Chemical production self-adaptive control system for process quality detection

Abstract

The invention discloses a chemical production self-adaptive control system for process quality detection, and relates to the technical field of production control, and the system comprises a fermentation record calling module which is used for calling process records of continuous multi-stage fermentation of target chemical raw materials through serial fermentation tanks; the first record data extraction module is used for extracting fermentation state data and control data in the fermentation record; the fermentation control evaluation module is used for introducing an evaluation function to analyze control data to obtain production fitness; the prediction model construction module is used for training a production fitness prediction model based on the fermentation state data and the fitness; the production fitness acquisition module is used for predicting a real-time fermentation state by using the model to obtain predicted production fitness; the control instruction generation module is used for sending out a control instruction if the predicted fitness does not meet the threshold value; and the fermentation production adjusting module adjusts the fermentation production process according to the control instruction. Therefore, the technical effects of improving regulation responsiveness and regulation efficiency and improving fermentation quality stability are achieved.

Description

Technical Field

[0001] The invention relates to the technical field of production control, and in particular to a chemical production adaptive control system for process quality detection. Background Art

[0002] In the chemical production process, the fermentation link is a key step in the production of many chemical products, and its quality control has an important impact on the performance and output of the final product. Traditional fermentation process control usually relies on manual experience or preset fixed parameters, and the parameters of the fermentation tank are monitored and adjusted. The main technical disadvantages are as follows: on the one hand, manual experience control lacks accurate grasp of the complex dynamic characteristics of the fermentation process, and it is difficult to adapt to changes in raw material characteristics and environmental conditions in real time, resulting in unstable fermentation quality; on the other hand, fixed parameter control cannot dynamically adjust fermentation parameters according to real-time monitoring data, and it is difficult to effectively respond to emergencies and abnormal fluctuations in the fermentation process. In addition, traditional control methods also have shortcomings in terms of control responsiveness and control efficiency, and cannot optimize and adjust the fermentation process in a timely and accurate manner, thus affecting the overall efficiency and product quality of chemical production. Summary of the invention

[0003] The present invention provides a chemical production adaptive control system for process quality detection to solve the technical problems of low regulation responsiveness and regulation efficiency and unstable fermentation quality in the prior art, and achieves the technical effect of improving regulation responsiveness and regulation efficiency and improving fermentation quality stability.

[0004] The present invention provides a chemical production adaptive control system for process quality detection, comprising: The fermentation record retrieval module is used to retrieve the fermentation record, wherein the fermentation record refers to the record of continuous multi-stage fermentation of the target chemical raw material through the series-connected fermentation tanks.

[0005] The first record data extraction module is used to extract the first record data in the fermentation record, where the first record data includes first fermentation state data and first fermentation control data.

[0006] The fermentation control evaluation module is used to introduce a fermentation control evaluation function to evaluate and analyze the first fermentation control data to obtain a first production fitness.

[0007] The prediction model building module is used to perform machine supervised learning on the first data group formed based on the first fermentation state data and the first production fitness to obtain a production fitness prediction model.

[0008] The production fitness acquisition module is used to predict and analyze the real-time fermentation state data of the target chemical raw material dynamically monitored through the production fitness prediction model to obtain the predicted real-time production fitness.

[0009] The control instruction generating module is used to issue a first adaptive control instruction if the predicted real-time production fitness does not meet a predetermined fitness threshold.

[0010] A fermentation production adjustment module is used to adjust the fermentation production of the target chemical raw material based on the first adaptive control instruction.

[0011] In a feasible implementation, the execution steps of the system include: A preheating tower-flash tank dual device is set up to preheat the chemical raw materials, and the waste liquid is heated by secondary steam to obtain preheated pre-produced materials.

[0012] The liquefaction plan is read, and the preheated and pre-produced materials are liquefied according to the liquefaction plan to obtain the target chemical raw materials, wherein the target chemical raw materials refer to the chemical raw materials after pretreatment.

[0013] In a feasible implementation, the expression of the fermentation control evaluation function is as follows: ; Wherein, F refers to the first production fitness, E refers to the first fermentation efficiency in the first fermentation control data, refers to the maximum possible value of the fermentation efficiency, P refers to the first ethanol purity in the first fermentation control data, refers to the maximum possible value of ethanol purity, C refers to the first fermentation energy consumption in the first fermentation control data, and are the weight coefficients of fermentation efficiency, ethanol purity, fermentation energy consumption and comprehensive efficiency-purity-energy consumption ratio, respectively, and .

[0014] In a feasible implementation, machine supervised learning is performed on the first data group formed based on the first fermentation state data and the first production fitness to obtain a production fitness prediction model, including: A first feature of the predetermined fermentation state features is traversed in the first fermentation state data to obtain a first state parameter.

[0015] When the first state parameter is at a first predetermined parameter threshold of the first feature, the first state parameter is added to the first data set.

[0016] The predetermined fermentation state characteristics include fermentation temperature, pH value, sugar concentration, yeast concentration and foreign bacteria rate.

[0017] In a feasible implementation, when the first state parameter is not within the first predetermined parameter threshold, an abnormal control warning is issued for the first feature.

[0018] In a feasible implementation, after predicting and analyzing the real-time fermentation state data of the target chemical raw material dynamically monitored by the production fitness prediction model to obtain the predicted real-time production fitness, the method further includes: If the predicted real-time production fitness meets the predetermined fitness threshold, the real-time distillation status data of the target chemical raw material is dynamically monitored.

[0019] The real-time distillation state data is traversed in the differential pressure distillation database to obtain the most similar distillation state data, and the most similar distillation control data corresponding to the most similar distillation state data is obtained.

[0020] The predetermined distillation control feature is read, and feature matching is performed on the most similar distillation control data based on the predetermined distillation control feature to obtain a distillation control feature parameter.

[0021] The distillation control characteristic parameters are normalized and weighted to obtain the distillation fitness.

[0022] If the distillation fitness does not meet the predetermined fitness threshold, a second adaptive control instruction is issued.

[0023] The differential pressure distillation of the target chemical raw material is adjusted based on the second adaptive control instruction.

[0024] In a feasible implementation, the real-time distillation state data is traversed in a differential pressure distillation database to obtain the most similar distillation state data, and the most similar distillation control data corresponding to the most similar distillation state data is obtained, including: Extract any data set in the differential pressure distillation database.

[0025] Vectorization processing is performed on the real-time distillation state data and any distillation state data in the arbitrary data group in sequence to obtain a real-time distillation state vector and an arbitrary distillation state vector, respectively.

[0026] An arbitrary similarity between the real-time distillation state vector and the arbitrary distillation state vector is calculated.

[0027] When any similarity meets the predetermined traversal constraint, any distillation state data is used as the most similar distillation state data.

[0028] Any distillation control data in the arbitrary data group is used as the most similar distillation control data.

[0029] In a feasible implementation, the predetermined distillation control characteristics include at least distillation efficiency, steam usage and ethanol purity.

[0030] The invention discloses a chemical production adaptive control system for process quality detection, comprising: a fermentation record retrieval module: used for retrieving fermentation records, wherein the records are process records of continuous multi-stage fermentation of target chemical raw materials by fermentation tanks in series; a first record data extraction module: extracting first record data in the fermentation record, including first fermentation state data and first fermentation control data; a fermentation control evaluation module: introducing a fermentation control evaluation function, evaluating and analyzing the first fermentation control data, and obtaining a first production fitness; a prediction model construction module: constructing a first data group based on the first fermentation state data and the first production fitness, and using a machine supervised learning method to train a production fitness prediction module Model; production fitness acquisition module: through the production fitness prediction model, the real-time fermentation status data of the dynamically monitored target chemical raw materials are predicted and analyzed to obtain the predicted real-time production fitness; control instruction generation module: if the predicted real-time production fitness does not meet the predetermined fitness threshold, a first adaptive control instruction is issued; fermentation production adjustment module: based on the first adaptive control instruction, the fermentation production process of the target chemical raw materials is adjusted. The chemical production adaptive control system for process quality detection disclosed in the present invention solves the technical problems of low control responsiveness and control efficiency and unstable fermentation quality, and achieves the technical effect of improving control responsiveness and control efficiency and improving fermentation quality stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a schematic diagram of the structure of a chemical production adaptive control system for process quality detection according to the present invention; Figure 2 The present invention is a schematic diagram of a process for obtaining the most similar distillation control data in a chemical production adaptive control system for process quality detection.

[0032] Explanation of the reference numerals: fermentation record retrieval module 11, first record data extraction module 12, fermentation control evaluation module 13, prediction model construction module 14, production fitness acquisition module 15, control instruction generation module 16, fermentation production adjustment module 17. DETAILED DESCRIPTION

[0033] The above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods of the specification to better understand the above technical solution. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments used only to explain the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In addition, it should be noted that, for the convenience of description, only the parts related to the present invention are shown in the drawings, rather than all of them.

[0034] Example, Figure 1 The present invention is a schematic diagram of a process quality detection process of a chemical production adaptive control system, wherein the system comprises: The fermentation record retrieval module 11 is used to retrieve the fermentation record, wherein the fermentation record refers to the record of continuous multi-stage fermentation of the target chemical raw material through the serially connected fermentation tanks.

[0035] Specifically, first, the data records related to the fermentation process are retrieved and obtained from the storage system of the production management end of the target scenario. The fermentation records refer to the data records generated by continuous multi-stage fermentation of target chemical raw materials through multiple fermentation tanks connected in series during the chemical production process, covering the key information of the fermentation process; exemplarily, they usually include control parameters, control operations, timestamps and other information during the fermentation process, such as changes in the temperature, pH value, stirring speed, sugar concentration, yeast concentration, etc. of the fermentation tank, which are used to provide basic data for subsequent analysis and processing.

[0036] By retrieving fermentation records, the system can obtain comprehensive and detailed fermentation process data, providing a rich information basis for subsequent fermentation control evaluation and prediction model construction.

[0037] The first record data extraction module 12 is used to extract the first record data in the fermentation record, where the first record data includes first fermentation state data and first fermentation control data.

[0038] Specifically, the first record data refers to the key data related to the fermentation process status and control operation selected from the retrieved fermentation records, such as the data with the highest frequency of working conditions, or the data with the best quality stability. Among them, the first fermentation status data refers to the parameter data reflecting the current operating status of the fermentation process, such as the temperature, pH value, sugar concentration, yeast concentration, etc. in the fermentation tank; the first fermentation control data refers to the operating parameters used to control the fermentation conditions during the fermentation process, such as stirring intensity, ventilation volume, type and amount of added nutrients, etc.

[0039] The purpose of extracting the first recorded data is to provide accurate data support for the evaluation and prediction of the fermentation process. By screening out key representative fermentation status data and control data, the system can more accurately evaluate the operation of the fermentation process, analyze the effectiveness of fermentation control operations, and provide high-quality input data for the subsequent construction of production fitness prediction models.

[0040] The fermentation control evaluation module 13 is used to introduce a fermentation control evaluation function to evaluate and analyze the first fermentation control data to obtain a first production fitness.

[0041] Specifically, the fermentation control evaluation function is a predefined function used to map the multi-dimensional and multi-dimensional parameters in the first fermentation control data into comprehensive and intuitive numerical evaluation values, namely, the first production fitness, so as to quantify the control effect of the fermentation process and facilitate comparison and judgment of the fermentation production levels under different conditions.

[0042] The above quantitative evaluation path not only improves the objectivity and accuracy of the fermentation process evaluation, but also provides high-quality training data for the subsequent prediction model construction. By calculating the first production fitness, the system can more accurately identify the advantages and disadvantages of the fermentation process, thereby providing a scientific basis for adaptive control.

[0043] In some embodiments, the expression of the fermentation control evaluation function is as follows: ; Wherein, F refers to the first production fitness, E refers to the first fermentation efficiency in the first fermentation control data, refers to the maximum possible value of the fermentation efficiency, P refers to the first ethanol purity in the first fermentation control data, refers to the maximum possible value of ethanol purity, C refers to the first fermentation energy consumption in the first fermentation control data, and are the weight coefficients of fermentation efficiency, ethanol purity, fermentation energy consumption and comprehensive efficiency-purity-energy consumption ratio, respectively, and .

[0044] Specifically, the fermentation control evaluation function includes multiple sub-items and corresponding weight coefficients for evaluating fermentation efficiency, product purity, fermentation energy consumption and energy consumption ratio efficiency, wherein the sum of the weight coefficients is 1, ensuring the balance of the evaluation function.

[0045] Specifically, For evaluating the fermentation efficiency of the first fermentation control data, wherein, is the maximum possible value of the fermentation efficiency, which can be set based on experimental data or theoretical analysis and represents the best fermentation efficiency. The closer the first fermentation efficiency in the first fermentation control data is to , it can be considered that the better the performance in terms of efficiency, The closer the value is to 1.

[0046] Specifically, Used to evaluate the product purity of the first fermentation control data (i.e., the purity of the ethanol produced), wherein: is the maximum possible value of ethanol purity. Similarly, based on process design or experimental settings, the closer the first ethanol purity in the first fermentation control data is to , it can be considered that the better the performance in terms of efficiency, The closer the value is to 1.

[0047] Specifically, The first fermentation energy consumption used to evaluate the first fermentation control data, that is, the evaluation of the absolute value of the energy consumption. The lower the energy consumption, the greater the value of this sub-item; It is an evaluation of the energy consumption efficiency of the first fermentation control data, that is, the energy consumption ratio efficiency. The higher the fermentation efficiency and product purity produced by unit energy consumption, the higher the overall efficiency of the fermentation process can be considered, and the larger the value of this sub-item.

[0048] By using the fermentation control evaluation function, the system can quantify the performance of each key fermentation control data and calculate the comprehensive fitness F based on the degree of deviation of each data; it ensures the balance of various process conditions and avoids the excessive influence of a single indicator on production results. At the same time, it provides accurate input data for predictive models and adaptive control, thereby improving the overall quality control level and production efficiency of the fermentation process.

[0049] The prediction model building module 14 is used to perform machine supervised learning on the first data set formed based on the first fermentation state data and the first production fitness to obtain a production fitness prediction model.

[0050] Specifically, by taking the first fermentation state data marked with the first production fitness as training data, a production fitness prediction model based on machine learning is trained to obtain a mapping relationship between the input data (first fermentation state data) and the output target (first production fitness), thereby having the ability to quickly predict the production fitness of the real-time fermentation process based on the input real-time fermentation state data, providing a basis for subsequent adaptive control, and helping to timely discover potential problems in the fermentation process and dynamically adjust the problems.

[0051] In some embodiments, the steps performed by the prediction model building module 14 include: A first feature in the predetermined fermentation state features is traversed in the first fermentation state data to obtain a first state parameter; when the first state parameter is at a first predetermined parameter threshold of the first feature, the first state parameter is added to the first data group; wherein the predetermined fermentation state features include fermentation temperature, pH value, sugar concentration, yeast concentration and foreign bacteria rate.

[0052] Specifically, the predetermined fermentation state characteristics refer to the key parameters pre-set during the fermentation process, such as fermentation temperature, pH value, sugar concentration, yeast concentration and foreign bacteria rate, etc. They are the core indicators for monitoring and controlling the fermentation process and have an important impact on the quality and efficiency of the fermentation process; the first characteristic is any specific parameter selected from the predetermined fermentation state characteristics.

[0053] Specifically, the first state parameter refers to a numerical value corresponding to the first feature extracted from the first fermentation state data, such as a fermentation temperature value or a pH value at a certain time point; the first predetermined parameter threshold is a reasonable range or standard value set for the first state parameter, which is used to determine whether the parameter meets the requirements of the fermentation process, so as to perform a preliminary screening of the traversed first state parameters.

[0054] Specifically, first, a first feature (such as fermentation temperature) is selected from the predetermined fermentation state features, and then the feature is traversed in the first fermentation state data to extract the corresponding first state parameters (such as the fermentation temperature value at each time point); then, it is determined whether these first state parameters are within the first predetermined parameter threshold range. If a first state parameter meets the threshold condition, it is added to the first data group, so as to screen out key state parameters that meet the requirements of the fermentation process, ensure that the data used to train the prediction model is high-quality and representative, and then improve the accuracy and reliability of the model, so that it can better predict production fitness, wherein the first data group is a basic data set for subsequent machine supervised learning, which contains fermentation state parameters that meet the requirements and the corresponding first production fitness.

[0055] In some embodiments, the execution step of the prediction model building module 14 also includes: when the first state parameter is not within the first predetermined parameter threshold, performing abnormal control warning on the first feature.

[0056] Specifically, if the first state parameter is not within the first predetermined parameter threshold, it means that the current first state parameter has significantly deviated from the preset normal range, and the system will issue an alarm or prompt to remind the operator or the automated control system to pay attention to the abnormal situation, so as to promptly discover and deal with potential problems, prevent the fermentation process from further deviating from the normal state, and avoid product quality degradation or reduced production efficiency.

[0057] Specifically, the abnormal control warning mechanism is triggered to carry out abnormal control warning, including generating an alarm signal (such as an audible alarm, a visual prompt or a system log record) and sending relevant information to an operator or an automated control system.

[0058] The production fitness acquisition module 15 is used to predict and analyze the real-time fermentation state data of the target chemical raw material monitored dynamically through the production fitness prediction model to obtain the predicted real-time production fitness.

[0059] Specifically, the fermentation status data of the target chemical raw materials are collected in real time and continuously during the fermentation process. The real-time fermentation status data reflects the real-time operation status of the fermentation process, including various key parameters in the fermentation process, such as fermentation temperature, pH value, sugar concentration, yeast concentration, etc.; further, the real-time fermentation status data is predicted and analyzed through the production fitness prediction model to obtain the predicted real-time production fitness, that is, the production fitness value predicted by the model according to the current fermentation status, which is used to evaluate whether the fermentation process is in an ideal state.

[0060] The purpose of this process is to apply the prediction model to the actual fermentation process to achieve real-time monitoring and evaluation of the fermentation process; through the predictive analysis of the model, the system can quickly determine whether the current fermentation process meets the expected production fitness requirements, thereby providing timely and accurate basis for subsequent adaptive control, helping to timely discover potential problems in the fermentation process and take corresponding adjustment measures to ensure the stability of the fermentation process and product quality.

[0061] In some embodiments, the execution steps of the production fitness acquisition module 15 include: If the predicted real-time production fitness meets the predetermined fitness threshold, dynamically monitor the real-time distillation state data of the target chemical raw material; traverse the real-time distillation state data in the differential pressure distillation database to obtain the most similar distillation state data, and obtain the most similar distillation control data corresponding to the most similar distillation state data; read the predetermined distillation control characteristics, and perform feature matching on the most similar distillation control data based on the predetermined distillation control characteristics to obtain distillation control feature parameters; perform normalized weighted calculation on the distillation control feature parameters to obtain distillation fitness; if the distillation fitness does not meet the predetermined fitness threshold, issue a second adaptive control instruction; and adjust the differential pressure distillation of the target chemical raw material based on the second adaptive control instruction.

[0062] Specifically, the differential pressure distillation database is a database that stores a large amount of historical distillation state data and corresponding control data, which is used to provide reference data; the most similar distillation state data refers to the data record in the above-mentioned differential pressure distillation database that is closest to the current real-time distillation state data; the predetermined distillation control characteristics refer to the key control parameters pre-set in the distillation process, such as distillation efficiency, steam consumption, ethanol purity, etc.; distillation adaptability is a comprehensive evaluation indicator used to measure whether the control effect of the current distillation process meets expectations.

[0063] Specifically, real-time distillation status data include: distillation temperature, which directly affects the separation efficiency of ethanol. Too high a temperature may lead to ethanol volatilization loss, while too low a temperature will affect the separation effect; distillation pressure. The pressure settings of different distillation towers in differential pressure distillation technology will affect the utilization efficiency of steam and the purity of ethanol; reflux ratio, which refers to the ratio of the amount of top condensate returned to the tower to the amount of top product produced. The size of the reflux ratio will affect the purity and output of ethanol; steam flow rate. The amount of steam flow rate directly affects the speed and efficiency of distillation; ethanol concentration. The change in ethanol concentration during the distillation process can reflect whether the distillation effect is good.

[0064] Specifically, when the predicted real-time production fitness meets the predetermined fitness threshold, the system dynamically monitors the real-time distillation state data of the target chemical raw material through a preset sensor network; then, the system traverses the differential pressure distillation database according to the acquired real-time distillation state data to find the distillation state data most similar to the current state, and obtains the corresponding most similar distillation control data; then, the predetermined distillation control feature is read, feature matching is performed on the most similar distillation control data, the distillation control feature parameters are extracted, and the extracted parameters are normalized and weighted to obtain the distillation fitness. If the distillation fitness does not meet the predetermined fitness threshold, a second adaptive control instruction is issued to adjust the differential pressure distillation process until the distillation fitness meets the predetermined fitness threshold.

[0065] Specifically, if the distillation fitness meets the predetermined fitness threshold, it can be considered that the current distillation control characteristics meet the needs and no adjustment is required. At this time, the system continuously monitors and collects the distillation process according to the predetermined monitoring granularity, and repeats the above-mentioned production fitness evaluation process.

[0066] The purpose of the above process is to combine the monitoring of the fermentation process with the monitoring of the distillation process, so as to realize comprehensive monitoring and dynamic adjustment of the entire chemical production process. By matching with historical data, it can be adjusted more accurately according to past successful operating conditions, avoiding the deviation caused by over-reliance on static parameter settings. At the same time, through the guidance of adaptive control instructions, the control efficiency of the distillation process is improved, which can significantly reduce energy consumption and improve product quality, ensuring that the entire production process remains coordinated and efficient in multiple process steps.

[0067] In some implementations, the predetermined distillation control characteristics include at least distillation efficiency, steam usage, and ethanol purity.

[0068] In some implementations, such as Figure 2 As shown, the real-time distillation state data is traversed in the differential pressure distillation database to obtain the most similar distillation state data, and the most similar distillation control data corresponding to the most similar distillation state data is obtained, including: Extract any data group from the differential pressure distillation database; perform vectorization processing on the real-time distillation state data and any distillation state data in the arbitrary data group in turn to obtain a real-time distillation state vector and an arbitrary distillation state vector respectively; calculate any similarity between the real-time distillation state vector and the arbitrary distillation state vector; when the arbitrary similarity meets the predetermined traversal constraint, use the arbitrary distillation state data as the most similar distillation state data; use any distillation control data in the arbitrary data group as the most similar distillation control data.

[0069] Specifically, first, any set of data is extracted from the differential pressure distillation database, including distillation state data and corresponding distillation control data; then, the distillation state data monitored in real time and any distillation state data in the database are respectively vectorized, that is, the distillation state data including multiple dimensional state features are converted into vectors in a multi-dimensional space, so as to obtain a real-time distillation state vector and an arbitrary distillation state vector, which are convenient for mathematical calculation and similarity comparison; then, the similarity between the two vectors is calculated. If the similarity meets the predetermined traversal constraint condition, it indicates that the similarity between the current real-time distillation state and a certain historical data group is high enough, and the group of historical data can be used as a reference for the current distillation state, and then the distillation state data in the database is marked as the most similar distillation state data, and the corresponding distillation control data is marked as the most similar distillation control data.

[0070] Specifically, similarity calculation can be performed based on, for example, Euclidean distance, cosine similarity, etc., thereby providing a quantitative basis for determining which historical data best represents the current distillation process.

[0071] The purpose of the above process is to find the historical data most similar to the current distillation state through data matching, so as to provide a reference for subsequent distillation control. In this way, the system can draw on historical experience, optimize the current distillation control strategy, and improve the stability and efficiency of the distillation process.

[0072] The control instruction generating module 16 is configured to issue a first adaptive control instruction if the predicted real-time production fitness does not meet a predetermined fitness threshold.

[0073] Specifically, if the predicted real-time production fitness does not meet the predetermined fitness threshold, it can be considered that the current production process is at risk and needs to be regulated, and the first adaptive control instruction is further generated to dynamically adjust the fermentation process to optimize the fermentation conditions.

[0074] The fermentation production adjustment module 17 is used to adjust the fermentation production of the target chemical raw material based on the first adaptive control instruction.

[0075] Exemplarily, the fermentation production of the target chemical raw material is adjusted. First, the system determines the fermentation parameters that need to be adjusted (such as temperature, pH value, stirring speed, etc.) according to the first adaptive control instruction; then, combined with the optimization algorithm (such as genetic algorithm, particle swarm optimization algorithm, etc.), under the guidance of the production fitness prediction model (i.e., with the production fitness prediction model as the cost function), the previously determined fermentation parameters are iteratively optimized and adjusted, wherein the optimization algorithm dynamically adjusts the fermentation parameters according to the output of the prediction model (i.e., the predicted production fitness) to find the optimal parameter combination.

[0076] The purpose of the above process is to optimize the fermentation process by dynamically adjusting the fermentation parameters to ensure the stability and efficiency of the fermentation process. Among them, combined with the optimization algorithm and the production fitness prediction model, the system can reduce manual intervention, improve the level of production automation, adjust the fermentation parameters more accurately, improve fermentation efficiency, reduce energy consumption, and improve product quality.

[0077] In some embodiments, the system performs the following steps: A preheating tower-flash tank dual device is set up to preheat the chemical raw materials, and the waste liquid is heated by secondary steam to obtain preheated pre-produced materials; the liquefaction plan is read, and the preheated pre-produced materials are liquefied according to the liquefaction plan to obtain the target chemical raw materials, wherein the target chemical raw materials refer to the chemical raw materials after pretreatment.

[0078] Specifically, the preheating tower-flash tank double device is a combination of equipment used to preheat chemical raw materials. Through the synergistic effect of the preheating tower and the flash tank, the raw materials are heated with secondary steam to improve energy utilization efficiency; the preheated chemical raw materials are called preheated pre-production materials.

[0079] Specifically, the liquefaction plan refers to a specific plan prepared in advance for liquefying preheated and pre-produced materials, including parameters such as liquefaction method, temperature, and enzyme addition amount. Exemplarily, the liquefaction plan adopts a one-time enzyme addition and one-time injection liquefaction method, and improves the liquefaction efficiency by adjusting the liquefaction tank capacity and liquefaction temperature (above 87°C), thereby solving the problems of low enzyme activity and difficult liquefaction.

[0080] The above steps, by optimizing the pretreatment steps, can improve the liquefaction efficiency and quality of chemical raw materials, provide high-quality raw materials for the subsequent fermentation process, and ensure that the raw materials are in good condition when entering the fermentation stage, thereby improving the fermentation efficiency and product quality.

[0081] In summary, the chemical production adaptive control system for process quality detection provided by the present invention has the following technical effects: First, by introducing the production fitness prediction model constructed by machine supervised learning, it is possible to quickly and accurately predict and analyze the real-time fermentation status data, thereby realizing real-time monitoring and dynamic adjustment of the fermentation process. The above data-driven predictive analysis can effectively overcome the regulation lag problem caused by insufficient manual experience or unreasonable fixed parameter settings in traditional control methods, and significantly improve the regulation responsiveness; secondly, the adaptive control system can dynamically generate control instructions according to the prediction results, adjust the fermentation parameters in time, and optimize the fermentation process, thereby improving the regulation efficiency and ensuring the stability of the fermentation process and the reliability of product quality; at the same time, it can also adapt to the characteristics of different batches of raw materials and changes in the fermentation environment, has strong versatility and adaptability, and provides strong support for the intelligent control of chemical production processes. In general, the present invention achieves the technical effects of improving the regulation responsiveness and efficiency and improving the stability of fermentation quality.

[0082] It should be understood that the embodiments disclosed in the present invention and the above description can enable those skilled in the art to use the present invention to implement the present invention. At the same time, the present invention is not limited to the above-mentioned embodiments. It should be understood that those skilled in the art can still modify the technical solutions recorded in the above-mentioned embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A chemical production adaptive control system for process quality detection, characterized in that: include: A fermentation record retrieval module, used to retrieve fermentation records, wherein the fermentation records refer to records of continuous multi-stage fermentation of target chemical raw materials through serially connected fermentation tanks; A first record data extraction module, used to extract first record data in the fermentation record, wherein the first record data includes first fermentation state data and first fermentation control data; A fermentation control evaluation module, used for introducing a fermentation control evaluation function to evaluate and analyze the first fermentation control data to obtain a first production fitness; A prediction model building module, used for performing machine supervised learning on a first data set formed based on the first fermentation state data and the first production fitness to obtain a production fitness prediction model; A production fitness acquisition module is used to predict and analyze the real-time fermentation state data of the target chemical raw material dynamically monitored through the production fitness prediction model to obtain a predicted real-time production fitness; A control instruction generating module, configured to issue a first adaptive control instruction if the predicted real-time production fitness does not meet a predetermined fitness threshold; A fermentation production adjustment module is used to adjust the fermentation production of the target chemical raw material based on the first adaptive control instruction.

2. According to the chemical production adaptive control system for process quality detection according to claim 1, it is characterized in that: include: A preheating tower-flash tank dual device is set up to preheat the chemical raw materials, and the waste liquid is heated by secondary steam to obtain preheated pre-production materials; The liquefaction plan is read, and the preheated and pre-produced materials are liquefied according to the liquefaction plan to obtain the target chemical raw materials, wherein the target chemical raw materials refer to the chemical raw materials after pretreatment.

3. According to the chemical production adaptive control system for process quality detection according to claim 1, it is characterized in that: The expression of the fermentation control evaluation function is as follows: ； Wherein, F refers to the first production fitness, E refers to the first fermentation efficiency in the first fermentation control data, refers to the maximum possible value of the fermentation efficiency, P refers to the first ethanol purity in the first fermentation control data, refers to the maximum possible value of ethanol purity, C refers to the first fermentation energy consumption in the first fermentation control data, and are the weight coefficients of fermentation efficiency, ethanol purity, fermentation energy consumption and comprehensive efficiency-purity-energy consumption ratio, respectively, and .

4. According to the chemical production adaptive control system for process quality detection of claim 1, it is characterized in that: Performing machine supervised learning on a first data set formed based on the first fermentation state data and the first production fitness to obtain a production fitness prediction model, including: Traversing a first feature of the predetermined fermentation state features in the first fermentation state data to obtain a first state parameter; When the first state parameter is within a first predetermined parameter threshold of the first feature, adding the first state parameter to the first data set; The predetermined fermentation state characteristics include fermentation temperature, pH value, sugar concentration, yeast concentration and foreign bacteria rate.

5. According to the chemical production adaptive control system for process quality detection according to claim 4, it is characterized in that: When the first state parameter is not within the first predetermined parameter threshold, an abnormal control warning is issued for the first feature.

6. According to the chemical production adaptive control system for process quality detection of claim 1, it is characterized in that: After predicting and analyzing the real-time fermentation state data of the target chemical raw material monitored dynamically through the production fitness prediction model to obtain the predicted real-time production fitness, the method further includes: If the predicted real-time production fitness meets the predetermined fitness threshold, dynamically monitoring the real-time distillation state data of the target chemical raw material; Traversing the real-time distillation state data in a differential pressure distillation database to obtain the most similar distillation state data, and acquiring the most similar distillation control data corresponding to the most similar distillation state data; Reading a predetermined distillation control feature, and performing feature matching on the most similar distillation control data based on the predetermined distillation control feature to obtain a distillation control feature parameter; Performing normalized weighted calculation on the distillation control characteristic parameters to obtain distillation fitness; If the distillation fitness does not meet the predetermined fitness threshold, issuing a second adaptive control instruction; The differential pressure distillation of the target chemical raw material is adjusted based on the second adaptive control instruction.

7. A chemical production adaptive control system for process quality detection according to claim 6, characterized in that: Traversing the real-time distillation state data in the differential pressure distillation database to obtain the most similar distillation state data, and obtaining the most similar distillation control data corresponding to the most similar distillation state data, including: Extracting any data set in the differential pressure distillation database; Sequentially performing vectorization processing on the real-time distillation state data and any distillation state data in the arbitrary data group to obtain a real-time distillation state vector and an arbitrary distillation state vector respectively; Calculate and obtain any similarity between the real-time distillation state vector and the arbitrary distillation state vector; When any similarity meets the predetermined traversal constraint, taking any distillation state data as the most similar distillation state data; Any distillation control data in the arbitrary data group is used as the most similar distillation control data.

8. According to claim 6, a chemical production adaptive control system for process quality detection is characterized in that: The predetermined distillation control characteristics include at least distillation efficiency, steam usage and ethanol purity.

Images