuracy and system stability. Overall, the system significantly improves the yield and quality of high-activity selenium yeast, reduces resource waste and production fluctuation risk, and exhibits excellent industrial application value and technical popularization potential, providing an innovative solution for intelligent production.

[0079] Embodiment Two:

[0080] Please refer to Figure 2 The application provides a specific application embodiment of a high-activity selenium yeast intelligent production control system based on gradient feeding monitoring in an industrial production environment. This embodiment takes a production line of a certain biological pharmaceutical enterprise with an annual output of 500 tons of high-activity selenium yeast as the application scenario, and details the cooperative working mechanism of each technical module. The specific scheme is as follows: real-time acquisition of process production data; extraction of key control parameters of process production data through a control parameter identification model; acquisition of real-time gradient values of each parameter in process production data and six-dimensional gradient vectors; construction of a multi-dimensional gradient vector field model to analyze the six-dimensional gradient vectors and obtain process state dynamic data; weighted fusion of the six-dimensional gradient vectors based on the parameter importance weight and time sequence decay weight of the key control parameters to generate control gradient indicators; determination of the type and amount of additives according to the control gradient indicators to generate addition control decisions; identification of the best addition time window based on the addition control decisions and process state dynamic data to generate active addition control instructions; and addition of reaction raw materials and functional additives according to the active addition control instructions.

[0081] In the high-activity selenium yeast production line reactor, the volume is 10 cubic meters and is equipped with complete automation control equipment. The data acquisition module monitors the changes of key process parameters in the fermentation process through a distributed sensor network. The entire production cycle is 72 hours, divided into three stages: strain culture period (0-24 hours), rapid proliferation period (24-48 hours) and selenium enrichment period (48-72 hours), and there are significant differences in process parameter control requirements in each stage.

[0082] The data acquisition module adopts a high-precision sensor array, including temperature sensors, pressure sensors, flow sensors, online concentration detectors, pH meters, and dissolved oxygen analyzers, to synchronously collect process production data at a frequency of 10 times per second. In the application scenario, the process production data specifically includes: yeast cell concentration, glucose concentration, selenomethionine concentration, nutrient salt additive concentration, oxygen solubility, and pH value, forming a continuous time series data stream to provide a data basis for subsequent intelligent analysis.

[0083] The control parameter acquisition module constructs a four-layer architecture control parameter identification model, which is specifically designed for the optimization of the complex characteristics of the high-activity selenium yeast fermentation process. The feature extraction layer uses a hybrid frequency domain analysis method combining wavelet transform and fast Fourier transform to preprocess six types of process production data, extract time domain feature parameters including mean, variance, skewness, kurtosis, and frequency domain feature parameters including main frequency component, frequency spectrum energy distribution, and frequency domain correlation, and generate a 48-dimensional feature vector.

[0084] The weight calculation layer introduces a self-attention mechanism to evaluate the importance of the 48-dimensional feature vector through a multi-head attention network. In the specific implementation, 8 attention heads are set, each focusing on feature extraction in different time scales and frequency ranges, and the influence weight coefficients of each parameter are calculated through the attention weight matrix. In the culture period of the strain, the temperature and pH parameter weights are higher (0.35 and 0.28, respectively); in the rapid proliferation period, the dissolved oxygen and glucose concentration weights are dominant (0.42 and 0.31, respectively); in the selenium enrichment period, the selenomethionine concentration and nutrient salt concentration weights are the most critical (0.48 and 0.25, respectively).

[0085] The parameter selection layer dynamically selects the key control parameter set for each production stage based on the set weight threshold (0.15) and Pearson correlation analysis (threshold 0.7). The dynamic adjustment layer updates the parameter weights using the exponential moving average algorithm based on the feedback data of the key control parameters of the last 30 batches, with a learning rate of 0.1, enabling the system to have self-adaptive optimization capabilities.

[0086] The gradient calculation module realizes a breakthrough multi-scale gradient vector field intelligent perception technology. This technology calculates the real-time change rate of each process parameter by constructing a 30-second sliding sampling mechanism, forming a six-dimensional gradient vector. Compared with the traditional single-point monitoring method, this technology can capture the small trends of parameter changes, with a 35% improvement in prediction accuracy.

[0087] The vector decomposition layer of the multi-dimensional gradient vector field model uses a method combining principal component analysis and independent component analysis to intelligently decompose the six-dimensional gradient vector according to the biological process characteristics. The heat transfer state component reflects the metabolic heat generation and heat dissipation balance in the fermentation process, the fluid dynamics state component embodies the stirring and aeration effect, the material transport state component represents the mass transfer process of nutrients, the chemical reaction state component indicates the growth and metabolism state of the yeast, and the oxidation-reduction state component reflects the cell respiration and selenium conversion process.

[0088] The field strength calculation layer introduces a biological process physical model to calculate the corresponding field strength distribution based on each gradient component. The temperature field strength is calculated by a heat conduction model corrected by the Arrhenius equation, considering the temperature sensitivity of the biological reaction; the flow field strength combines the Rheological model to describe the non-Newtonian fluid characteristics of the fermentation broth; the concentration field strength combines the Fick diffusion law and the Monod kinetics model to accurately reflect the mass transfer and consumption of nutrients; and the dissolved oxygen field strength is calculated by a kLa mass transfer coefficient dynamic model to real-time evaluate the oxygen mass transfer efficiency.

[0089] The state fusion layer uses tensor decomposition technology to adaptively weight and fuse the six field strength components, and the weight coefficients are automatically adjusted according to the current fermentation stage and historical data. The prediction output layer constructs a state prediction model based on the long short-term memory network (LSTM), which can predict the process state change 15-30 minutes in advance, generate a stability evaluation index (range 0-1), a trend prediction vector, and an abnormal warning level (normal / attention / warning / danger four levels).

[0090] The intelligent control module realizes an innovative multi-weight fusion adaptive control algorithm. This algorithm considers both the parameter importance weight and the time sequence decay weight to achieve scientific weighted fusion of real-time gradient values.

[0091] In the parameter importance weight acquisition process, the system collects historical process data of 500 batches to establish an association database containing 18 process parameters and product quality indicators (selenium content, protein content, activity). The random forest regression model is used to calculate the contribution of each parameter to the final product quality, and the specific weight distribution is as follows: yeast concentration weight 0.28, glucose concentration weight 0.22, selenium methionine concentration weight 0.25, pH value weight 0.12, dissolved oxygen concentration weight 0.08, and nutrient salt concentration weight 0.05.

[0092] The time sequence decay weight uses an exponential decay function, with a decay coefficient of 0.05 / min. The specific weight distribution is as follows: current time weight 1.0, previous 1 minute weight 0.95, previous 5 minutes weight 0.78, previous 10 minutes weight 0.61, and previous 30 minutes weight 0.22, ensuring that recent data play a dominant role in control decisions.

[0093] The control gradient index acquisition process adopts a weighted combination algorithm: first, the parameter importance weight and the time sequence decay weight are normalized, then multiplied by the real-time gradient value to obtain six weighted gradient control components; second, an adaptive weight linear combination method is used to fuse the six components to form the control gradient value; finally, an improved hyperbolic tangent function is used for standardization to map the value to the [0, 1] interval to generate the standardized control gradient index.

[0094] The addition control decision generation process establishes a precise five-level mapping relationship: when the control gradient index is less than 0.2, select phosphate buffer additives to maintain pH stability; when the index is in [0.2, 0.4), select yeast paste nutritional additives to promote bacterial growth; when the index is in [0.4, 0.6), select selenomethionine regulating additives to regulate selenium content; when the index is in [0.6, 0.8), select glucose catalytic additives to improve metabolic activity; when the index is greater than or equal to 0.8, select ethanol inhibitory additives to prevent over-fermentation.

[0095] The dose precise calculation adopts a PID control algorithm, with an output dosage range of 0.1-0.5 kg / h for buffer type, 0.5-2.0 kg / h for nutritional type, 0.2-1.0 kg / h for regulating type, 1.0-3.0 kg / h for catalytic type, and 0.1-0.8 kg / h for inhibitory type. At the same time, safety constraints are set: the maximum single dosage does not exceed 0.5% of the reactor volume, and the dosage rate does not exceed 100 kg / h.

[0096] The feed control module realizes dynamic timing window intelligent optimization scheduling technology, which realizes the accurate identification of the best dosing time through multi-dimensional analysis.

[0097] The stability index calculation adopts a comprehensive evaluation model based on the weighted evaluation of the fluctuation degree of the six parameters in the process state dynamic data.

[0098] The dosing time window prediction is based on the biological mechanism of different additives. The buffer additive (phosphate) is set to a 45-minute prediction window, considering the establishment time of its pH buffering effect; the nutritional additive (yeast paste) is set to a 90-minute prediction window, considering its metabolic decomposition and absorption utilization process; the regulating additive (selenomethionine) is set to a 60-minute prediction window, considering its cell uptake and transformation period; the catalytic additive (glucose) is set to a 30-minute prediction window, considering its rapid metabolic characteristics; the inhibitory additive (ethanol) is set to a 20-minute prediction window, considering its rapid response to inhibition effect.

[0099] The time conflict detection mechanism establishes an accurate conflict matrix, and determines that there is a time conflict when the time interval of multiple additives is less than 15 minutes. The priority sorting rule is: inhibition class > buffering class > regulation class > catalysis class > nutrition class. Dynamic queue management is adopted, high-priority additives are executed immediately, and low-priority additives are automatically delayed for 15 minutes to ensure maximum additive effect.

[0100] The dynamic programming algorithm adopts a state transition equation for optimization solution, and the objective function is a weighted combination of maximizing production efficiency and minimizing cost. The state space includes four dimensions of system stable state, additive type, dosing time and dosing amount, and the optimal dosing time window of each additive is solved by Bellman optimization principle to realize global optimization control.

[0101] The dosing control program adopts a distributed execution architecture to realize real-time analysis of active addition control instructions, extract key parameters such as additive code, target dosing amount, set dosing rate and execution priority, and automatically schedule the corresponding peristaltic pump, electric regulating valve and stainless steel pipeline system to establish a complete dosing channel.

[0102] The closed-loop control system adopts a mass flow controller to realize accurate dosing, and the deviation between the set value and the actual value is controlled within ±2%. During the dosing process, the process parameters are monitored at a frequency of 1Hz, and when the change rate of any parameter exceeds the preset threshold, adaptive control is started immediately to dynamically adjust the dosing rate. After the dosing is completed, a standardized reset program is executed, and detailed logs including timestamp, additive type, actual dosing amount, execution time and system response are recorded.

[0103] Through systematic technical innovation, the intelligent, precise and efficient control of the production process of high-activity selenium yeast is realized. The multi-dimensional gradient sensing technology significantly improves the understanding and prediction ability of the system to complex biological processes, making the control strategy more scientific and reasonable. The multi-weight fusion algorithm ensures the accuracy and timeliness of the control decision, effectively dealing with various changes and disturbances in the production process. The dynamic scheduling optimization technology realizes the collaborative management of multiple additives, maximizing the effect of various additives. Overall, this system not only significantly improves product quality and production efficiency, but also greatly reduces production cost and resource consumption, providing an advanced technical solution for the industrial production of high-activity selenium yeast. Referring to Table 1, it has important industrial application value and promotion prospect.

[0104] Table 1 Comparison of performance of intelligent control system and traditional control technology of the application

[0105]

[0106] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A high-activity selenium yeast intelligent production control system based on gradient feed monitoring, characterized in that, The method comprises the following steps: a data acquisition module is used to acquire process production data in real time; a control parameter acquisition module extracts key control parameters of the process production data through a control parameter identification model; a gradient calculation module is used to acquire real-time gradient values of each parameter in the process production data and acquire a six-dimensional gradient vector; a multi-dimensional gradient vector field model is constructed to analyze the six-dimensional gradient vector and obtain process state dynamic data; the six-dimensional gradient vector acquisition process comprises the following steps: parameters in the process production data are sampled in a continuous time sequence, the change rates of the parameters in a continuous time window are calculated, and temperature change gradients, pressure change gradients, flow change gradients, concentration change gradients, pH change gradients and dissolved oxygen change gradients are obtained to form the six-dimensional gradient vector; the multi-dimensional gradient vector field model comprises the following steps: a vector decomposition layer is used to decompose the six-dimensional gradient vector and extract gradient components, including heat transfer state components, fluid dynamics state components, material conveying state components, chemical reaction state components, ion balance state components and oxidation-reduction state components; a field strength calculation layer is used to calculate corresponding field strength distributions based on the gradient components and obtain multi-dimensional field strengths, including temperature field strengths, pressure field strengths, flow field strengths, concentration field strengths, pH field strengths and dissolved oxygen field strengths; 2. The high-activity selenium yeast intelligent production control system based on gradient feed monitoring according to claim 1, characterized in that: a state fusion layer is used to perform weighted fusion on the multi-dimensional field strengths by using tensor operation to obtain fusion field strength characteristics; a prediction output layer is used to predict process state dynamic data based on the fusion field strength characteristics, including stability evaluation, trend prediction and abnormal early warning; an intelligent control module is used to perform weighted fusion on the six-dimensional gradient vector based on parameter importance weights and time sequence attenuation weights of the key control parameters to generate a control gradient index; the type and amount of an additive are determined based on the control gradient index to generate an addition control decision; 3. The high activity selenium yeast intelligent production control system based on gradient feed monitoring according to claim 1, characterized in that: a feed control module is used to identify an optimal addition time window based on the addition control decision and the process state dynamic data to generate an active addition control instruction; and the process production data comprises solid concentration, reactant concentration, target product concentration, additive concentration, gas solubility and pH value; the control parameter identification model comprises the following steps: a feature extraction layer is used to pre-process and feature vectorize the process production data to extract time domain feature parameters and frequency domain feature parameters; a weight calculation layer is used to evaluate the importance of the time domain feature parameters and the frequency domain feature parameters by using an attention mechanism and calculate influence weight coefficients of the parameters; a parameter screening layer is used to obtain key control parameters based on weight threshold values and correlation analysis; a dynamic adjustment layer is used to update the influence weight coefficients based on historical key control parameter feedback. the parameter importance weight acquisition process comprises the following steps: historical process production data is acquired, the contribution of each parameter to quality is calculated, and parameter importance weights are assigned according to the contribution size; and the time sequence attenuation weight acquisition process comprises the following steps: a time attenuation strategy is set, and weights are assigned according to time distances. The control gradient index acquisition process comprises: weighted calculation, normalizing the parameter importance weight and the time sequence decay weight, combining the real-time gradient value to obtain a weighted gradient control component; multi-dimensional fusion operation, analyzing the weighted gradient control component to obtain a control gradient value; index standardization processing, processing the control gradient value using a hyperbolic tangent function to generate a standardized control gradient index.

4. The high-activity selenium yeast intelligent production control system based on gradient feed monitoring according to claim 1, characterized in that: The generation process of the addition control decision comprises: Type mapping judgment, establishing a corresponding relationship between the control gradient index and the type of the additive, selecting a buffering additive when the index value is less than a first threshold value, selecting a nutrient additive when the index value is greater than or equal to the first threshold value and less than a second threshold value, selecting a regulating additive when the index value is greater than or equal to the second threshold value and less than a third threshold value, selecting a catalytic additive when the index value is greater than or equal to the third threshold value and less than a fourth threshold value, and selecting an inhibitory additive when the index value is greater than or equal to the fourth threshold value; Dose accurate calculation, based on the control gradient index, combining a proportional-integral-derivative control algorithm to optimize the dosing amount of each type of additive to generate an addition control decision, including additive type code, accurate dosing amount value, execution priority order and estimated cost information.

5. The high activity selenium yeast intelligent production control system based on gradient feed monitoring according to claim 1, characterized in that: The optimal addition time window identification process comprises: Based on the process state dynamic data, a stability index is calculated to analyze the fluctuation degree of each parameter and determine the stability level; According to the action mechanism and response characteristics of different additives, the dosing time window is predicted; A time conflict detection mechanism is established among multiple additives to obtain a conflict detection result; when the predicted dosing time interval of two and / or multiple additives is less than a preset time, it is determined as a time conflict, and a priority queue management mechanism is used to solve the conflict according to the priority order in the addition control decision, with high-priority additives being executed first and low-priority additives being executed later; Considering the stability index, stability level, dosing time window and conflict detection result, a dynamic programming algorithm is used to obtain the optimal addition time window of each additive.

6. The high activity selenium yeast intelligent production control system based on gradient feed monitoring according to claim 1, characterized in that: The dosing process comprises: The active addition control instruction is analyzed to extract the dosing parameters and execution sequence; the corresponding metering pump, valve and pipeline system are dispatched according to the dosing requirements; a closed-loop control method is used to monitor the dosing flow and cumulative amount in real time to ensure the dosing accuracy; the process parameter changes are continuously monitored during the dosing process to timely adjust the dosing rate and method; after the dosing is completed, the system is reset and the state is confirmed, and the dosing execution log is recorded.

Images