Intelligent control method and system for bacon production smoking house

Abstract

This invention belongs to the field of intelligent control technology, specifically an intelligent control method and system for smoking rooms in cured meat production. The method includes: collecting multi-dimensional and coupled interference data from the smoking room; constructing a basic data twin through dual-weighting and calibration; extracting parameters to build a process gene library and generate process curve gene templates; decomposing the data into individual smoking room templates; designing a control and compensation mechanism through a four-level deviation judgment; and collecting deviation gene datasets; analyzing the datasets to locate causative genes, optimizing the gene library, and updating the templates; building a sensory mapping model to quantify manual evaluation and convert experience into parameters, generating flavor customization instructions; and compiling a process gene ledger to achieve self-evolution of the gene library and update the templates, forming a closed-loop iterative process. This invention constructs an intelligent control system for smoking rooms through digital twins and process genes, achieving flavor customization and process self-evolution, ensuring stable cured meat quality, meeting diverse needs, and promoting the intelligent upgrading of the industry.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control technology, specifically an intelligent control method and system for smoking rooms used in the production of cured meat. Background Technology

[0002] Smoking cured meat is a typical experience-based food processing technique. Traditional smoking room control relies heavily on manual experience. Even existing simple intelligent control solutions only achieve simple collection and closed-loop control of core parameters of a single smoking room, ignoring coupling interferences such as temperature field superposition, smoke flow mutual interference, and air pressure linkage in the smoking room. Furthermore, the data collection does not consider the attenuation of equipment accuracy and lacks the ability to calibrate the triple data distortion of equipment, environment, and transmission. The collected data is isolated and has a high distortion rate, which cannot accurately reflect the real operating status of the smoking room group. This leads to a disconnect between parameter control and actual production, resulting in large quality differences between batches of cured meat.

[0003] Existing control methods mostly use fixed thresholds for process parameters, lacking a dynamic adaptation system. This makes it impossible to differentiate control based on the equipment precision, spatial layout, and historical operating characteristics of different smokehouses, and also lacks targeted parameter matching for raw material types and production environments. Furthermore, parameter deviations during production are only adjusted with simple feedback; a systematic collection and causal analysis system for deviation data has not been established, making it impossible to pinpoint the core root cause of deviations. This easily leads to secondary deviations, resulting in insufficient stability and adaptability of process control. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, this invention proposes an intelligent control method for smoking chambers used in the production of cured meat. This invention primarily addresses the problems of traditional smoking chambers lacking precise data and self-optimization, being decoupled and lacking iteration, and being difficult to maintain stable quality and customization.

[0005] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows: The present invention provides an intelligent control method for smoking rooms in the production of cured meat, including: S1: collecting multi-dimensional detection and coupling interference data of each smoking room, assigning dual weights in combination with the equipment accuracy attenuation coefficient, generating a multi-label accurate detection dataset, and constructing a basic data twin of the smoking room group.

[0006] S2: Extract tagged parameters from the basic data twin of the smokehouse group and input them into the process gene library. Set thresholds for clustering, transform meat sampling data into quality digital fingerprints, bind parameters to dual trigger indicators, and generate process curve gene templates.

[0007] S3: The process curve gene template is split into a single smokehouse-specific sub-template. Real-time parameters and coupling interference data are collected to determine the four-level deviation. A graded regulation and compensation mechanism is designed, and the deviation data with multiple labels is retained to obtain the deviation gene dataset.

[0008] S4: Classify and statistically analyze the deviation gene dataset, combine it with the basic data of the fumigation chamber group and the full data of the twins to perform a four-dimensional association analysis, locate the core causative genes, optimize the threshold of the process gene library, and update the gene template of the process curve.

[0009] S5: Based on the process gene library, a three-level gene mapping model for sensory indicators is built, which transforms human sensory evaluation into quantitative characteristic factors, sets flavor gene ranges, transforms human experience into parameter increment gene values, and generates flavor customization fine-tuning instructions.

[0010] S6: Organize the process gene library and flavor customization fine-tuning instructions into a standardized process gene ledger, improve the digital twin of the whole chain, screen high-quality genes to build a library and achieve self-evolution, and update the process curve gene template to form an iterative closed loop.

[0011] According to the intelligent control method for smoking rooms in cured meat production provided by the present invention, the specific steps for constructing a basic data twin of the smoking room group in step S1 are as follows:

[0012] S11: Collect multi-dimensional detection data from sensors and zone instruments in each fumigation chamber, collect coupling interference data generated by the interaction between the fumigation chambers, and form full raw data.

[0013] S12: Based on the full amount of original data and the accuracy attenuation coefficient of each detection device, weights are assigned to the detection data and the coupling interference data respectively, forming a dual-weight data system.

[0014] S13: Perform three-level calibration on the dual-weighted data system to eliminate the triple data distortion problem caused by equipment error, environmental interference, and data transmission, and obtain three-level calibration data.

[0015] S14: Based on the three-level calibration data, label the data with fumigation room number, detection dimension, and collection time label to generate a precise detection dataset.

[0016] S15: Map the precise detection dataset to the physical state, data associations and coupling relationships of each fumigation chamber, and construct a basic data twin of the fumigation chamber group in the digital space that corresponds one-to-one with the physical fumigation chamber group.

[0017] According to the intelligent control method for smoking rooms in cured meat production provided by the present invention, the specific steps for generating the process curve gene template in step S2 are as follows:

[0018] S21: Based on the digital base of the basic data twin of the fumigation room group, screen the core control parameters of the entire fumigation process, clean and organize the data, and form a standardized parameter dataset.

[0019] S22: Based on the standardized parameter dataset, classify and enter the smoking process steps into the process gene library, build a parameter classification storage and retrieval system, perform secondary cluster analysis according to gene characteristics, and preset the process parameter control range.

[0020] S23: Within the range of process parameter control, collect sample data of meat products during the smoking process, and transform the quality characteristics into quantifiable digital fingerprints through digital analysis of quality testing indicators.

[0021] S24: Associate and match the process parameter control range with the quality digital fingerprint to form a dual trigger index. Combine the historical high-quality parameter sequences in the process gene library to fit the parameter change pattern and generate a process curve gene template.

[0022] According to the intelligent control method for smoking rooms used in cured meat production provided by the present invention, the specific steps for forming the dual-trigger index in step S24 are as follows:

[0023] Based on the range of process parameter control and the digital fingerprint of meat quality, the dimensional hierarchy of the range of process parameter control is decomposed, the core quantitative features of the digital fingerprint of quality are deconstructed, and a mapping framework between parameter dimensions and quality characteristics is established.

[0024] Based on the mapping framework, the upper and lower limits and the optimal range of process parameter control are associated one by one with the feature thresholds of the quality digital fingerprint, forming a basic correlation matrix between parameters and quality.

[0025] The validity of the basic correlation matrix is ​​verified, invalid correlation items are eliminated, the correspondence of deviations is corrected, the linkage between parameter changes and quality characteristic fluctuations is strengthened, and parameter-quality correlation rules are formed.

[0026] Based on the parameter-quality correlation rules, a two-way triggering logic is built, setting a quality early warning threshold when the parameter exceeds the control range and a parameter adjustment threshold when the quality fingerprint deviates from the standard, thus forming a dual-trigger indicator.

[0027] According to the intelligent control method for smoking rooms in cured meat production provided by the present invention, the specific steps for obtaining the deviation gene dataset in step S3 are as follows:

[0028] S31: Based on the process curve gene template and the equipment precision, spatial layout, and historical operating characteristics of each individual fumigation chamber, a unique sub-template adapted to each fumigation chamber is decomposed.

[0029] S32: Real-time collection of process parameter data and coupling interference data of each fumigation chamber, comparison with the preset parameters of the single fumigation chamber's dedicated sub-template according to the four-level deviation judgment standard, identification of parameter deviations in each dimension, and obtaining the four-level deviation judgment result.

[0030] S33: Based on the four-level deviation judgment results, design differentiated graded control strategies according to the deviation level, and establish a coupled interference deviation linkage compensation mechanism.

[0031] S34: Collect all deviation data labeled with fumigation chamber number, deviation level, interference type, and regulation method during the compensation process of the coupling interference deviation linkage compensation mechanism, standardize and organize the data to obtain the deviation gene dataset.

[0032] According to the intelligent control method for smoking rooms in cured meat production provided by the present invention, the specific steps of optimizing the process gene library threshold and updating the process curve gene template in step S4 are as follows:

[0033] S41: Perform multi-dimensional classification statistics based on the deviation gene dataset, summarize the frequency of deviation occurrence, degree of impact and regulatory effect data of each dimension, and form deviation statistical results.

[0034] S42: Based on the deviation statistics, retrieve the full data of the twin of the basic data of the fumigation chamber group, and conduct correlation analysis from four dimensions: process parameters, equipment status, environmental conditions, and coupling interference. Explore the inherent correlation between deviation and data in each dimension, and identify the key influencing factors that cause deviation.

[0035] S43: Based on the key influencing factors, locate the core causative gene that causes process deviation, determine the parameter threshold deviation problem corresponding to the core causative gene, and form a threshold optimization scheme.

[0036] S44: Adjust the threshold range and optimal range in the process gene library according to the threshold optimization scheme, and refit the parameter change law of the entire fumigation process based on the full data of the optimized process gene library, and iteratively update the process curve gene template.

[0037] According to the intelligent control method for a smoking room in cured meat production provided by the present invention, the specific steps for generating flavor customization fine-tuning instructions in step S5 are as follows:

[0038] S51: Extract sensory index parameters, quality digital fingerprints, and deviation gene data of meat products from the process gene library, and build a three-level gene mapping model between sensory indicators and process genes.

[0039] S52: Collect artificial sensory evaluation data based on the three-level gene mapping model, and convert the evaluators' descriptions of color depth, flavor intensity, and tenderness into quantifiable sensory characteristic factors through model analysis.

[0040] S53: Based on the sensory quantitative characteristic factors and the parameter range of high-quality flavors in the process gene library, set the flavor gene range corresponding to different flavor requirements, determine the core parameter range and characteristic factor threshold of the flavor range, and define the flavor customization boundary.

[0041] S54: Based on the flavor customization boundary, summarize the experience of optimizing flavor in human sensory evaluation, and transform human experience into incremental gene values ​​of process parameters through a three-level gene mapping model to generate flavor customization fine-tuning instructions.

[0042] According to the intelligent control method for smoking rooms in cured meat production provided by the present invention, the specific steps for building a three-level gene mapping model in step S51 are as follows:

[0043] Extract process parameters, quality digital fingerprints, and deviation gene data from the process gene library, remove outliers, and unify dimensions to obtain standard sensory correlation data.

[0044] Based on sensory correlation data, the evaluation points of the three major sensory dimensions of color, flavor, and taste are decomposed, the correspondence between process genes, quality fingerprints, and deviation data is determined, and a basic mapping framework is built.

[0045] Based on the mapping framework, a three-level hierarchical structure of process genes, quantitative features, and sensory indicators is constructed. The mapping logic, quantification standards, and analysis rules of each level are determined, and a three-level gene-based mapping model between sensory indicators and process genes is built.

[0046] According to the intelligent control method for smoking rooms in cured meat production provided by the present invention, the specific steps in step S6 of updating the process curve gene template to form an iterative closed loop are as follows:

[0047] S61: Classify and organize the full data of the process gene library and the generated flavor customization fine-tuning instructions according to process links, gene types and instructions, unify data formats, labeling standards and traceability information, and form a process gene ledger.

[0048] S62: Integrate the data in the process gene ledger into the full-chain digital twin of the smokehouse group, supplement sensory genes and flavor customization, and obtain a complete full-chain digital twin.

[0049] S63: Based on the comprehensive digital twin screening of high-quality process genes, a gene self-evolution algorithm is built to obtain the gene pool self-evolution results.

[0050] S64: Based on the self-evolution results of the gene bank and the latest data in the standardized process gene ledger, refit the parameter change pattern of the entire fumigation process, iteratively update the original process curve gene template, and form an iterative closed loop.

[0051] This invention also provides an intelligent control system for a smoking room used in the production of cured meat, comprising:

[0052] The twin-based foundation module is used to collect multi-dimensional detection and coupling interference data from each fumigation chamber. It assigns dual weights based on the equipment accuracy attenuation coefficient, generates a multi-labeled accurate detection dataset, and constructs a twin of basic data for the fumigation chamber group.

[0053] The gene modeling module is used to extract tagged parameters from the twin data of the smokehouse group and input them into the process gene library. It sets thresholds for clustering, transforms meat sampling data into quality digital fingerprints, binds parameters into dual trigger indicators, and generates process curve gene templates.

[0054] The deviation control module is used to split the process curve gene template into a single smokehouse-specific sub-template, collect real-time parameters and coupling interference data to determine the deviation at four levels, design a graded regulation and compensation mechanism, and retain the deviation data with multiple labels to obtain the deviation gene dataset.

[0055] The gene optimization module is used to classify and statistically analyze the deviation gene dataset, combine it with the basic data of the smokehouse group and the full data of twins to perform four-dimensional correlation analysis, locate the core causative genes, optimize the threshold of the process gene library, and update the gene template of the process curve.

[0056] The flavor customization module is used to build a three-level gene mapping model of sensory indicators based on the process gene library, transform human sensory evaluation into quantitative characteristic factors, set flavor gene ranges, transform human experience into parameter increment gene values, and generate flavor customization fine-tuning instructions.

[0057] The gene iteration module is used to organize the process gene library and flavor customization fine-tuning instructions into a standardized process gene ledger, improve the digital twin of the whole chain, screen high-quality genes to build a library and achieve self-evolution, and update the process curve gene template to form an iterative closed loop.

[0058] This invention provides an intelligent control method for smoking rooms used in the production of cured meat. The beneficial effects of this invention are as follows:

[0059] 1. This invention achieves precise control of the smoking process by constructing a full-dimensional data calibration and digital twin. By collecting sensor and instrument detection data, as well as cross-smoking chamber coupling interference data, and assigning dual weights based on equipment accuracy attenuation coefficients, a three-level calibration eliminates triple distortions from equipment, environment, and transmission. This generates a precise detection dataset and constructs a basic data twin of the smoking chamber group, achieving a precise digital replication of the physical smoking chamber group. This allows for real-time and precise traceability of core parameters such as temperature, humidity, and smoke concentration during the smoking process. Simultaneously, it restores the coupling interference relationships within the smoking chambers, solving the problem of uneven smoking quality caused by inaccurate data and neglect of cross-equipment interference in traditional control methods. It ensures that the control of each process parameter is based on real and comprehensive operational data, guaranteeing the accuracy of the smoking process from the data source and significantly reducing quality defects caused by parameter deviations.

[0060] 2. This invention achieves dynamic adaptation of the process system through the construction of a process gene library and end-to-end self-optimization. Smoking process parameters are transformed into clusterable and iterative process genes. Secondary clustering is used to preset parameter thresholds, combined with meat quality digital fingerprints to form dual-trigger indicators, generating process curve gene templates. Specific sub-templates are then decomposed for the characteristics of individual smoking rooms. Simultaneously, a deviation gene dataset is collected through four-level deviation judgment, and core causative genes are located through four-dimensional correlation analysis. The process gene library thresholds are optimized and the templates are updated. Furthermore, based on sensory evaluation, human experience is transformed into incremental parameter gene values, achieving three-level self-optimization of the process template. This system transforms process parameters from fixed values ​​into gene sequences that can be dynamically adjusted according to equipment status, raw material characteristics, and quality feedback. It adapts to equipment differences between different smoking rooms and cross-smoking room coupling interference, while also digitizing and standardizing human optimization experience and integrating it into the process system. This transforms the process from fixed execution to dynamic adaptation, ensuring stable and consistent quality of cured meat from different batches and smoking rooms.

[0061] 3. This invention achieves diversified and long-term upgrades in cured meat production through flavor gene customization and gene pool self-evolution. The method overcomes the pain points of traditional cured meat production, such as the lack of flavor diversity and the difficulty in passing on manual fine-tuning experience. By constructing a three-level gene mapping model of sensory indicators, it transforms manual sensory evaluation into quantifiable characteristic factors, sets flavor gene ranges for different flavor requirements, and converts manual fine-tuning experience into executable parameter increment gene values, generating differentiated flavor customization fine-tuning instructions. This enables batch customization of cured meat flavors, meeting diversified market demands. Simultaneously, by collecting full-chain data to form a process gene ledger and improving the full-chain digital twin, it selects high-quality process genes to achieve gene pool self-evolution, forming an iterative closed loop of data collection, model optimization, regulation execution, gene evolution, and template updating. This allows the process system to continuously accumulate high-quality experience and eliminate invalid parameters in production practice, achieving long-term self-upgrading of process capabilities. This transforms cured meat production from traditional experience-based production to data-driven intelligent production, laying a core foundation for the large-scale, standardized, and intelligent development of the industry. Attached Figure Description

[0062] The invention will now be further described with reference to the accompanying drawings.

[0063] Figure 1 This is a flowchart illustrating the steps of an intelligent control method for a smoking room used in the production of cured meat, provided by an embodiment of the present invention.

[0064] Figure 2 This is a flowchart of an intelligent control method for a smoking room used in the production of cured meat, provided by an embodiment of the present invention;

[0065] Figure 3 This is a module diagram of an intelligent control system for a smoking room used in the production of cured meat, provided in an embodiment of the present invention. Detailed Implementation

[0066] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below according to specific embodiments.

[0067] like Figures 1 to 3 As shown in the embodiment of the present invention, an intelligent control method for a smoking room used in the production of cured meat is provided. The method includes:

[0068] S1: Collect multi-dimensional sensor and zone instrument detection data and coupling interference data in each fumigation chamber, assign dual weights based on the equipment accuracy attenuation coefficient, eliminate triple data distortion through three-level calibration, generate a multi-label accurate detection dataset, and construct a basic data twin of the fumigation chamber group.

[0069] S11: Collect multi-dimensional detection data from sensors and zone instruments in each fumigation chamber, collect coupling interference data generated by the interaction between the fumigation chambers, and form full raw data.

[0070] Real-time sampling data from key sensors in each smoking chamber, including temperature and humidity, smoke concentration, furnace temperature, internal air pressure, and wind speed, are collected. Simultaneously, operational data from instruments controlling zone temperature, smoke volume, and ventilation are also collected, with data structured and recorded in second-level time sequences. A cross-smoking chamber data acquisition module captures coupled interference data generated during the operation of adjacent and same-area smoking chambers, such as temperature field superposition, smoke flow interference, and air pressure linkage, and collects these data categorized according to smoking chamber-related pairing relationships. Finally, the independent detection data from each smoking chamber is integrated with the cross-smoking chamber coupled interference data, establishing a unified data framework based on the data source, data type, and time sequence dimension, forming a complete set of raw data with basic identifiers.

[0071] S12: Based on the full set of raw data and the accuracy attenuation coefficients of each testing device, weights are assigned to the testing data and coupling interference data respectively, forming a dual-weighted data system. The entire lifecycle information of the acquisition devices corresponding to the full set of raw data is analyzed, and combined with the equipment's service life, historical calibration records, and on-site operational wear and tear, the real-time accuracy attenuation coefficients of each sensor and instrument are calculated using an equipment accuracy attenuation algorithm, establishing a precise ledger with a one-to-one correspondence between equipment and coefficients. Then, based on the acquisition source identifier of the full set of raw data, the testing data of each fumigation chamber sensor and instrument is matched with the corresponding equipment accuracy attenuation coefficient, assigning basic data weights. The coupling interference data is matched with the average accuracy attenuation of related fumigation chamber equipment, assigning correlation correction weights. Finally, the weighted testing data and coupling interference data are fused according to data dimensions and acquisition time sequence to establish a dual-weighted data system that reflects the true validity of the data, achieving a precise match between data weights and the actual testing capabilities of the equipment.

[0072] S13: A three-level calibration process is performed on the dual-weighted data system to eliminate the triple data distortion problem caused by equipment errors, environmental interference, and data transmission, resulting in three-level calibration data. First, a first-level calibration is performed on the dual-weighted data system. The factory reference values ​​and recent calibration reference values ​​of each testing device are retrieved. A reference value deviation compensation algorithm is used to eliminate inherent hardware errors of sensors and instruments, as well as accuracy drift errors after long-term use. Second, a second-level calibration is performed based on the data from the first-level calibration. Environmental parameters such as temperature, humidity, atmospheric pressure, and dust concentration during data acquisition are considered. An environmental interference correction model is used to correct real-time interference deviations caused by the external environment. Finally, a third-level calibration is performed. Relying on error monitoring data from the data transmission link, methods such as packet loss compensation, signal offset correction, and transmission delay calibration are used to eliminate data distortion during transmission from the acquisition end to the terminal, ultimately obtaining three-level calibration data without triple distortion.

[0073] S14: Based on the Level 3 calibration data, label the data with fumigation chamber number, detection dimension, and collection time label to generate a precise detection dataset. Using the Level 3 calibration data as a foundation, assign a unique fumigation chamber number label to each data point based on the physical fumigation chamber from which the data was collected, determining the physical unit to which each data set belongs and achieving precise binding between the data and the physical fumigation chamber. Then, label each Level 3 calibration data point with a detection dimension label based on detection index types such as temperature and humidity, smoke concentration, furnace temperature, and air pressure, clearly defining the data attributes for easier subsequent data classification and extraction. Finally, label each data point with a collection time label based on the specific year, month, day, hour, minute, and second of the actual data collection, achieving time-series identification of the data. Integrate the multi-label information with the Level 3 calibration data, organize and store it in a structured format, generating a traceable, categorizable, and time-series precise detection dataset.

[0074] S15: Map the precise detection dataset to the physical state, data associations, and coupling relationships of each smoking room, constructing a basic data twin of the smoking room group in the digital space that corresponds one-to-one with the physical smoking room group. Based on the precise detection dataset, extract data such as physical operating parameters, real-time equipment conditions, and core detection indicators for each smoking room. According to the physical structure and operating logic of the physical smoking rooms, construct an independent digital parameter mapping layer for each smoking room to achieve precise digital replication of the physical state of the physical smoking rooms. Further explore the correlation patterns between detection indicators within each smoking room in the precise detection dataset, as well as the coupling and interference relationships between cross-smoking room data. Through data modeling, construct a logical mapping layer for data association and coupling relationships to restore the overall operating data logic of the smoking room group. Finally, merge the digital parameter mapping layers of each smoking room with the logical mapping layer of the smoking room group to build a real-time data interaction channel between the digital space and the physical smoking room group, achieving parameter synchronization and status linkage between the digital layer and the physical layer, and ultimately constructing a basic data twin of the smoking room group that corresponds one-to-one with the physical smoking room group.

[0075] S2: Extract tagged parameter data from the basic data twin of the smokehouse group and input it into the process gene library. Based on the genetic characteristics of raw materials and environment, perform secondary clustering with preset threshold ranges, transform meat sampling data into quality digital fingerprints, bind real-time parameters to form dual-trigger indicators, and generate process curve gene templates with multiple tags.

[0076] S21: Based on the digital foundation of the fumigation chamber cluster's basic data twin, core control parameters for the entire fumigation process are selected, and data cleaning and organization are performed to form a standardized parameter dataset. Relying on the digital foundation of the fumigation chamber cluster's basic data twin, focusing on the entire fumigation process of preheating, fumigation, and simmering, core control parameters such as temperature and humidity, smoke concentration, furnace temperature, and fumigation time are selected, and all raw parameter data with labels for fumigation chamber, detection dimension, and collection time sequence are extracted. Multi-dimensional cleaning is performed on the extracted raw data, using the 3σ principle to remove outliers, interpolation to fill in missing values, and standardizing dimensions and data according to industry standards. The cleaned standardized data is then structured according to process stage and detection indicators, and data attribution and attributes are labeled to form a standardized parameter dataset with multiple labels.

[0077] S22: Based on the standardized parameter dataset, the smoking process steps are categorized and entered into the process gene library. A parameter classification, storage, and retrieval system is established, and secondary cluster analysis is performed based on gene characteristics. The control range of process parameters is preset. Based on the process stages and indicator attributes of the standardized parameter dataset, process steps are categorized according to smoking preheating temperature control, smoking smoke concentration adjustment, and simmering temperature control, and each step is entered into the process gene library for systematic collection. A multi-level parameter classification storage module and a multi-dimensional retrieval system are built in the gene library, supporting rapid parameter retrieval by raw materials, environment, and process step, achieving standardized parameter management. Based on the full parameter data in the process gene library, secondary cluster analysis is performed according to the gene characteristics of raw material category and environmental temperature and humidity. The K-means algorithm is used to calculate cluster centers and boundaries, and the upper and lower limits and optimal intervals of process parameters under each cluster dimension are preset to define the reasonable control range of process parameters.

[0078] S23: Within the controlled range of process parameters, collect sample data of meat products during the smoking process. Through digital analysis of quality testing indicators, transform quality characteristics into quantifiable digital fingerprints. Using the preset controlled range of process parameters as the on-site execution benchmark, collect meat samples at key stages such as the end of smoking preheating, the middle of smoking, and the completion of braising, following random sampling principles. Measure core quality indicators such as meat tenderness, flavor compound content, color Lab* value, and water activity using professional testing equipment. Digitally analyze each quality testing indicator, transforming qualitative descriptions into quantitative values, establishing a correspondence between quality characteristics and unique numerical sequences, and eliminating the subjectivity of manual evaluation. Integrate all quantified quality indicators according to weights into a structured feature set, forming a meat quality digital fingerprint that can be accurately compared and traced throughout the entire process.

[0079] S24: Associate and match the process parameter control range with the quality digital fingerprint to form a dual trigger index. Combine the historical high-quality parameter sequences in the process gene library to fit the parameter change pattern and generate a process curve gene template.

[0080] The specific steps to form a dual-trigger indicator are as follows:

[0081] Based on the range of process parameter control and the digital fingerprint of meat quality, the dimensional hierarchy of the range of process parameter control is decomposed, the core quantitative features of the digital fingerprint of quality are deconstructed, and a mapping framework between parameter dimensions and quality characteristics is established.

[0082] Based on the mapping framework, the upper and lower limits and the optimal range of process parameter control are associated one by one with the feature thresholds of the quality digital fingerprint, forming a basic correlation matrix between parameters and quality.

[0083] The validity of the basic correlation matrix is ​​verified, invalid correlation items are eliminated, the correspondence of deviations is corrected, the linkage between parameter changes and quality characteristic fluctuations is strengthened, and parameter-quality correlation rules are formed.

[0084] Based on the parameter-quality correlation rules, a two-way triggering logic is built, setting a quality early warning threshold when the parameter exceeds the control range and a parameter adjustment threshold when the quality fingerprint deviates from the standard, thus forming a dual-trigger indicator.

[0085] Based on the association rules of dual-trigger indicators, historical high-quality parameter sequences under the same raw materials and environment are retrieved from the process gene library. The dynamic change trend of these parameters with the smoking process is analyzed using a time series fitting algorithm to fit the parameter change pattern that aligns with the predetermined quality target. Using the fitted parameter change pattern as the core framework, combined with the quality constraints of the dual-trigger indicators, the parameter thresholds and high-quality sequence information from the process gene library are integrated to generate a dynamically adjustable process curve gene template with multiple tags for raw materials, environment, and process steps.

[0086] S3: The process curve gene template is split into single-fumigation-specific sub-templates according to the gene characteristics of the fumigation chamber. Real-time parameters and coupling interference data are collected to make four-level deviation judgments. A hierarchical regulation + coupling deviation linkage compensation mechanism is designed. Deviation data with multiple labels are retained to form a deviation gene dataset.

[0087] S31: Based on the process curve gene template and the equipment precision, spatial layout, and historical operating characteristics of each individual fumigation chamber, a unique sub-template adapted to each fumigation chamber is derived. According to the process curve gene template, the core characteristics of each individual fumigation chamber are analyzed, and key information such as equipment precision attenuation data, internal spatial layout parameters, parameter fluctuation patterns during historical operation, and failure frequency are statistically analyzed for each chamber. Then, the threshold values ​​and timing patterns of the general parameters in the process curve gene template are adjusted specifically to the unique characteristics of each fumigation chamber to adapt to the equipment capabilities and operating conditions of different chambers. Finally, the template decomposition is completed, generating a unique sub-template corresponding to each fumigation chamber, labeling the fumigation chamber number and core adaptation basis for the sub-template, ensuring that the sub-template can be directly implemented.

[0088] S32: Real-time acquisition of process parameter data and coupling interference data for each smoking chamber. Comparison with preset parameters in the dedicated sub-template for each smoking chamber according to a four-level deviation judgment standard to identify deviations in each dimension and obtain a four-level deviation judgment result. Real-time data of process parameters such as temperature, humidity, and smoke concentration throughout the smoking process in each smoking chamber, as well as coupling interference data such as temperature field superposition and smoke flow mutual interference generated by the operation of adjacent smoking chambers, are collected and structured for storage at the second-level time sequence. The dedicated sub-template for each smoking chamber is retrieved, and the preset parameter thresholds and time-series change standards are extracted from the sub-template as a benchmark for deviation judgment. Based on this benchmark, the collected real-time process parameters and coupling interference data are compared and analyzed one by one according to the four-level deviation judgment standard to accurately identify the magnitude and duration of deviations in each parameter dimension, ultimately forming a four-level deviation judgment result that includes deviation level, deviation dimension, and degree of interference impact.

[0089] S33: Based on the four-level deviation judgment results, design differentiated graded control strategies according to deviation levels and establish a coupled interference deviation linkage compensation mechanism. Classify and analyze the four-level deviation judgment results, determine the parameter offset range, interference type, and degree of impact on smoking quality corresponding to different deviation levels, and classify the control priorities corresponding to four deviation levels: slight, moderate, severe, and critical. Based on the control priorities of each deviation level, design differentiated graded control strategies, determine the adjustment range, frequency, and execution nodes of control parameters for each level, and ensure that the control strategies conform to the actual deviation situation. Combine the coupled interference type and intensity in the judgment results to establish a coupled interference deviation linkage compensation mechanism, linking the control strategies corresponding to the deviation levels to achieve synergistic linkage between deviation control and interference compensation, avoiding secondary deviations caused by single control.

[0090] S34: Collect all deviation data tagged with fumigation chamber number, deviation level, interference type, and regulation method during the coupling interference deviation linkage compensation mechanism compensation process. Standardize and organize the data to obtain a deviation gene dataset. After the coupling interference deviation linkage compensation mechanism is initiated, the deviation data acquisition channel is opened to collect various relevant data during the regulation compensation process. Each data point is uniformly tagged with four core labels: fumigation chamber number, deviation level, interference type, and regulation method, ensuring data traceability and classification. The collected tagged deviation data is standardized through outlier removal, missing value completion, unit unification, and time series organization to eliminate data distortion and inconsistent formats. The standardized deviation data is structured and stored according to fumigation chamber number and deviation level, ultimately forming a standardized, complete deviation gene dataset that can be used for subsequent causal analysis and process optimization.

[0091] S4: The deviation gene dataset is classified and statistically analyzed in multiple dimensions. A four-dimensional correlation analysis is performed by combining the basic data of the fumigation chamber group with the full data of twins. The core causative genes are located through a combination algorithm. The threshold of the process gene library is optimized and an attenuation early warning model is established. The gene template of the process curve is updated to achieve three-level self-optimization.

[0092] S41: Based on the deviation gene dataset, multi-dimensional classification and statistics are performed according to fumigation room number, deviation level, interference type, and regulation method. Data on the frequency of deviation occurrence, impact degree, and regulation effect of each dimension are summarized to form structured deviation statistical results. The classification and statistical dimensions are determined based on the deviation gene dataset, focusing on four core dimensions: fumigation room number, deviation level, interference type, and regulation method. The dataset is then split and categorized. Each category of data is then statistically analyzed separately, summarizing the frequency of deviation occurrence, duration, impact on fumigation quality, and key data such as the implementation effect of corresponding regulation strategies and the deviation correction rate after regulation. Finally, the statistical data of each type are structured and organized to form deviation statistical results that include statistical dimensions, core statistical indicators, and data trends. Data sources and statistical standards are labeled to provide accurate data support for subsequent correlation analysis.

[0093] S42: Based on the deviation statistics, retrieve the full data of the twin of the basic data of the fumigation chamber group, and conduct correlation analysis from four dimensions: process parameters, equipment status, environmental conditions, and coupling interference. Explore the inherent correlation between deviation and data in each dimension, and identify the key influencing factors that cause deviation.

[0094] The deviation statistics were analyzed to identify the dimensions with concentrated deviations and the types of deviations with high impact, thus determining the key areas for correlation analysis. Based on these key areas, the entire dataset of the fumigation chamber group's basic data twin was retrieved, and historical process parameter data, equipment operating status data, environmental condition data, and coupling interference data matching the deviation statistics were selected. Finally, in-depth analysis was conducted using data correlation algorithms across four dimensions: process parameter rationality, equipment accuracy and stability, environmental condition volatility, and coupling interference intensity. This revealed the inherent correlation patterns between various deviations and the data in each dimension, eliminated irrelevant influencing factors, identified the key influencing factors causing the deviations, and generated a correlation analysis report.

[0095] S43: Based on key influencing factors, identify the core causative genes causing process deviations, determine the parameter threshold deviations corresponding to these core causative genes, and formulate threshold optimization schemes. Based on the correspondence between key influencing factors and parameters and genes in the process gene library, trace the process genes corresponding to each key influencing factor, distinguishing between secondary and core causative genes. Then, conduct focused analysis on the core causative genes, clarifying the specific differences between the corresponding process parameter threshold ranges, optimal ranges, and actual deviations, and identifying core issues such as unreasonable parameter threshold settings and threshold deviations. Finally, combining the control effect data from the deviation statistics, develop targeted parameter threshold adjustment schemes for the core causative genes, clarifying the adjustment range, optimization standards, and verification methods, forming a complete threshold optimization scheme.

[0096] S44: Adjust the threshold ranges and optimal ranges in the process gene library according to the threshold optimization scheme. Based on the full data of the optimized process gene library, refit the parameter variation patterns of the entire fumigation process and iteratively update the process curve gene template. The threshold optimization scheme adjusts the parameter threshold ranges, upper and lower limits, and optimal ranges of the corresponding core causative genes in the process gene library one by one, and simultaneously updates the parameter classification, storage, and retrieval system of the gene library to ensure that the optimized thresholds can be quickly retrieved. Then, retrieve the full data of the optimized process gene library, and combine it with deviation statistics and correlation analysis patterns. Using a time series fitting algorithm, refit the parameter variation patterns of the entire fumigation preheating, fumigation, and simmering process. Finally, based on the refitted parameter variation patterns and the specific characteristics of each individual fumigation chamber, iteratively update the original process curve gene template, annotating the updated content and optimization basis to ensure that the updated template can adapt to actual production needs and achieve precise upgrades in process control.

[0097] S5: Based on the three-level self-optimized process gene library, a three-level gene mapping model for sensory indicators is built, which transforms artificial sensory evaluation into quantitative characteristic factors, sets flavor gene intervals, transforms artificial fine-tuning experience into parameter increment gene values, and generates differentiated batch flavor customization fine-tuning instructions.

[0098] S51: Extract sensory index parameters, quality digital fingerprints, and deviation gene data related to meat products from the process gene library, and construct a three-level gene mapping model between sensory indicators and process genes. First, retrieve the optimized and upgraded process gene library to accurately screen core data directly related to meat sensory indicators, including smoking process parameters, meat quality digital fingerprints, and deviation gene datasets. Standardize and streamline the screened data, removing irrelevant and redundant information. Then, combining meat smoking industry standards and market demands, identify three core sensory dimensions: color, flavor, and texture, and break down the sub-evaluation points for each dimension. Finally, based on the streamlined data and defined sensory dimensions, construct a three-level gene mapping model of process genes, quantitative characteristics, and sensory indicators, clarifying the correspondence, quantitative standards, and analytical logic at each level to ensure the model can achieve bidirectional mapping between process genes and sensory indicators.

[0099] S52: Based on a three-level genotyping model, artificial sensory evaluation data is collected. The descriptions of color depth, flavor intensity, and tenderness by evaluators are converted into quantifiable sensory characteristic factors through model analysis. The specific process and evaluation standards for artificial sensory evaluation are determined according to the three-level genotyping model. Professional evaluators conduct sensory scoring of meat products under different smoking stages and process parameters. The system collects qualitative evaluation data such as color depth, flavor intensity, and tenderness, and labels the corresponding smoking room number, process parameters, and other related information. The collected qualitative evaluation data is then entered into the three-level genotyping model, and the model's built-in algorithm analyzes and transforms the vague qualitative descriptions into specific quantitative values. Finally, the transformed quantitative values ​​are categorized and organized to form accurately comparable and traceable sensory characteristic factors, which are then linked to corresponding process gene data, completing the digital transformation of artificial sensory evaluation.

[0100] S53: Based on sensory quantitative characteristic factors and the parameter ranges of high-quality flavors in the process gene library, flavor gene ranges corresponding to different flavor requirements are set, the core parameter ranges and characteristic factor thresholds of the flavor ranges are determined, and the boundaries of flavor customization are defined. Based on sensory quantitative characteristic factors and the process parameter ranges corresponding to high-quality flavor meat products in the process gene library, the intrinsic relationship between high-quality flavor and sensory characteristic factors and process parameters is analyzed, clarifying the core characteristics of different flavor requirements. Based on the correlation analysis results, flavor gene ranges adapted to different flavor requirements are set, and the core process parameter ranges and sensory characteristic factor thresholds corresponding to each flavor range are determined one by one, clarifying the matching standards between parameters and characteristic factors. Finally, the parameter and threshold information of all flavor gene ranges is integrated to define the reasonable boundaries of flavor customization, clarifying which parameters can be adjusted, their adjustment range, and the corresponding flavor changes, providing a clear basis for subsequent fine-tuning of flavor customization.

[0101] S54: Based on the flavor customization boundary, summarize the experience of flavor optimization in human sensory evaluation, and transform this human experience into incremental gene values ​​of process parameters through a three-level gene mapping model to generate flavor customization fine-tuning instructions. Based on the defined flavor customization boundary, summarize the flavor optimization experience accumulated during human sensory evaluation, select high-quality experiences that meet the customization boundary and are reusable, classify and organize these experiences, and label the corresponding flavor requirements and sensory feedback. Then, substitute the organized human flavor optimization experience into the three-level gene mapping model, and through model analysis, transform abstract experiences into specific, executable incremental gene values ​​of process parameters, clarifying the parameter adjustment range, direction, and adjustment node for each experience. Finally, integrate all incremental gene values ​​of process parameters, combine them with flavor gene interval standards, and generate flavor customization fine-tuning instructions with multiple labels that can directly guide production, labeling the flavor requirements and parameter adjustment details of the instructions to ensure that the instructions are implementable and traceable.

[0102] S6: Collect all-link data and organize it into a process gene ledger according to five-dimensional labels, improve the digital twin of the entire smoking process, screen high-quality process gene combinations through improved clustering and include them in the high-quality gene library, realize the self-evolution of the process gene library through gene recombination and mutation, and solidify and update the process curve gene template to form an iterative closed loop.

[0103] S61: Classify and organize the full data of the process gene library and the generated flavor customization fine-tuning instructions according to process links, gene types and instructions, unify data formats, labeling standards and traceability information, and form a process gene ledger.

[0104] The entire process gene library and the generated flavor customization fine-tuning instructions are analyzed to identify their core attributes. A three-dimensional classification is completed based on the smoking preheating, smoking, and simmering process stages, process genes, deviation genes, incremental gene types, and the purpose of flavor customization and process control instructions. The classified data and instructions are then standardized in format and labeled, and traceability information such as data collection time and instruction execution scenario is added. Finally, all content is integrated according to the classification logic to form a structured, traceable, and easily searchable process gene ledger, achieving systematic management of data and instructions.

[0105] S62: Integrate the data from the process gene ledger into the full-chain digital twin of the smokehouse group, supplementing sensory genes and flavor customization to obtain a complete full-chain digital twin. Extract core data such as sensory genes, flavor customization instructions, and process parameter thresholds from the process gene ledger, and complete the data format adaptation and conversion with the twin interface. Synchronously integrate the adapted data into the full-chain digital twin of the smokehouse group, supplementing the two core modules of sensory gene mapping and flavor customization control, and establishing a real-time linkage channel between the ledger data and the twin. Verify the integrity of the twin data and the effectiveness of linkage, correct data docking deviations, and obtain a complete full-chain digital twin covering all dimensions of the process and the entire customization process.

[0106] S63: Based on a comprehensive end-to-end digital twin, high-quality process genes are screened, and a gene self-evolution algorithm is built to obtain the self-evolution results of the gene library. Based on the end-to-end digital twin, high-quality process gene screening criteria are set, including core indicators such as flavor compliance rate, parameter control stability, and deviation correction efficiency. All process genes in the twin are screened from multiple dimensions. The screened high-quality process genes are classified according to function and applicable scenarios, and a dedicated high-quality gene sub-library is built to achieve precise aggregation of high-quality genes. Based on the operational data and iteration patterns of high-quality genes, a gene self-evolution algorithm is built to achieve dynamic screening, updating, and optimization of the gene library, resulting in a self-evolving gene library that can iterate autonomously.

[0107] S64: Based on the self-evolution results of the gene bank and the latest data in the standardized process gene ledger, the parameter variation patterns of the entire smoking process are refitted, and the original process curve gene template is iteratively updated to form an iterative closed loop. First, the latest data from the self-evolution results of the gene bank and the standardized process gene ledger are retrieved, and core information such as high-quality gene parameters, the latest flavor customization instructions, and process operation deviation data are integrated as the basis for template iteration. Then, based on the integrated full data, a time series fitting algorithm is used, combined with the specific characteristics of each individual smoking room, to refit the parameter variation patterns of the entire smoking process and optimize the parameter control timing and threshold range. Finally, based on the new parameter variation patterns, the original process curve gene template is iteratively updated in all dimensions to verify the adaptability of the template to actual production, forming a closed-loop iterative process chain of data collection - model optimization - control execution - gene evolution - template update.

[0108] like Figure 3 As shown, the present invention also provides a collaborative optimization system for plasticizer production based on the Industrial Internet, comprising:

[0109] The twin-based foundation module is used to collect multi-dimensional detection and coupling interference data from each fumigation chamber. It assigns dual weights based on the equipment accuracy attenuation coefficient, generates a multi-labeled accurate detection dataset, and constructs a twin of basic data for the fumigation chamber group.

[0110] The gene modeling module is used to extract tagged parameters from the twin data of the smokehouse group and input them into the process gene library. It sets thresholds for clustering, transforms meat sampling data into quality digital fingerprints, binds parameters into dual trigger indicators, and generates process curve gene templates.

[0111] The deviation control module is used to split the process curve gene template into a single smokehouse-specific sub-template, collect real-time parameters and coupling interference data to determine the deviation at four levels, design a graded regulation and compensation mechanism, and retain the deviation data with multiple labels to obtain the deviation gene dataset.

[0112] The gene optimization module is used to classify and statistically analyze the deviation gene dataset, combine it with the basic data of the smokehouse group and the full data of twins to perform four-dimensional correlation analysis, locate the core causative genes, optimize the threshold of the process gene library, and update the gene template of the process curve.

[0113] The flavor customization module is used to build a three-level gene mapping model of sensory indicators based on the process gene library, transform human sensory evaluation into quantitative characteristic factors, set flavor gene ranges, transform human experience into parameter increment gene values, and generate flavor customization fine-tuning instructions.

[0114] The gene iteration module is used to organize the process gene library and flavor customization fine-tuning instructions into a standardized process gene ledger, improve the digital twin of the whole chain, screen high-quality genes to build a library and achieve self-evolution, and update the process curve gene template to form an iterative closed loop.

[0115] In summary, this embodiment provides an intelligent control method and system for smoking chambers in cured meat production. By constructing a full-dimensional data calibration and digital twin, it achieves precise control of the smoking process. By collecting sensor and instrument detection data, as well as cross-smoking chamber coupling interference data, and assigning dual weights based on equipment accuracy attenuation coefficients, a three-level calibration eliminates triple distortions from equipment, environment, and transmission, generating a precise detection dataset and constructing a basic data twin of the smoking chamber group, achieving a precise digital replication of the physical smoking chamber group. This allows for real-time and precise traceability of core parameters such as temperature, humidity, and smoke concentration during the smoking process. Simultaneously, it restores the coupling interference relationships within the smoking chambers, solving the problem of uneven smoking quality caused by inaccurate data and neglect of cross-equipment interference in traditional control methods. It ensures that the control of each process parameter is based on real and comprehensive operational data, guaranteeing the accuracy of the smoking process from the data source and significantly reducing quality defects caused by parameter deviations.

[0116] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0117] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart control method for a smoking room used in the production of cured meat, characterized in that, include: S1: Collect multi-dimensional detection and coupling interference data from each fumigation chamber, assign dual weights based on the equipment accuracy attenuation coefficient, generate a multi-labeled accurate detection dataset, and construct a basic data twin of the fumigation chamber group; S2: Extract tagged parameters from the basic data twin of the smoking room group and input them into the process gene library. Set thresholds for clustering, convert meat sampling data into quality digital fingerprints, bind parameters into dual trigger indicators, and generate process curve gene templates. S3: The process curve gene template is split into a single smokehouse-specific sub-template, real-time parameters and coupling interference data are collected to determine the four-level deviation, a graded regulation and compensation mechanism is designed, and deviation data with multiple labels is retained to obtain the deviation gene dataset. S4: Classify and statistically analyze the deviation gene dataset, combine it with the full data of twins of the basic data of the fumigation chamber group to perform four-dimensional correlation analysis, locate the core causative genes, optimize the threshold of the process gene library, and update the gene template of the process curve. S5: Based on the process gene library, build a three-level gene mapping model for sensory indicators, transform artificial sensory evaluation into quantitative characteristic factors, set flavor gene intervals, transform artificial experience into parameter increment gene values, and generate flavor customization fine-tuning instructions. S6: Organize the process gene library and flavor customization fine-tuning instructions into a standardized process gene ledger, improve the whole-chain digital twin, screen high-quality genes to build a library and achieve self-evolution, and update the process curve gene template to form an iterative closed loop.

2. The intelligent control method for a smoking room used in the production of cured meat according to claim 1, characterized in that: In step S1, the specific steps for constructing the basic data twin of the fumigation chamber group are as follows: S11: Collect multi-dimensional detection data from sensors and zone instruments in each fumigation chamber, collect coupling interference data generated by the interaction between the fumigation chambers, and form full raw data; S12: Based on the full amount of original data and the accuracy attenuation coefficient of each detection device, weights are assigned to the detection data and the coupling interference data respectively to form a dual-weight data system; S13: Perform three-level calibration on the dual-weighted data system to eliminate the triple data distortion problem of equipment error, environmental interference and data transmission, and obtain three-level calibration data; S14: Based on the three-level calibration data, label the data with fumigation room number, detection dimension, and collection time label to generate a precise detection dataset; S15: Map the precise detection dataset to the physical state, data association and coupling relationship of each fumigation chamber, and construct a basic data twin of the fumigation chamber group in the digital space that corresponds one-to-one with the physical fumigation chamber group.

3. The intelligent control method for a smoking room used in the production of cured meat according to claim 1, characterized in that: In step S2, the specific steps for generating the process curve gene template are as follows: S21: Based on the digital base of the basic data twin of the fumigation room group, screen the core control parameters of the entire fumigation process, perform data cleaning and organization, and form a standardized parameter dataset. S22: Based on the standardized parameter dataset of the label, classify and enter the smoking process steps into the process gene library, build a parameter classification storage and retrieval system, perform secondary cluster analysis according to gene characteristics, and preset the process parameter control range; S23: Within the range of the process parameter control, collect sample data of meat products during the smoking process, and convert the quality characteristics into quantifiable quality digital fingerprints through digital analysis of quality detection indicators. S24: The process parameter control range is associated and matched with the quality digital fingerprint to form a dual trigger index. The parameter change pattern is fitted by combining the historical high-quality parameter sequences in the process gene library to generate a process curve gene template.

4. The intelligent control method for a smoking room used in the production of cured meat according to claim 3, characterized in that: In step S24, the specific steps for forming the dual-trigger indicator are as follows: Based on the range of process parameter control and the digital fingerprint of meat quality, the dimensional hierarchy of the range of process parameter control is decomposed, the core quantitative features of the digital fingerprint of quality are deconstructed, and a mapping framework between parameter dimensions and quality features is established. Based on the mapping framework, the upper and lower limits and the optimal range of the process parameter control range are associated with the feature thresholds of the quality digital fingerprint one by one, forming a basic correlation matrix between parameters and quality. The validity of the basic correlation matrix is ​​verified, invalid correlation items are removed, the correspondence of deviations is corrected, the linkage between parameter changes and quality characteristic fluctuations is strengthened, and parameter-quality correlation rules are formed. Based on the parameter-quality correlation rules, a two-way triggering logic is built, setting a quality early warning threshold when the parameter exceeds the control range and a parameter adjustment threshold when the quality fingerprint deviates from the standard, thus forming a dual-trigger indicator.

5. The intelligent control method for a smoking room used in the production of cured meat according to claim 1, characterized in that: In step S3, the specific steps for obtaining the biased gene dataset are as follows: S31: Based on the process curve gene template and the equipment precision, spatial layout, and historical operating characteristics of each individual fumigation chamber, a unique sub-template adapted to each fumigation chamber is decomposed. S32: Real-time collection of process parameter data and coupling interference data of each smoking room, comparison with the preset parameters of the single smoking room's dedicated sub-template according to the four-level deviation judgment standard, identification of parameter deviations in each dimension, and obtaining the four-level deviation judgment result; S33: Based on the four-level deviation judgment results, design differentiated graded control strategies according to the deviation level, and build a coupled interference deviation linkage compensation mechanism. S34: Collect all deviation data labeled with fumigation chamber number, deviation level, interference type, and regulation method during the compensation process of the coupling interference deviation linkage compensation mechanism, standardize and organize the data to obtain the deviation gene dataset.

6. The intelligent control method for a smoking room used in the production of cured meat according to claim 1, characterized in that: In step S4, the specific steps for optimizing the process gene library threshold and updating the process curve gene template are as follows: S41: Perform multi-dimensional classification and statistics based on the aforementioned deviation gene dataset, summarize the frequency of deviation occurrence, degree of influence and regulatory effect data of each dimension, and form deviation statistical results; S42: Based on the deviation statistics, retrieve the full data of the twin of the basic data of the fumigation chamber group, and conduct correlation analysis from four dimensions: process parameters, equipment status, environmental conditions, and coupling interference. Explore the inherent correlation between the deviation and the data of each dimension, and identify the key influencing factors of the deviation. S43: Based on the key influencing factors, locate the core causative gene that causes process deviation, determine the parameter threshold deviation problem corresponding to the core causative gene, and form a threshold optimization scheme. S44: Adjust the threshold range and optimal range in the process gene library according to the threshold optimization scheme, and refit the parameter change law of the entire fumigation process based on the full data of the optimized process gene library, and iteratively update the process curve gene template.

7. The intelligent control method for a smoking room used in the production of cured meat according to claim 1, characterized in that: In step S5, the specific steps for generating flavor customization fine-tuning instructions are as follows: S51: Extract sensory index parameters, quality digital fingerprints, and deviation gene data of meat products from the process gene library, and build a three-level gene mapping model between sensory indicators and process genes. S52: Collect artificial sensory evaluation data according to the three-level gene mapping model, and convert the evaluation personnel's descriptions of color depth, flavor intensity and tenderness into quantifiable sensory characteristic factors through model analysis. S53: Based on the sensory quantitative characteristic factors and the parameter range of high-quality flavors in the process gene library, set the flavor gene range corresponding to different flavor requirements, determine the core parameter range and characteristic factor threshold of the flavor range, and define the flavor customization boundary. S54: Based on the flavor customization boundary, summarize the experience of optimizing flavor in human sensory evaluation, and transform human experience into incremental gene values ​​of process parameters through a three-level gene mapping model to generate flavor customization fine-tuning instructions.

8. The intelligent control method for a smoking room used in the production of cured meat according to claim 7, characterized in that: In step S51, the specific steps for building the three-level gene mapping model are as follows: Extract process parameters, quality digital fingerprints, and deviation gene data related to color, flavor, and taste from the process gene library, and remove outliers and unify dimensions to obtain standard sensory correlation data. Based on the sensory association data, the evaluation points of the three major sensory dimensions of color, flavor, and taste are decomposed, the correspondence between process genes, quality fingerprints, and deviation data is determined, and a basic mapping framework is built. Based on the aforementioned mapping framework, a three-level hierarchical structure of process genes, quantitative features, and sensory indicators is constructed. The mapping logic, quantification standards, and parsing rules for each level are determined, and a three-level gene-based mapping model between sensory indicators and process genes is built.

9. The intelligent control method for a smoking room used in the production of cured meat according to claim 1, characterized in that: In step S6, the specific steps for updating the process curve gene template to form an iterative closed loop are as follows: S61: Classify and organize the full data of the process gene library and the generated flavor customization fine-tuning instructions according to process links, gene types and instructions, unify data formats, labeling standards and traceability information, and form a process gene ledger. S62: Integrate the data in the process gene ledger into the digital twin of the entire smoking room group, supplement the sensory genes and flavor customization, and obtain a complete digital twin of the entire chain; S63: Based on the improved full-link digital twin, select high-quality process genes, build a gene self-evolution algorithm, and obtain the gene library self-evolution results; S64: Based on the self-evolution results of the gene bank and the latest data in the standardized process gene ledger, refit the parameter change pattern of the entire fumigation process, iteratively update the original process curve gene template, and form an iterative closed loop.

10. An intelligent control system for a smoking room for cured meat production, comprising an intelligent control method for a smoking room for cured meat production as described in any one of claims 1 to 9, characterized in that, The control system includes: The twin-based foundation module is used to collect multi-dimensional detection and coupling interference data from each fumigation chamber. It combines the equipment accuracy attenuation coefficient to assign dual weights, generate a multi-labeled accurate detection dataset, and construct a twin of basic data for the fumigation chamber group. The gene molding module is used to extract tagged parameters from the basic data twins of the smokehouse group and input them into the process gene library. Clustering is set with thresholds to convert meat sampling data into quality digital fingerprints, bind parameters into dual trigger indicators, and generate process curve gene templates. The deviation control module is used to split the process curve gene template into a single smokehouse-specific sub-template, collect real-time parameters and coupling interference data to determine the deviation at four levels, design a graded regulation and compensation mechanism, and retain the deviation data with multiple labels to obtain the deviation gene dataset. The gene optimization module is used to classify and statistically analyze the deviation gene dataset, perform four-dimensional correlation analysis in combination with the full data of twins of the smokehouse group, locate the core causative genes, optimize the threshold of the process gene library, and update the gene template of the process curve. The flavor customization module is used to build a three-level gene mapping model of sensory indicators based on the process gene library, transform artificial sensory evaluation into quantitative feature factors, set flavor gene intervals, transform artificial experience into parameter increment gene values, and generate flavor customization fine-tuning instructions. The gene iteration module is used to organize the process gene library and flavor customization fine-tuning instructions into a standardized process gene ledger, improve the digital twin of the whole chain, screen high-quality genes to build a library and achieve self-evolution, and update the process curve gene template to form an iterative closed loop.

Images