Deep management-based health food production line control method and system

Abstract

The invention discloses a health food production line control method and system based on deep management, and relates to the technical field of food production control. A raw material feature database is established, a first analysis result is output as a raw material deviation evaluation reference, data of a current batch of raw materials are collected, differences are compared, and a raw material deviation result is output; the method comprises the following steps: collecting process data, analyzing whether the process data is abnormal or not by combining with raw material deviation, generating an adjustment suggestion, executing formula and parameter adjustment, triggering a self-repairing mechanism when the process data is abnormal, identifying an intermediate product state, evaluating a quality risk level and triggering early warning. Dynamic optimization of a formula and process parameters is achieved, meanwhile, an image and spectrum fusion early warning technology is introduced, the quality control ability of intermediate products is improved, finally, optimization parameter configuration is generated for the next batch through whole-process data chain construction and causal reasoning analysis, and self-adaptive control and quality steady-state improvement in a production closed loop are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of food production control, and in particular to a health food production line control method and system based on in-depth management. Background Art

[0002] As the functions of health food products become increasingly diverse and the sources of raw materials become increasingly complex, traditional production line control methods often use static formulas and fixed process parameters, which cannot achieve real-time response to raw material fluctuations and process disturbances, resulting in poor product quality consistency, strong process control rigidity, and difficulty in timely identification of abnormal intermediate products. Existing systems often rely on manual judgment or preset rules, lack deep data modeling and closed-loop optimization capabilities, and are unable to support the actual needs of modern health foods for precise, flexible and intelligent production. Therefore, there is an urgent need for a health food production line control method that can achieve dynamic identification, intelligent adjustment and traceable optimization based on data-driven methods.

[0003] At present, the Chinese invention patent with application number CN202411887620.1 discloses an automated management method and system for food production lines based on intelligent control. The method includes: extracting production lines that can produce target products based on product names, and then formulating production plans and allocating production lines for production orders based on preset rules. Before the first production order is produced, the control information and initial operating parameters of each process on the target production line are obtained, and based on the control information and initial operating parameters, it is judged whether the target production line meets the preset standards. If it meets the standards, production is started. During the order production process, the production progress information and product quality information of the first production order are collected and monitored. If the production progress or product quality is abnormal, a production plan adjustment suggestion or reminder information is generated and sent to relevant personnel. After the first production order is completed, a shipping reminder is generated.

[0004] The above technologies make it difficult to achieve full-process intelligent identification, dynamic regulation and closed-loop optimization control of raw material differences, process disturbances and intermediate quality status in health food production. Summary of the Invention

[0005] The technical problem solved by the present invention is that the above-mentioned technology is difficult to achieve full-process intelligent identification, dynamic regulation and closed-loop optimization control of raw material differences, process disturbances and intermediate quality states in the production of health food.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: A control method for a health food production line based on in-depth management comprises the following steps: Step S1, establishing a raw material feature database and outputting a first analysis result as a raw material deviation assessment benchmark; Step S2: Collect the raw material data of the current batch, compare the differences, and output the raw material deviation results; Step S3: Collect process data and analyze whether there are any abnormalities based on raw material deviations, and generate adjustment suggestions; Step S4, executing recipe and parameter adjustment, triggering the self-repair mechanism in case of abnormality; Step S5, identifying the status of the intermediate product, assessing the quality risk level and triggering an early warning; Step S6: Establish a full-process data chain, perform causal analysis, and generate the next batch of optimized parameters.

[0007] Preferably, step S1 includes the following sub-steps: Step S101: Retrieving raw material data from a raw material database that has been assessed as high-quality batches during historical production, extracting key parameter information for each batch of raw materials, including moisture content, trace element composition, particle size distribution, storage time, and batch number, and performing structured processing on the key parameter information to output a standard data set; Step S102: Cluster and classify the standard data set using a cluster analysis method based on similarity measurement to identify high-consistency raw material categories, form feature vector templates for the high-consistency raw material categories, and construct the feature vector templates and corresponding label data into a raw material feature database. The raw material feature database serves as a reference benchmark, and at the same time, outputs the feature template results of the cluster analysis method as the first analysis result.

[0008] Preferably, step S2 includes the following sub-steps: Step S201: Collect the moisture content, trace element composition, particle size distribution, storage time, and batch number of the current batch of raw materials at the raw material storage stage, format them, and output them as a structured data set of the current batch of raw materials; In step S202, the structured data set is input into the feature comparison model, and a multi-dimensional feature difference comparison is performed with the corresponding feature vector template in the raw material feature database. A comprehensive difference score is calculated based on the degree of deviation of the feature data, and a judgment is made based on a preset threshold whether it constitutes a significant raw material deviation. The raw material deviation result is output and marked for archiving.

[0009] Preferably, step S3 includes the following sub-steps: Step S301: Collect temperature, humidity, stirring rate, pH value, flow rate, and viscosity data during the production process, upload the collected results to the edge computing node, and generate a real-time process data set; Step S302: Receive the real-time process data set and the raw material deviation results, perform a difference comparison analysis against the preset standard process execution model, and determine whether there is an abnormal state that deviates from the normal process curve. If there is an abnormal state that deviates from the normal process curve, it is marked as a process deviation, and the deviation judgment result is output and the process link affected by the deviation is recorded; Step S303: Receive the deviation judgment result and combine it with the first analysis result, call the deep adjustment model to perform multi-parameter simulation on the process link where the deviation occurs, calculate the optimal adjustment strategy based on the historical successful experience path and the current batch of raw material structured data set, and output formula adjustment suggestions and process parameter correction suggestions. The formula adjustment suggestions and process parameter correction suggestions include raw material ratio change suggestions, stirring or heating rate correction suggestions, and time node adjustment suggestions.

[0010] Preferably, step S4 includes the following sub-steps: Step S401: Output a recipe adjustment signal based on the recipe adjustment suggestion, obtain a real-time raw material difference index, correct the target ratio and form a corrected recipe for the current batch, and output the corrected recipe for the current batch as a recipe correction signal; Step S402: updating the stirring time, heating temperature, and drying time according to the recipe correction signal, collecting real-time process data and generating a process parameter execution plan based on the recipe correction signal, verifying the parameter range and the upper and lower safety limits, and forming a corrected process parameter set; Step S403: After executing the modified process parameter set, monitor the key process parameters. If there is a persistent deviation in the key process parameters or the intermediate product status is abnormal, the current operation process is suspended, buffer mark data is generated, and the self-repair logic mechanism is triggered. The historical optimal parameter set is called for comparison and analysis, and a feasible parameter fallback solution and process bypass strategy are recommended.

[0011] Preferably, step S5 includes the following sub-steps: Step S501, collecting production status image data of the intermediate product, wherein the collection process nodes of the production status image data include a mixing node, a heating node, and a drying node, extracting color saturation, appearance morphology contour, flow state, and surface viscosity texture of the production status image data and outputting an intermediate product image feature vector; Step S502 , performing spectral scanning on the intermediate product, extracting the active ingredient concentration, impurity distribution, and dissolution uniformity, and outputting them as a chemical state feature vector; Step S503: Perform quality risk analysis on the image feature vector and chemical state feature vector inputs. The logic of the quality risk analysis is: The image feature vector and chemical state feature vector are compared with the historical high-quality sample database in multiple dimensions to calculate the risk level score of the current intermediate product. If the risk level score is higher than the set threshold, the second analysis result is output and marked as a warning state, and the corresponding image feature vector and chemical state feature vector are recorded.

[0012] Preferably, if the risk level score is higher than the set threshold, a quality warning mechanism is triggered, which sends a stop operation instruction or a slowdown operation instruction to the process control end and records the corresponding relevant risk level, time node, corresponding intermediate product image features and chemical state data.

[0013] Preferably, step S6 includes the following sub-steps: Step S601: Obtain the final product quality inspection results. Number the current batch of raw material structured data sets, real-time process data sets, recipe adjustment suggestions, process parameter correction suggestions, production status image data, chemical state feature vectors, and final product quality inspection results. Construct a structured data chain in chronological order to establish a full-process data chain, which includes the current batch data. Step S602: Perform causal reasoning analysis on the entire process data chain. The causal reasoning analysis is: Use a multivariate association model to identify the source variables of abnormal or fluctuating finished product quality and output the abnormal path, which includes raw material characteristics, process nodes, and adjustment response delays; Step S603: Perform a comparative analysis based on the execution data of the abnormal path and the historical optimal batch, and output the optimization parameter set for the next batch in combination with the system experience rule library. The optimization parameter set includes the recommended initial ratio of the recipe, the target value of the recommended process parameters, and the updated quality warning threshold range, and is synchronized to the next round of process initialization configuration.

[0014] Preferably, the optimization parameter set includes an initial recipe suggestion generated based on historical high-quality batch data and current abnormal path analysis results, the initial recipe suggestion includes initial process parameter settings, and the process parameter settings include initial control values of time and temperature of process units; The optimization parameter set also includes a dynamic warning threshold.

[0015] A control system for a health food production line based on deep management, including a basic data acquisition module, a current raw material deviation module, an adjustment suggestion output module, an execution adjustment and repair module, an intermediate product identification module, and a subsequent batch update module; The basic data acquisition module is used to establish a raw material feature database and output a first analysis result as a raw material deviation assessment benchmark; The current raw material deviation module is used to collect the current batch of raw material data, compare the differences, and output the raw material deviation results; The adjustment suggestion output module is used to collect process data and analyze whether there are abnormalities in combination with raw material deviations to generate adjustment suggestions; The execution adjustment and repair module is used to execute recipe and parameter adjustments and trigger the self-repair mechanism in case of abnormality; The intermediate product identification module is used to identify the status of intermediate products, assess the quality risk level and trigger an early warning; The subsequent batch update module is used to establish a full-process data chain, perform cause-effect analysis and generate the next batch of optimized parameters.

[0016] Beneficial effects of the present invention: The present invention integrates raw material clustering modeling, real-time process perception, deviation identification and intelligent adjustment mechanism to achieve dynamic optimization of formula and process parameters. At the same time, it introduces image and spectral fusion early warning technology to improve the quality control capability of intermediate products. Finally, through the construction of a full-process data chain and causal reasoning analysis, it generates optimized parameter configuration for the next batch, realizing adaptive control and steady-state quality improvement in the production closed loop. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flowchart of a method for controlling a health food production line based on in-depth management according to an embodiment of the present invention; Figure 2 A schematic diagram of the basic flow of a health food production line control system based on in-depth management provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0019] Example 1, with reference to Figure 1 , provides a health food production line control method based on in-depth management, comprising the following steps: Step S1: Establish a raw material feature database and output a first analysis result as a raw material deviation assessment benchmark.

[0020] Step S2: Collect the raw material data of the current batch, compare the differences, and output the raw material deviation results.

[0021] Step S3: Collect process data and analyze whether there are any abnormalities based on raw material deviations, and generate adjustment suggestions.

[0022] Step S4: execute recipe and parameter adjustment, and trigger the self-repair mechanism in case of abnormality.

[0023] Step S5: Identify the status of the intermediate product, assess the quality risk level and trigger an early warning.

[0024] Step S6: Establish a full-process data chain, perform causal analysis, and generate the next batch of optimized parameters.

[0025] Step S1 includes the following sub-steps: Step S101: retrieve raw material data that has been assessed as high-quality batches during historical production processes from the raw material database, extract key parameter information for each batch of raw materials, including moisture content, trace element composition, particle size distribution, storage time, and batch number, and perform structured processing on the key parameter information to output a standard data set.

[0026] Step S101 is used to extract key parameter information of historical high-quality batches of raw materials, perform standardization processing, and construct a structured standard data set to provide an original high-quality data source for subsequent raw material clustering and feature modeling.

[0027] Step S102: Use a cluster analysis method based on similarity measurement to cluster and classify the standard data set, identify high-consistency raw material categories, form a feature vector template for the high-consistency raw material categories, and construct the feature vector template and the corresponding label data into a raw material feature database. The raw material feature database is used as a reference benchmark, and the feature template results of the cluster analysis method are output as the first analysis result.

[0028] Step S102 is used to classify and identify the standard data set through cluster analysis, extract the feature vector template of the high-consistency raw material category, and build a raw material feature database, while outputting the first analysis result as a reference benchmark for subsequent raw material comparison and deviation judgment.

[0029] Step S1 is used to establish a high-consistency raw material feature database. Through structured processing and cluster analysis of historical high-quality batch raw material data, a benchmark model for subsequent raw material deviation comparison and process control is constructed, providing stable data support for the system to realize raw material difference identification and process personalized adjustment.

[0030] Step S2 includes the following sub-steps: Step S201 : The moisture content, trace element composition, particle size distribution, storage time and batch number of the current batch of raw materials at the raw material storage stage are collected and formatted, and output as a structured data set of the current batch of raw materials.

[0031] Step S201 is used to collect the core parameter data of the current batch of raw materials during the raw material warehousing process, and perform unified formatting processing to generate a structured data set to ensure that the current raw material information is comparable and calculable.

[0032] In step S202, the structured data set is input into the feature comparison model, and a multi-dimensional feature difference comparison is performed with the corresponding feature vector template in the raw material feature database. A comprehensive difference score is calculated based on the degree of deviation of the feature data, and a judgment is made based on a preset threshold whether it constitutes a significant raw material deviation. The raw material deviation result is output and marked for archiving.

[0033] Step S202 is used to compare the current structured data set with the highly consistent templates in the raw material feature database, identify the degree of difference in each parameter dimension, calculate the comprehensive difference score, determine whether it constitutes a significant raw material deviation, and output the raw material deviation results, while archiving them for subsequent traceability analysis and strategy linkage.

[0034] Step S2 obtains the structured parameter data for the current batch of raw materials, compares it with the raw material feature database, and outputs the raw material deviation results. By calculating the multi-dimensional feature differences between the current raw materials and historical high-quality templates, a rapid assessment of raw material consistency and adaptability is achieved, providing a key basis for subsequent process adjustments.

[0035] Step S3 includes the following sub-steps: Step S301: Collect temperature, humidity, stirring rate, pH value, flow rate and viscosity data during the production process, upload the collected results to the edge computing node, and generate a real-time process data set.

[0036] Step S301 is used to collect key process parameters in real time during the production process and upload them to the edge computing node for processing, generating a real-time process data set for subsequent deviation analysis, ensuring that the process status has time synchronization and real-time response.

[0037] Step S302: Receive the real-time process data set and the raw material deviation results, perform a difference comparison analysis against the preset standard process execution model, and determine whether there is an abnormal state that deviates from the normal process curve. If there is an abnormal state that deviates from the normal process curve, it is marked as a process deviation, output the deviation judgment result, and record the process links affected by the deviation.

[0038] Step S302 is used to input the real-time process data set and the current raw material deviation result into the standard process execution model, and determine whether there is an abnormal process state by comparing whether the process state deviates from the preset range. If so, the deviation judgment result is output and the specific process link corresponding to the abnormality is recorded as the subsequent adjustment target.

[0039] Step S303: Receive the deviation judgment result and combine it with the first analysis result, call the deep adjustment model to perform multi-parameter simulation on the process link where the deviation occurs, calculate the optimal adjustment strategy based on the historical successful experience path and the current batch of raw material structured data set, and output formula adjustment suggestions and process parameter correction suggestions. The formula adjustment suggestions and process parameter correction suggestions include raw material ratio change suggestions, stirring or heating rate correction suggestions, and time node adjustment suggestions.

[0040] Step S303 is used to receive the deviation judgment result and the first analysis result, call the deep adjustment model to perform multi-parameter simulation on the abnormal process node, combine the historical successful path and the current raw material characteristics, and generate the optimal formula adjustment suggestion and process parameter correction suggestion. The output content covers specific execution strategies such as raw material ratio, stirring / heating rate and time node adjustment.

[0041] Step S3 enables real-time sensing of multiple parameters during the production process, identifies process anomalies, and generates intelligent adjustment suggestions. The perception layer collects process data, the control layer identifies deviations, and the decision layer combines raw material deviations with historical experience to output recipes and parameter adjustment suggestions. This enables coordinated control from identifying deviations to generating executable optimization solutions, providing a decision-making basis for dynamic process adjustments.

[0042] Step S4 includes the following sub-steps: Step S401: output a recipe adjustment signal according to the recipe adjustment suggestion, obtain a real-time raw material difference index, correct the target ratio and form a corrected recipe for the current batch, and output the corrected recipe for the current batch as a recipe correction signal.

[0043] Step S401 is used to generate a recipe adjustment signal based on the recipe adjustment suggestion output by the decision layer, and combine the real-time raw material difference index to correct the target ratio and form a corrected recipe for the current batch. The signal is output to the downstream control module as a recipe correction signal to provide a basic basis for parameter adjustment.

[0044] Step S402: update the stirring time, heating temperature and drying time according to the recipe correction signal, collect real-time process data and generate a process parameter execution plan in combination with the recipe correction signal, verify the parameter range and the upper and lower safety limits, and form a corrected process parameter set.

[0045] Step S402 is used to automatically update parameters such as stirring time, heating temperature and drying time of key process units according to the recipe correction signal, generate an executable process parameter execution plan based on the real-time collected process data, and verify its rationality and safety to form the final corrected process parameter set.

[0046] Step S403: After executing the modified process parameter set, monitor the key process parameters. If there is a persistent deviation in the key process parameters or the intermediate product status is abnormal, the current operation process is suspended, buffer mark data is generated, and the self-repair logic mechanism is triggered. The historical optimal parameter set is called for comparison and analysis, and a feasible parameter fallback solution and process bypass strategy are recommended.

[0047] Step S403 is used to continuously monitor key process indicators during the execution of parameter correction. If a persistent deviation or abnormal intermediate product status is detected, the current process is automatically paused, buffer mark data is generated, and self-repair logic is triggered. The historical optimal parameter set is called for comparison and analysis, and a feasible parameter fallback plan or process bypass strategy is output to ensure that the system is restored to a controllable state.

[0048] Step S4 dynamically adjusts the raw material ratio and core process parameters for the current batch based on the system-generated recipe adjustment suggestions and process correction strategies, enabling personalized process adjustments driven by raw material variations. This step automatically executes recipe modifications, parameter updates, and abnormality feedback control to ensure that the production process can consistently deliver intermediate and final products that meet quality requirements despite raw material fluctuations.

[0049] Step S5 includes the following sub-steps: Step S501, collect the production status image data of the intermediate product. The collection process nodes of the production status image data include a mixing node, a heating node and a drying node. The color saturation, appearance morphology contour, flow state and surface viscosity texture of the production status image data are extracted and the intermediate product image feature vector is output.

[0050] Step S501 is used to collect production status image data at key process nodes of intermediate products, extract key visual features such as color saturation, morphological contours, flow state and surface viscosity texture through image processing algorithms, and construct image feature vectors to provide a perceptual basis for quality risk identification.

[0051] Step S502 : Spectral scanning is performed on the intermediate product to extract the active ingredient concentration, impurity distribution, and dissolution uniformity and output them as a chemical state feature vector.

[0052] Step S502 is used to perform near-infrared or multi-channel spectral scanning on the intermediate product to extract parameters such as active ingredient concentration, impurity distribution, and dissolution uniformity that reflect the intrinsic quality of the product, and uniformly represent them as chemical state feature vectors, which together with image features construct a complete quality state description.

[0053] Step S503: Perform quality risk analysis on the image feature vector and chemical state feature vector inputs. The logic of the quality risk analysis is: The image feature vector and chemical state feature vector are compared with the historical high-quality sample database in multiple dimensions to calculate the risk level score of the current intermediate product. If the risk level score is higher than the set threshold, the second analysis result is output and marked as a warning state, and the corresponding image feature vector and chemical state feature vector are recorded.

[0054] If the risk level score is higher than the set threshold, the quality warning mechanism is triggered. The quality warning mechanism sends a stop operation instruction or a slow operation instruction to the process control end and records the corresponding relevant risk level, time node, corresponding intermediate product image characteristics and chemical state data.

[0055] Step S503 is used to input the image feature vector and chemical state feature vector of the intermediate product into the quality risk analysis module, perform multi-dimensional feature comparison with the historical high-quality sample database, and calculate the risk level score of the current sample; when the score is higher than the set threshold, the system outputs the second analysis result and marks it as a warning state, and at the same time triggers the quality warning mechanism, issues an operation instruction to the control end, and records the risk level, time node and related feature data for subsequent tracing and model self-update.

[0056] Step S5 collects dual-channel image and spectral features of the intermediate product during the production process and performs quality risk analysis based on multi-dimensional feature comparison. If the image or composition characteristics of the intermediate product significantly deviate from high-quality standard samples, the system will output a risk level score and actively trigger an early warning mechanism when it exceeds a preset threshold. This allows potential quality issues to be identified before the product enters the final process, ensuring safety and stability in downstream processes.

[0057] Step S6 includes the following sub-steps: Step S601, obtain the final finished product quality inspection results, number the current batch of raw material structured data set, real-time process data set, recipe adjustment suggestions, process parameter correction suggestions, production status image data, chemical state feature vectors and final finished product quality inspection results, build a structured data chain in chronological order, and establish a full-process data chain, which includes the current batch data.

[0058] Step S601 is used to uniformly number the raw material structured data, real-time process data, recipe adjustment suggestions, process parameter correction suggestions, intermediate product status characteristics and final product quality inspection results involved in the current batch, and construct a structured data chain in chronological order of production to form a full-process data chain covering the entire production process, providing data support for subsequent reasoning and optimization.

[0059] Step S602: Perform causal reasoning analysis on the entire process data chain. The causal reasoning analysis is: A multivariate association model is used to identify the source variables of abnormal or fluctuating finished product quality and output the abnormal path, which includes raw material characteristics, process nodes, and adjustment response delays.

[0060] Step S602 is used to perform causal reasoning analysis on the constructed full-process data chain, using a multivariate association model to identify key influencing factors that cause quality fluctuations or anomalies from dimensions such as raw material characteristics, process nodes, and adjustment response delays, output abnormal paths, and provide precise positioning for optimization.

[0061] Step S603: Perform a comparative analysis based on the execution data of the abnormal path and the historical optimal batch, and output the optimized parameter set for the next batch in combination with the system experience rule library. The optimized parameter set includes the recommended initial ratio of the recipe, the target value of the recommended process parameters, and the updated quality warning threshold range, and is synchronized to the next round of process initialization configuration.

[0062] The optimization parameter set includes an initial recipe suggestion generated based on historical high-quality batch data and current abnormal path analysis results. The initial recipe suggestion includes the initial process parameter setting, and the process parameter setting includes the starting control values of time and temperature of the process unit.

[0063] The optimization parameter set also includes dynamic warning thresholds.

[0064] Step S603 compares the abnormal path with the historical optimal batch data, and combines it with the system's empirical rule base to generate a set of optimized parameters for the next batch. This set of optimized parameters includes the recommended initial recipe ratio, target control values for each process unit, and the updated dynamic warning thresholds. These parameters are then synchronized to the process initialization module for the next round of production configuration, ensuring continued improvement in product quality and process stability.

[0065] Step S6 is used to build a full-process data chain for the current batch at the end of the product production closed loop, conduct causal reasoning analysis on the data of various stages in the production, such as raw materials, processes, adjustments, and quality inspection, identify the key paths that lead to fluctuations in finished product quality, and generate an optimized parameter set for the next batch based on comparative analysis, thereby realizing dynamic optimization and knowledge transfer across batches, thereby improving the robustness and yield rate of the entire production line.

[0066] Example 2, reference Figure 2 , provides a health food production line control system based on deep management, including a basic data acquisition module, a current raw material deviation module, an adjustment suggestion output module, an execution adjustment and repair module, an intermediate product identification module and a subsequent batch update module.

[0067] The basic data acquisition module is used to establish a raw material feature database and output a first analysis result as a raw material deviation assessment benchmark.

[0068] The current raw material deviation module is used to collect the current batch of raw material data, compare the differences, and output the raw material deviation results.

[0069] The adjustment suggestion output module is used to collect process data and analyze whether there are any abnormalities based on raw material deviations to generate adjustment suggestions.

[0070] The execution adjustment and repair module is used to execute recipe and parameter adjustments, and trigger the self-repair mechanism in case of abnormalities.

[0071] The intermediate product identification module is used to identify the status of intermediate products, assess the quality risk level and trigger early warning.

[0072] The subsequent batch update module is used to establish the full process data chain, perform causal analysis and generate the next batch optimization parameters.

[0073] The present invention establishes a high-consistency raw material feature database through cluster analysis, realizes multi-dimensional comparison between current raw materials and high-quality templates, supports personalized adjustment of the entire process, division of labor and coordination, integrates raw material and process data for dynamic anomaly identification and adjustment suggestion generation, can automatically adjust recipe ratios and core process parameters according to deviations, and has built-in process rollback and self-repair mechanisms to ensure stable system operation, integrates image recognition and spectral analysis to realize multimodal quality status judgment, identifies potential risks in advance and outputs explainable warnings, builds batch-level data chains and performs causal path analysis, generates initial configurations suitable for the next batch, and improves continuous production quality stability and autonomous optimization capabilities.

[0074] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium may be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0075] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A control method for a health food production line based on in-depth management, characterized in that: The steps include: Step S1, establishing a raw material feature database and outputting a first analysis result as a raw material deviation assessment benchmark; Step S2: Collect the raw material data of the current batch, compare the differences, and output the raw material deviation results; Step S3: Collect process data and analyze whether there are any abnormalities based on raw material deviations, and generate adjustment suggestions; Step S4, executing recipe and parameter adjustment, triggering the self-repair mechanism in case of abnormality; Step S5, identifying the status of the intermediate product, assessing the quality risk level and triggering an early warning; Step S6: Establish a full-process data chain, perform causal analysis, and generate the next batch of optimized parameters.

2. A method for controlling a health food production line based on in-depth management according to claim 1, characterized in that: The step S1 includes the following sub-steps: Step S101: Retrieving raw material data from a raw material database that has been assessed as high-quality batches during historical production, extracting key parameter information for each batch of raw materials, including moisture content, trace element composition, particle size distribution, storage time, and batch number, and performing structured processing on the key parameter information to output a standard data set; Step S102: Cluster and classify the standard data set using a cluster analysis method based on similarity measurement to identify high-consistency raw material categories, form feature vector templates for the high-consistency raw material categories, and construct the feature vector templates and corresponding label data into a raw material feature database. The raw material feature database serves as a reference benchmark, and at the same time, outputs the feature template results of the cluster analysis method as the first analysis result.

3. A method for controlling a health food production line based on in-depth management as claimed in claim 2, characterized in that: The step S2 includes the following sub-steps: Step S201: Collect the moisture content, trace element composition, particle size distribution, storage time, and batch number of the current batch of raw materials at the raw material storage stage, format them, and output them as a structured data set of the current batch of raw materials; In step S202, the structured data set is input into the feature comparison model, and a multi-dimensional feature difference comparison is performed with the corresponding feature vector template in the raw material feature database. A comprehensive difference score is calculated based on the degree of deviation of the feature data, and a judgment is made based on a preset threshold whether it constitutes a significant raw material deviation. The raw material deviation result is output and marked for archiving.

4. A method for controlling a health food production line based on in-depth management as claimed in claim 3, characterized in that: The step S3 includes the following sub-steps: Step S301: Collect temperature, humidity, stirring rate, pH value, flow rate, and viscosity data during the production process, upload the collected results to the edge computing node, and generate a real-time process data set; Step S302: Receive the real-time process data set and the raw material deviation results, perform a difference comparison analysis against the preset standard process execution model, and determine whether there is an abnormal state that deviates from the normal process curve. If there is an abnormal state that deviates from the normal process curve, it is marked as a process deviation, and the deviation judgment result is output and the process link affected by the deviation is recorded; Step S303: Receive the deviation judgment result and combine it with the first analysis result, call the deep adjustment model to perform multi-parameter simulation on the process link where the deviation occurs, calculate the optimal adjustment strategy based on the historical successful experience path and the current batch of raw material structured data set, and output formula adjustment suggestions and process parameter correction suggestions. The formula adjustment suggestions and process parameter correction suggestions include raw material ratio change suggestions, stirring or heating rate correction suggestions, and time node adjustment suggestions.

5. A method for controlling a health food production line based on in-depth management as claimed in claim 4, characterized in that: The step S4 includes the following sub-steps: Step S401: Output a recipe adjustment signal based on the recipe adjustment suggestion, obtain a real-time raw material difference index, correct the target ratio and form a corrected recipe for the current batch, and output the corrected recipe for the current batch as a recipe correction signal; Step S402: updating the stirring time, heating temperature, and drying time according to the recipe correction signal, collecting real-time process data and generating a process parameter execution plan based on the recipe correction signal, verifying the parameter range and the upper and lower safety limits, and forming a corrected process parameter set; Step S403: After executing the modified process parameter set, monitor the key process parameters. If there is a persistent deviation in the key process parameters or the intermediate product status is abnormal, the current operation process is suspended, buffer mark data is generated, and the self-repair logic mechanism is triggered. The historical optimal parameter set is called for comparison and analysis, and a feasible parameter fallback solution and process bypass strategy are recommended.

6. A method for controlling a health food production line based on in-depth management according to claim 5, characterized in that: The step S5 includes the following sub-steps: Step S501, collecting production status image data of the intermediate product, wherein the collection process nodes of the production status image data include a mixing node, a heating node, and a drying node, extracting color saturation, appearance morphology contour, flow state, and surface viscosity texture of the production status image data and outputting an intermediate product image feature vector; Step S502 , performing spectral scanning on the intermediate product, extracting the active ingredient concentration, impurity distribution, and dissolution uniformity, and outputting them as a chemical state feature vector; Step S503: Perform quality risk analysis on the image feature vector and chemical state feature vector inputs. The logic of the quality risk analysis is: The image feature vector and chemical state feature vector are compared with the historical high-quality sample database in multiple dimensions to calculate the risk level score of the current intermediate product. If the risk level score is higher than the set threshold, the second analysis result is output and marked as a warning state, and the corresponding image feature vector and chemical state feature vector are recorded.

7. A method for controlling a health food production line based on in-depth management according to claim 6, characterized in that: If the risk level score is higher than the set threshold, the quality warning mechanism is triggered. The quality warning mechanism sends a stop operation instruction or a slow operation instruction to the process control end and records the corresponding relevant risk level, time node, corresponding intermediate product image characteristics and chemical state data.

8. A method for controlling a health food production line based on in-depth management according to claim 7, characterized in that: The step S6 includes the following sub-steps: Step S601: Obtain the final product quality inspection results. Number the current batch of raw material structured data sets, real-time process data sets, recipe adjustment suggestions, process parameter correction suggestions, production status image data, chemical state feature vectors, and final product quality inspection results. Construct a structured data chain in chronological order to establish a full-process data chain, which includes the current batch data. Step S602: Perform causal reasoning analysis on the entire process data chain. The causal reasoning analysis is: Use a multivariate association model to identify the source variables of abnormal or fluctuating finished product quality and output the abnormal path, which includes raw material characteristics, process nodes, and adjustment response delays; Step S603: Perform a comparative analysis based on the execution data of the abnormal path and the historical optimal batch, and output the optimization parameter set for the next batch in combination with the system experience rule library. The optimization parameter set includes the recommended initial ratio of the recipe, the target value of the recommended process parameters, and the updated quality warning threshold range, and is synchronized to the next round of process initialization configuration.

9. A method for controlling a health food production line based on in-depth management according to claim 8, characterized in that: The optimization parameter set includes an initial recipe suggestion generated based on historical high-quality batch data and current abnormal path analysis results, the initial recipe suggestion includes initial process parameter settings, and the process parameter settings include initial control values of time and temperature of the process unit; The optimization parameter set also includes a dynamic warning threshold.

10. A control system for a health food production line based on depth management, which is applied to a control method for a health food production line based on depth management as claimed in any one of claims 1 to 9, characterized in that: It includes basic data acquisition module, current raw material deviation module, adjustment suggestion output module, execution adjustment and repair module, intermediate product identification module and subsequent batch update module; The basic data acquisition module is used to establish a raw material feature database and output a first analysis result as a raw material deviation assessment benchmark; The current raw material deviation module is used to collect the current batch of raw material data, compare the differences, and output the raw material deviation results; The adjustment suggestion output module is used to collect process data and analyze whether there are abnormalities in combination with raw material deviations to generate adjustment suggestions; The execution adjustment and repair module is used to execute recipe and parameter adjustments and trigger the self-repair mechanism in case of abnormality; The intermediate product identification module is used to identify the status of intermediate products, assess the quality risk level and trigger an early warning; The subsequent batch update module is used to establish a full-process data chain, perform cause-effect analysis and generate the next batch of optimized parameters.

Images