Low-freezing-point oleic acid production quality tracing and early warning system under industrial internet architecture

Abstract

The invention discloses a low-freezing-point oleic acid production quality tracing and early warning system under an industrial internet architecture, and relates to the technical field of grease chemical industry production quality management and control, and the technical key points are as follows: distributing a global unique batch identification code conforming to an industrial internet identification analysis system for each independent production batch; irreversible strong binding is carried out on identification codes, whole-process production source data, influence weights of all nodes on finished product freezing points and finished product core quality index measured data through a Hash encryption algorithm, and it is ensured that traceability data cannot be tampered; meanwhile, a tracing index with a freezing point influence factor as a correlation dimension is established by taking a batch identification code as a core, and the production data and the finished product quality index are deeply correlated, so that the method is not limited to sequential storage and simple query of the production data; specific process nodes and single group of production data which cause anomalies can be directly positioned based on a tracing link, and the checking and handling period of quality problems is shortened.

Description

Technical Field

[0001] This invention relates to the field of quality control technology in oleochemical production, and in particular to a traceability and early warning system for low-freezing-point oleic acid production quality under the industrial internet architecture. Background Technology

[0002] Low-freezing-point oleic acid is a core raw material in high-end lubricating oil base oils, food emulsifiers, cosmetic active ingredients, and pharmaceutical intermediates. Its core quality indicator is the freezing point. High-end applications in the industry typically require a freezing point ≤5℃, which places extremely high demands on the stability and precision of the production process. The production of low-freezing-point oleic acid is a continuous chemical production process. The core steps include raw material pretreatment, hydrolysis reaction, hydrogenation reaction, distillation separation, freeze crystallization, pressure filtration purification, and finished product filling. Fluctuations in the process parameters of each step will have a conductive impact on the freezing point index of the final product, and the influence weight of different steps varies significantly.

[0003] However, during the implementation of the above technical solution, at least the following technical problems were discovered: Firstly, the system uses the final freezing point detection of the finished product as the core control point, but it fails to quantify the transmission effect of quality fluctuations between various processes, nor does it establish a correlation prediction model between the process parameters and the freezing point index of the finished product. It can only achieve alarms for parameters exceeding limits in a single process, and cannot predict the comprehensive impact of process fluctuations on the final freezing point of the finished product during the production process. In other words, in actual production, batch non-conformity issues can only be discovered after the finished product has completed the entire production process and the freezing point detection results are issued. At this time, the raw materials, energy, labor, and equipment capacity of the entire batch have already been invested. The rework and scrapping of non-conforming products will cause a huge waste of production costs, and will also lead to delays in production delivery cycles. It cannot meet the production requirements of high stability and high pass rate of high-end low freezing point oleic acid. Secondly, existing traceability systems can only achieve simple storage and post-event query of batch production data, without deeply binding production data with the impact weight of the finished product's freezing point. This results in a disconnect between traceability data and quality control requirements. When the finished product's freezing point exceeds the standard, the traceability system cannot directly locate the core process and key parameters causing the anomaly. Technical personnel must manually investigate each process and parameter, with the root cause identification cycle taking several hours to several days, leading to extremely low efficiency in handling quality issues. At the same time, existing systems cannot iteratively optimize process standards using historical production data from qualified batches, nor can they improve control rules using data from quality anomalies. They can only adjust the process through repeated manual trials, resulting in long optimization cycles and high trial-and-error costs. This makes it impossible to achieve self-optimization of the production process and continuous improvement of product quality. Therefore, we propose a low-freezing-point oleic acid production quality traceability and early warning system under the industrial internet architecture. Summary of the Invention

[0004] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a quality traceability and early warning system for low-freezing-point oleic acid production under an industrial internet architecture, which solves the core technical problems of existing technologies, such as post-event control, lack of quality correlation in traceability, rigid data collection mode, lack of early warning capability, and lack of iterative capability.

[0005] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: The Industrial Internet architecture-based low-freezing-point oleic acid production quality traceability and early warning system includes the following components: The dynamic acquisition module is used to deploy edge acquisition terminals at each process node of the entire low-freezing-point oleic acid production process through an industrial internet architecture. It pre-constructs the influence weight matrix of each process on the core quality indicators of the product, and sets dynamic differentiated acquisition rules for the production source data of each process node according to the influence weight matrix, and synchronously collects multi-source heterogeneous production source data of the entire low-freezing-point oleic acid production process. The traceability link construction module is used to assign a globally unique batch identification code to each independent production batch of low freezing point oleic acid. The batch identification code is irreversibly and strongly bound to the production source data of each process node in the entire production process of that batch and the influence weight of each process on the core quality indicators of the product. A full-process traceability index is established with the batch identification code as the retrieval subject and the influence factor of the core quality indicators of the product as the correlation dimension, and traceability link information is generated. The coupling analysis module is used to perform coupling analysis on the quality fluctuations of each process node based on the production source data corresponding to the full-process traceability information and the pre-constructed quality risk transmission coefficient matrix of the entire process of low-freezing-point oleic acid production, and to calculate the predicted value of the core quality indicators of the product and the quality assessment value of the entire process. The early warning module is used to calculate the quality risk judgment value based on the predicted value of the product's core quality indicators, the quality assessment value of the whole process, and the deviation of the quality fluctuation at each process node. The control iteration module is used to determine whether the quality risk judgment value exceeds the preset graded control threshold range. If the quality risk judgment value exceeds the preset graded control threshold range, it is determined to be a quality abnormality state. According to the risk transmission level of the abnormal process node, the corresponding level of early warning response and production process control action are triggered. Otherwise, it is determined to be a quality qualified state, and the entire process production data is stored in the core production database for iterative optimization of the low freezing point oleic acid production process.

[0006] Preferably, the dynamic acquisition module includes: The raw material data acquisition unit is used to collect information on the source, batch, core component testing data, basic quality index data, and warehousing and storage environment data of the raw materials used in the production of low-freezing-point oleic acid, and to generate raw material attribute and component data. The key process data acquisition unit is used to collect real-time operating parameters of each key process in the entire production process of low-freezing-point oleic acid that has a direct impact on the core quality indicators of the product, and generate full-process production process parameter data. The equipment status acquisition unit is used to collect the operating status parameters, fault alarm information, maintenance records and operating accuracy data of the production equipment corresponding to the above key processes, and generate key equipment operating status data. The environmental data acquisition unit is used to collect real-time environmental parameters in the core process workshops of low-freezing-point oleic acid production and the finished product storage area, and generate full-scenario environmental monitoring data. The quality inspection and data acquisition unit is used to collect the intermediate control semi-finished product inspection data and the finished product full quality index inspection data during the production process of low freezing point oleic acid, and to generate process and finished product quality inspection data. The logistics data acquisition unit is used to collect outbound information, warehousing and transportation environment data, delivery information and terminal feedback data of low-freezing-point oleic acid products, and generate warehousing and logistics data. The dynamic data acquisition and control unit is used to construct an impact weight matrix of each process on the core quality indicators of the product based on historical production big data. The initial acquisition cycle and acquisition accuracy are set for each process node according to the impact weight matrix. At the same time, the fluctuation range of process parameters of each process node is monitored in real time. When the fluctuation of process parameters of a certain process node exceeds the preset stability threshold, the acquisition frequency and acquisition accuracy of the production source data corresponding to that process node are increased.

[0007] Preferably, the dynamic acquisition module also includes a full-link data standardization processing unit; The end-to-end data standardization processing unit is used to perform format standardization, outlier removal, and missing value completion processing on all types of production source data collected. According to the process nodes and material flow sequence of the entire production process of low freezing point oleic acid, all production source data are aligned with the process timestamp and material batch dual benchmarks to ensure that all production source data within the same process node are at the same time benchmark and material batch benchmark. The various production source data that have undergone standardization are classified and encrypted for storage according to batch identification codes and process nodes, providing a standardized data source for the construction of a full-process traceability chain and coupled analysis of quality risks.

[0008] Preferably, the traceability link construction module specifically includes: This is used to assign a globally unique batch identification code that conforms to the Industrial Internet Identifier Resolution System to each independent production batch of low freezing point oleic acid. The batch identification code is then irreversibly encrypted and bound to the corresponding batch's raw material batch information, production source data of each node in the entire process, the influence weight of each node on the core quality indicators of the product, and the measured data of the core quality indicators of the finished product through an encryption algorithm. Using batch identification codes as the core, and following the sequence of production processes, all production source data for the corresponding batch are sequentially associated according to process nodes, establishing a full-process traceability index with batch identification codes as the retrieval subject and product core quality indicator influencing factors as the association dimension. Based on the full-process traceability index, traceability link information covering the entire product lifecycle is generated. The traceability link information supports forward full-process traceability and reverse root cause localization. Forward traceability can fully display the full lifecycle information of the corresponding batch of products, while reverse root cause localization can directly locate the specific process node and single set of production source data that caused the quality abnormality based on the abnormal results of the core quality indicators of the finished product.

[0009] Preferably, the traceability link construction module also includes a multi-scenario batch traceability query unit; The multi-scenario batch traceability query unit supports hierarchical query services for entities with different permissions. It opens up traceability data query permissions according to the query needs of different entities. At the same time, it can generate standardized traceability reports and product quality compliance certificates that meet the corresponding requirements according to the needs of the querying entity.

[0010] Preferably, the coupling analysis module specifically includes: Used to extract standardized production source data for each node of the entire process of a corresponding production batch based on the traceability information of the entire process chain; Based on the quality risk transmission coefficient matrix of the entire process of low-freezing-point oleic acid production, which is pre-constructed using historical big data of production, the quality risk transmission coefficient matrix includes the quality fluctuation transmission coefficient between each process node and the direct impact coefficient of each process node on the core quality indicators of the product. By using a coupled analysis algorithm, the transmission effect of process parameter fluctuations at each process node on subsequent processes and the comprehensive impact on the core quality indicators of the final product are calculated, and the predicted values ​​of the core quality indicators of the corresponding production batch are generated. At the same time, based on the production source data of each node in the entire process, multi-dimensional quality assessment indicators for the corresponding production batch are calculated. By combining the predicted values ​​of the product's core quality indicators with multi-dimensional quality assessment indicators, the full-process quality assessment value of the corresponding production batch is obtained through weighted fusion calculation.

[0011] Preferably, the coupling analysis module also includes an intelligent identification unit for the root causes of quality anomalies; The intelligent identification unit for the root causes of quality abnormalities is used to pre-build a knowledge base for the root causes of quality abnormalities in low-freezing-point oleic acid. The knowledge base includes abnormal processes, abnormal parameters, root cause analysis logic and optimization solutions corresponding to different quality abnormalities. The multi-dimensional quality assessment indicators of the corresponding production batch are compared with the preset standard indicator range to identify abnormal assessment indicators that exceed the standard range and locate the corresponding abnormal process nodes and production source data. By combining the quality risk transmission coefficient matrix, we analyze the impact path and degree of quality fluctuations at abnormal process nodes on the core quality indicators of the product. We then match the knowledge base of root causes of quality anomalies to identify the core root causes of quality anomalies and generate process optimization solutions simultaneously.

[0012] Preferably, the early warning module specifically includes: Used to obtain the preset standard range of process parameters for each node in the entire process of low freezing point oleic acid production, and to calculate the deviation of the quality fluctuation between the real-time production source data of each process node and the standard range. Based on the predicted values ​​of the product's core quality indicators and the national quality standards and internal control quality standards of low-freezing-point oleic acid products, the deviation risk values ​​of the core quality indicators are calculated. By combining the overall quality assessment value, the quality fluctuation deviation of each process node, the deviation risk value of the core quality indicators, and the influence weight of each process node on the core quality indicators of the product, the quality risk discrimination value of the corresponding production batch is calculated. Simultaneously, at any process node in the production process, based on the production source data of that node and the preceding process, the predicted values ​​of the product's core quality indicators and the judgment values ​​of quality risks can be updated in real time.

[0013] Preferably, the early warning module also includes an adaptive iteration unit for the control system; The adaptive iteration unit of the control system is used to acquire real-time data on the status of production equipment for low-freezing-point oleic acid, data on changes in raw material composition, data on changes in the production environment, data on adjustments to the production process, and data on updates to product quality standards. Based on the real-time data obtained above, machine learning algorithms are used to adaptively and iteratively update the whole process quality risk transmission coefficient matrix, the influence weight of each process node on the core quality indicators of the product, and the hierarchical control threshold range.

[0014] Preferably, the control iteration module specifically includes: This is used to pre-set a hierarchical control threshold range containing multiple levels of thresholds. The hierarchical control threshold range is set based on the national product quality standard for low freezing point oleic acid, the enterprise's internal control quality standard, and historical production big data. The real-time calculated quality risk discrimination value is compared with the graded control threshold range. Based on the threshold range where the quality risk discrimination value is located, the quality abnormality level is determined and the corresponding control action is triggered. The control actions include early warning prompts, process parameter adjustments, suspension of high-risk processes, and emergency shutdown of the entire process. For production batches that meet quality standards, the full-process production source data, traceability information, and quality analysis results of the corresponding batch are stored in the core production database. The quality risk transmission coefficient matrix, process parameter standard ranges, and hierarchical control threshold ranges are iteratively optimized through machine learning algorithms.

[0015] (III) Beneficial Effects 1. Based on historical big data of low-freezing-point oleic acid production, a full-process quality risk transmission coefficient matrix is ​​constructed to clarify the transmission law of quality fluctuations between processes and the direct impact weight on the freezing point of the finished product. Then, through a coupling analysis algorithm, the transmission impact of process parameter fluctuations in each process on subsequent processes and the comprehensive effect on the final product freezing point are calculated in real time. At any process node in the entire production process, the predicted value of the frozen point of the finished product can be updated in real time based on the production data of the completed processes, and the risk of exceeding the freezing point standard can be identified without waiting for the final inspection of the finished product. This breaks the traditional control logic of determining the result by finished product inspection and handling non-conforming products after the fact in the industry. It can intervene in quality risks in advance during the production process, avoid ineffective input of raw materials, energy, labor and production capacity from the root, and significantly reduce the cost of production quality control. 2. Assign a globally unique batch identification code that conforms to the Industrial Internet Identifier Resolution System to each independent production batch. Through a hash encryption algorithm, the batch identification code is irreversibly and strongly bound to the production source data of the entire process of that batch, the influence weight of each node on the freezing point of the finished product, and the measured data of the core quality indicators of the finished product, ensuring that the traceability data cannot be tampered with. At the same time, with the batch identification code as the core, establish a full-process traceability index with the freezing point influence factor as the correlation dimension, deeply linking production data with the core quality indicators of the finished product. The traceability system is no longer limited to the time-series storage and simple query of production data. When a finished product quality abnormality occurs, the specific process node and single set of production data that caused the abnormality can be directly located based on the traceability link, which greatly shortens the investigation and handling cycle of quality problems. 3. First, an influence weight matrix of each process on the freezing point of the finished product is constructed using orthogonal experiments and grey relational analysis. Differentiated initial data collection cycles and accuracy are set for different processes. During the production operation phase, the fluctuation range of process parameters in each process is monitored in real time. When the parameter fluctuation of a certain process exceeds the preset stability threshold, the collection frequency and accuracy of the corresponding production source data for that process are increased. Through the implementation of this technical solution, the complete capture of transient data on process fluctuations is achieved for key processes such as freezing crystallization and hydrogenation reaction, which have extremely high influence weight on the freezing point, ensuring the integrity of core quality data. At the same time, the collection frequency is reduced for non-critical processes such as raw material pretreatment and finished product filling, avoiding the generation of a large amount of redundant data, reducing the occupation of system storage and computing resources, and improving the overall operating efficiency of the system.

[0016] 4. Through the adaptive iterative unit of the control system, real-time data on the operating status and accuracy degradation of production equipment, changes in the fatty acid composition of raw materials, seasonal environmental temperature and humidity changes, production process adjustment data, and product quality standard updates are acquired. Based on the above real-time data of all elements, the quality risk transmission coefficient matrix, process influence weight, and hierarchical control threshold range are continuously and adaptively updated using the random forest machine learning algorithm. This solves the core defect of existing technologies where fixed thresholds cannot adapt to changes in production conditions. The control system can be dynamically optimized with changes in all production elements, while effectively avoiding false alarms and missed alarms caused by changes in production conditions, ensuring the control accuracy and stability throughout the entire production cycle.

[0017] 5. Pre-set multi-level hierarchical control threshold ranges covering early warning, control, and shutdown. Match corresponding control actions based on real-time calculated quality risk assessment values, achieving full-level cyclical control from early warning prompts, cyclical adjustment of process parameters, suspension of high-risk processes to emergency shutdown of the entire process. Simultaneously, for qualified production batches, their entire production source data, traceability information, and quality analysis results are stored in the core production database. Through machine learning algorithms, the quality risk transmission model, process parameter standard ranges, and control thresholds are continuously iterated and optimized. Production processes can be continuously optimized using historical qualified batch production data, achieving continuous improvement in product quality stability while significantly shortening the process optimization cycle and reducing R&D costs. Attached Figure Description

[0018] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0019] Figure 1 This is an overall flowchart of an embodiment of the present invention; Figure 2 This is a flowchart illustrating the risk assessment process in an embodiment of the present invention. Figure 3 This is a flowchart of an embodiment of the present invention. Detailed Implementation

[0020] This application provides a traceability and early warning system for low-freezing-point oleic acid production quality under an industrial internet architecture. It addresses the core technical problems of existing technologies, such as post-production control, lack of quality correlation in traceability, rigid data collection methods, lack of early warning capabilities, and lack of iterative processing. Each independent production batch is assigned a globally unique batch identification code conforming to the industrial internet identifier resolution system. A hash encryption algorithm irreversibly binds the batch identification code to the entire production process source data of that batch, the influence weight of each node on the frozen point of the finished product, and the measured data of the core quality indicators of the finished product, ensuring that the traceability data is tamper-proof. Simultaneously, based on the batch identification code, a full-process traceability index is established with the frozen point influence factor as the correlation dimension, deeply linking production data with the core quality indicators of the finished product. The traceability system is no longer limited to the time-series storage and simple querying of production data. When an abnormality in the quality of the finished product occurs, the specific process node and single set of production data causing the abnormality can be directly located based on the traceability link, significantly shortening the investigation and handling cycle of quality problems.

[0021] Example: Reference Figures 1 to 3 As shown, the technical solution in this application embodiment addresses the core technical problems of existing technologies, such as post-event control, lack of quality correlation in traceability, rigid data collection modes, lack of early warning capabilities, and lack of iterative capabilities. The overall approach is as follows: To address the problems existing in the prior art, this invention provides a low-freezing-point oleic acid production quality traceability and early warning system under the industrial internet architecture. The system is based on a three-layer architecture of industrial internet edge acquisition layer, platform service layer, and application management layer. The edge acquisition layer is deployed on the production workshop site, the platform service layer is deployed in the enterprise's private cloud data center, and the application management layer is open to users in all scenarios. The three layers are interconnected using industrial Ethernet, and the transmission protocol adopts the standard Modbus and Profinet protocols.

[0022] To enable traceability and early warning analysis of low-freezing-point oleic acid, the system mainly consists of five modules: a multi-source heterogeneous production data dynamic acquisition module, a full lifecycle traceability link construction module, a quality risk transmission coupling analysis module, a quality anomaly early warning module, and a full-link cyclical control and iteration module. This forms a complete cycle from data acquisition, traceability analysis, early warning control to process iteration, covering the entire product lifecycle of low-freezing-point oleic acid production, from raw material entry, production and processing, finished product testing, warehousing and logistics to delivery to end customers. Among them, the multi-source heterogeneous production data dynamic acquisition module provides underlying data support for the system and subsequent analysis. It mainly uses industrial-grade edge acquisition terminals deployed at each process node of the entire production process, and then constructs a weight matrix of the impact of each process on the core quality indicators based on the enterprise's mass production history big data for more than 3 consecutive years. Then, according to the weight matrix, dynamic differentiated acquisition rules are set for each process node to complete the full coverage acquisition and standardized processing of multi-source heterogeneous production data throughout the entire process. This method provides a unified and highly reliable data source for subsequent modules. The full lifecycle traceability link construction module is the core of data traceability. By assigning a globally unique batch identification code that conforms to the Industrial Internet Identifier Resolution System to each independent production batch, the batch identification code is irreversibly and strongly bound to the full-process production source data of that batch, the influence weight of each process on the core quality indicators, and the measured data of the core quality indicators of the finished product. This builds a searchable traceability index for the entire process and generates a full lifecycle traceability link for the product. At the same time, it realizes forward full-process traceability and reverse quality root cause positioning, breaking through the industry pain point that traditional traceability systems can only record data and cannot associate quality impact. The quality risk transmission coupling analysis module is mainly used to analyze the acquired data. It adopts standardized production source data extracted based on the full life cycle traceability link, combined with the pre-constructed quality risk transmission coefficient matrix of the entire process of low freezing point oleic acid production, to conduct cross-process coupling analysis on the fluctuation of process parameters at each process node. It quantifies the direct impact of single process fluctuations, cross-process transmission impacts, and multi-process superposition impacts, calculates and generates the predicted values ​​of the core quality indicators of the finished product and the quality assessment values ​​of the whole process, and provides scientific data analysis support for early warning of quality anomalies, so as to facilitate intuitive detection and analysis. The quality anomaly early warning module combines the predicted values ​​of core quality indicators, the quality assessment values ​​of the whole process, and the deviation of quality fluctuations at each process node. Then, it uses a multi-factor weighted comprehensive algorithm to calculate and generate the quality risk judgment value of the corresponding production batch. This enables dynamic risk assessment and early warning of finished product quality anomalies at any node in the entire production process, solving the problem of the lagging control logic of traditional post-production detection and remediation. The end-to-end cyclical control and iteration module is the core of the system control. It pre-sets hierarchical control threshold ranges, determines the quality anomaly level based on the real-time calculated quality risk discrimination value, and triggers corresponding cyclical control actions. At the same time, it stores the full-process data of qualified batches into the core production database, and uses machine learning algorithms to achieve continuous iterative optimization of process parameter standards, analysis models, and control systems.

[0023] In the above, the data acquisition module provides the basic data source for the other modules; the tracing module provides data retrieval, time-series alignment, and location support for the analysis module; the analysis module provides core analysis results and predictive data support for the early warning module; the early warning module provides risk assessment criteria and early warning trigger signals for the control module; and the process optimization results from the control module feed back into the acquisition rules of the acquisition module and the model parameters of the analysis module, achieving dynamic optimization and continuous upgrading of the entire system. Specifically: Firstly, the multi-source heterogeneous production data dynamic acquisition module comprises seven units: raw material data acquisition unit, key process data acquisition unit, equipment status acquisition unit, environmental data acquisition unit, quality inspection data acquisition unit, logistics data acquisition unit, and dynamic acquisition and control unit. It primarily operates collaboratively around the goals of full-process coverage, multi-dimensional acquisition, differentiated control, and standardized output, so that all collected data can ultimately be converged into the dynamic acquisition and control unit for unified management. Specifically: The raw material data acquisition unit covers the origin, supplier qualifications, batch number, warehousing time, storage tank number, temperature and humidity data throughout the storage cycle, full component testing data of fatty acids, acid value, iodine value, moisture, impurities, peroxide value and other basic quality indicators of raw materials used in the production of low-freezing-point oleic acid. All data are collected at once when raw materials are put into storage, and storage environment data are collected once every hour. The collected data is synchronized through the connection with the enterprise ERP system and LIMS laboratory system. Finally, standardized raw material attribute and component data are output, providing full data support for raw material compliance assessment, process parameter preset and raw material traceability. After the data is collected, it is synchronized to the full life cycle traceability link construction module. The key process data acquisition unit covers core processes including raw material pretreatment, hydrolysis reaction, hydrogenation reaction, distillation separation, freeze crystallization, pressure filtration purification, and finished product filling. The acquired data includes real-time temperature, pressure, liquid level, flow rate, stirring rate, reaction time, reflux ratio, and feed rate of core production equipment such as reactors, distillation columns, freeze crystallizers, filter presses, and filling machines. The acquisition frequency matches the differentiated rules set by the dynamic acquisition and control unit, with a minimum acquisition frequency of no less than 10 seconds per acquisition for core processes. The acquired data is synchronized in real time through an edge gateway to the PLC controllers of each process. The final output is standardized full-process production parameter data, providing core data support for quality risk analysis, process control, and anomaly early warning. The acquired data is synchronized in real time to the quality risk transmission coupling analysis module and the quality anomaly early warning module. The equipment status acquisition unit covers the entire range of equipment status data, including operating speed, operating temperature, operating pressure, start / stop status, fault alarm codes, maintenance records, operating accuracy decay data, equipment service life, and motor load rate, for all production equipment corresponding to the seven core processes. The acquisition frequency is consistent with the acquisition frequency of the process parameters for the corresponding processes. Fault alarm data is triggered for acquisition and uploading in real time. The acquired data is synchronized with the equipment management system and the equipment's built-in sensors, and finally outputs standardized key equipment operating status data to identify the impact of equipment status fluctuations on product quality. This provides data support for preventive maintenance and dynamic adjustment of process parameters. The acquired data is synchronized in real time to the quality anomaly early warning module. The environmental data acquisition unit covers real-time environmental parameters such as temperature, humidity, cleanliness, ventilation status, and nitrogen protection pressure in core process workshops such as hydrogenation reaction workshop and frozen crystallization workshop, finished product storage warehouse, and cold chain transportation. The data acquisition frequency for workshops and warehouses is once per minute, and the data acquisition frequency for cold chain transportation is once every 10 minutes. The acquired data is synchronized by connecting to workshop environmental monitoring terminals, warehouse temperature and humidity monitoring systems, and cold chain transportation GPS terminals. The final output is standardized full-scenario environmental monitoring data, which is used to eliminate the interference of environmental fluctuations on the core quality indicators of the product. The acquired data is synchronized to the quality risk transmission coupling analysis module. The quality inspection and data collection unit covers the full range of test data for semi-finished products in each process, as well as the full range of quality indicators for finished products, including freezing point, acid value, iodine value, oleic acid purity, color, peroxide value, and oxidative stability. It also collects auxiliary information such as the operating status of testing equipment, testing time, testing personnel, and testing method standards. The data is synchronized after each process is completed, and the data for finished products is synchronized within 10 minutes after testing. The collected data is synchronized by connecting to the gas chromatograph, fully automatic freezing point tester, and Lovibond colorimeter in the central control laboratory and the finished product laboratory. Finally, it outputs standardized process and finished product quality inspection data, providing direct evidence for quality assessment, anomaly identification, and model verification. The collected data is synchronized to the full life cycle traceability link construction module and the full link cycle management and iteration module. The logistics data collection unit covers the entire process of data collection, including finished product outbound time, outbound quantity, storage location, temperature and humidity data throughout the cold chain transportation, transportation routes, transportation vehicle information, distributor information, end customer information, and end customer quality feedback data. Outbound and inbound data are synchronized in real time, transportation process data is synchronized once every 10 minutes, and feedback data is synchronized immediately upon receipt. The collected data is synchronized through the connection with the WMS warehouse management system and TMS transportation management system, and finally outputs standardized warehousing and logistics data to improve the product life cycle traceability. The collected data is synchronized to the full life cycle traceability link construction module. The dynamic data acquisition and control unit is the control structure of this module. It is mainly based on the historical big data of mass production of no less than 1,200 batches accumulated by the enterprise over three consecutive years. Through orthogonal experiments and grey relational analysis, it constructs a weight matrix of the impact of each process on the core quality indicators. Based on the weight matrix, it sets differentiated initial acquisition cycles and acquisition accuracies for each process. At the same time, it monitors the fluctuation range of process parameters in real time and presets process stability thresholds for the core parameters of each process. When the core process parameter of a certain process node exceeds the preset stability threshold for three consecutive acquisition cycles, the acquisition cycle of that process node is adjusted to 2 seconds / time, and the acquisition accuracy is improved to two decimal places. When the parameter returns to the stable threshold range and remains so for 10 minutes, the initial acquisition rules are restored. This achieves targeted allocation of acquisition resources, ensuring complete capture of data at the core quality impact nodes while avoiding invalid data redundancy in non-critical processes. It also coordinates the start and stop of acquisition, data transmission, and anomaly handling of the other six acquisition units.

[0024] In addition, the multi-source heterogeneous production data dynamic acquisition module is also equipped with a full-link data standardization processing unit to standardize all types of collected data, ensuring the accuracy of subsequent module analysis results; the specific steps are as follows: Firstly, for the collected raw material data, process parameter data, equipment status data, environmental data, quality inspection data, and logistics data, three basic processing steps are performed: format standardization, outlier removal, and missing value completion. Format standardization converts heterogeneous data from different acquisition terminals and devices into a system-compatible standardized format, unifying the units of measurement and time format of all data. The units of measurement follow national legal measurement standards, and the time format is standardized as YYYY-MM-DDHH:mm:ss, eliminating differences in dimensions and time sequence formats. Outlier removal uses the 3σ principle for identification. Abnormal data caused by sensor failure, network fluctuations, equipment malfunctions, etc. is removed. The mean μ and standard deviation σ of the data sequence are calculated, and outliers exceeding the range [μ-3σ, μ+3σ] are removed. An outlier removal logs are kept, and the reasons for the anomalies are marked. The removed data undergoes a second verification to avoid the wrong removal of valid data. For missing data caused by network interruption, temporary sensor failure, etc., linear interpolation of time-series data in the same process is used to complete the missing data to ensure the continuity of the data sequence. For data that is missing for more than 3 consecutive periods, a data anomaly warning is triggered, and maintenance personnel are notified to conduct on-site verification. After basic processing, all production source data is aligned with both process timestamps and material batches according to the process nodes and material flow sequence of low-freezing-point oleic acid production. Using process nodes as units and the standard time synchronized by the NTP network time server as the benchmark, all data from the same batch and process are unified to the same time and batch benchmarks, with a time synchronization error ≤10ms. This eliminates time and batch deviations between different acquisition terminals, achieving data correlation and time sequence unification. Finally, this unit categorizes and encrypts the standardized production source data according to batch identification codes and process nodes, using the AES-256 encryption algorithm. The storage architecture is linked to the core production database, clearly defining storage paths, retrieval rules, and access permissions, outputting standardized and regulated data sources to provide core data support for all subsequent modules.

[0025] Secondly, the full lifecycle traceability link construction module follows the core principles of unique batch identification, immutable data, and full-link traceability. Before batch feeding, this module assigns a globally unique batch identification code that conforms to the national industrial internet identifier resolution system to each independent production batch. The code structure is fixed as "national code - enterprise code - industry code - production year, month and date - batch serial number". The coding rules follow the relevant standards of the national industrial internet identifier resolution system to ensure the uniqueness, universality and industry compatibility of the code nationwide, providing a unique identification basis for full batch traceability. After the coding is assigned, this module uses the SHA-256 hash algorithm to irreversibly encrypt and bind the batch identification code with the corresponding batch of raw material batch information, production source data of each node in the entire process, the influence weight of each node on the core quality indicators, and the measured data of the core quality indicators of the finished product, generating a unique hash value. The hash value is then written to the Hyperledger Fabric consortium chain for blockchain notarization. The hash value is updated and notarized on the chain once after each production process is completed, which technically prevents data tampering. Any data tampering will result in a hash value mismatch. The system identifies and intercepts tampering operations to ensure the authenticity, integrity and immutability of traceability data. After data binding is completed, this module uses batch identification codes as the core and, according to the sequence of low-freezing-point oleic acid production processes, sequentially associates all production source data of the corresponding batch by process nodes. Each data node is marked with key information such as corresponding process information, collection time, and impact weight on core quality indicators. A full-process traceability index is established with batch identification codes as the retrieval subject and impact factors of core quality indicators as the association dimension. The index supports fast retrieval by multiple dimensions such as batch code, process node, parameter type, impact weight, and production time, with a retrieval response time of ≤100ms. Ultimately, this module generates traceability information covering the entire lifecycle of a product, from raw material entry, production and processing, finished product testing, warehousing and logistics, to final delivery, based on a full-process traceability index. This information can pinpoint individual production source data for each process, eliminating data gaps. It also achieves two core functions: first, forward full-process traceability, which uses batch identification codes to completely reconstruct the entire production, quality testing, and warehousing and logistics information of a product from raw material entry to final delivery, enabling transparent traceability throughout the product's lifecycle; second, reverse root cause localization, which, based on the abnormal detection results of core quality indicators of the finished product, sorts the impact weight of each process on the core quality indicators from high to low, directly locating the specific process node and individual production source data that caused the quality abnormality, with a localization time of ≤10 seconds.

[0026] In addition to the above, the full lifecycle traceability link construction module also includes a multi-scenario batch traceability query unit to meet the traceability query needs of different entities, balancing data security and ease of use. This unit sets differentiated hierarchical query permissions and account management mechanisms for four different permission-based entities: manufacturers, regulatory agencies, distributors, and end customers. Each entity has a unique account password and permission identifier. Permission applications, changes, and cancellations all require an approval process, and logs are maintained throughout the entire process. Manufacturers, as the production control entity, are granted full access to traceability data and some operational permissions, allowing them to query, modify, and export data throughout the entire process. Internal quality control, process optimization, and problem identification needs; regulatory agencies, as the main body of industry supervision, are granted read-only query and report export permissions for production records, quality testing data, traceability links, and other data related to compliance inspections to meet regulatory and compliance inspection needs, while also supporting standardized integration with local regulatory platforms; distributors and end customers, as the main bodies of product circulation and use, are granted read-only query and traceability certificate generation permissions for data such as raw material source information, finished product quality qualification reports, cold chain logistics information, and basic batch traceability information to meet product traceability and quality verification needs, while desensitizing non-essential publicly available data to ensure the security of enterprise business information; This unit can generate standardized traceability reports and product quality compliance certificates that meet the corresponding requirements based on the needs of different query subjects. For regulatory agencies, it generates standardized compliance testing reports that comply with national regulations, including core content such as full-process production data, quality testing results, and traceability information, with the format fully compliant with regulatory requirements. For distributors and end customers, it generates concise and clear product traceability certificates, including key content such as raw material sources, core process parameters, finished product quality certificates, and cold chain logistics information. The reports and certificates can be viewed, downloaded, and printed online, and can also generate electronically signed documents with anti-counterfeiting features to ensure the authenticity and immutability of the documents.

[0027] Third, the quality risk transmission coupling analysis module enables early prediction of finished product quality indicators and comprehensive assessment of production quality status through cross-process coupling analysis. First, it extracts standardized production source data from each node of the entire process of the corresponding production batch from the full-process traceability information. The extraction scope covers all categories of data such as raw material data, process parameter data, equipment status data, environmental data, quality inspection data, and logistics data. The extracted data undergoes a second integrity and accuracy verification to ensure that the data is unbiased, providing basic data support for coupling analysis. Subsequently, based on the historical big data of 1200 batches of qualified low-freezing-point oleic acid produced by the enterprise over three consecutive years, this module constructed a quality risk transmission coefficient matrix for the entire low-freezing-point oleic acid production process using partial least squares regression. The matrix contains two types of core coefficients. One type is the quality fluctuation transmission coefficient between each process node, used to quantify the impact of fluctuations in process parameters of the preceding process on the production process and quality indicators of the subsequent process. This includes the transmission coefficient of hydrogenation reaction temperature on distillation components (0.72), the transmission coefficient of distillation components on freezing point of crystallization (0.85), and the transmission coefficient of hydrolysis reaction conversion rate on hydrogenation effect (0.68), etc., which are transmission coefficients for the entire process. The other type is the direct impact coefficient of each process node on the core quality indicators of the product. This is completely consistent with the impact weight matrix constructed by the multi-source heterogeneous production data dynamic acquisition module, namely, freezing crystallization process 35%, hydrogenation reaction process 30%, distillation separation process 20%, hydrolysis reaction process 8%, raw material pretreatment process 4%, pressure filtration purification process 2%, and finished product filling process 1%. After the matrix was constructed, it was validated with 200 batches of independent historical data, providing core model support for coupled analysis. Based on the quality risk transmission coefficient matrix, this module conducts a comprehensive analysis of the fluctuations in process parameters at each process node through a coupling analysis algorithm. It not only quantifies the direct impact of fluctuations in single process parameters on core quality indicators, but also calculates the transmission impact of fluctuations in each process parameter on subsequent processes, as well as the combined impact of the superposition of fluctuations in multiple process parameters. The analysis process and data acquisition proceed simultaneously. After the parameter acquisition of each process node is completed, a coupling analysis is immediately carried out to achieve real-time dynamic updates of the analysis process. Through coupling analysis, this module finally generates the predicted values ​​of the core quality indicators for the corresponding batch. The predicted values ​​are calculated based on the pre-trained multiple linear regression model. The model uses the core process parameters of each process as independent variables and the measured value of the frozen point of the finished product as dependent variables. It is trained with historical data, and the model fit is R2≥0.98. The deviation between the predicted value and the final measured value is controlled within ±0.3℃, and it can be updated in real time at any process node in the entire production process. Meanwhile, based on production source data from the entire process, this module calculates multi-dimensional quality assessment indicators from six dimensions: raw material compliance, process stability, equipment reliability, environmental adaptability, quality compliance, and traceability integrity. Each dimension has a clear quantitative scoring standard and deduction rule, with a maximum score of 100 points for each indicator. The weight of each indicator is determined by the analytic hierarchy process, namely: raw material compliance 10%, process stability 35%, equipment reliability 15%, environmental adaptability 5%, quality compliance 30%, and traceability integrity 5%. The weighted fusion calculation yields the overall process quality assessment value, calculated as follows: Overall process quality assessment value = 0.6 × weighted score of six indicators + 0.4 × (1 - |freezing point prediction value - internal control standard value| / internal control standard value) × 100. This assessment value can comprehensively reflect the overall production quality status of the batch.

[0028] In addition to the above, it also includes setting up an intelligent identification unit for the root causes of quality anomalies, which is used to quickly generate solutions for identifying the root causes of quality anomalies; based on industry experience, process requirements and historical quality anomaly data in the production of low-freezing-point oleic acid, a comprehensive knowledge base for the root causes of quality anomalies in low-freezing-point oleic acid is built in advance. The knowledge base adopts a hierarchical classification architecture, which comprehensively covers various core quality anomalies such as high freezing point, excessive acid value, excessive iodine value and excessive color. For each type of anomaly, the corresponding possible abnormal process, key abnormal parameters, root cause analysis logic, standardized handling plan and long-term process optimization solution are clearly defined. The knowledge base has a strict knowledge entry and review mechanism. New anomaly cases and handling plans must be reviewed by the process manager before they can be entered. At the same time, it can be continuously supplemented and improved according to new problems and situations in production. Every time a new quality anomaly case occurs, it is entered into the knowledge base and the root cause analysis logic and handling plan are updated. This unit compares the multi-dimensional quality assessment indicators of the corresponding batch with the preset standard indicator range in real time, and compares the assessment indicators of each dimension with the standard range in real time. It identifies abnormal assessment indicators that exceed the standard range and locates the corresponding abnormal process nodes and specific production source data based on the abnormal indicators. The positioning accuracy can be down to a single set of parameters. Subsequently, this unit combines the quality risk transmission coefficient matrix to conduct an in-depth analysis of the transmission path of quality fluctuations at abnormal process nodes to subsequent processes, as well as the specific degree of impact on core quality indicators. It quantifies the transmission amplitude and scope of abnormal fluctuations, and clarifies the complete transmission link and core influencing factors of quality anomalies. Ultimately, this unit matches the quality anomaly root cause knowledge base, identifies the core root causes of quality anomalies through a similarity matching algorithm, and simultaneously generates targeted process optimization solutions and on-site emergency response measures. The solutions include specific parameter adjustment ranges, execution steps, execution entities, and verification standards, providing clear operational guidance for on-site quality anomaly handling. At the same time, a feedback loop for handling effect is set up. After the handling is completed, the handling effect is entered into the knowledge base to continuously optimize the accuracy of root cause identification and solution generation.

[0029] Fourth, the quality anomaly early warning module achieves early identification and early warning of finished product quality anomalies through multi-dimensional risk quantification. First, it pre-obtains the preset standard ranges of all process parameters at each node of the entire process of low-freezing-point oleic acid production. The standard ranges are comprehensively set based on national industry standards, enterprise internal control standards, and historical best data, covering key process parameters of all core processes such as freeze crystallization, hydrogenation reaction, and distillation separation. These include the cooling rate ±0.2℃ / h and constant temperature ±0.1℃ in the freeze crystallization process, the reaction temperature ±1℃ and hydrogen partial pressure ±0.02MPa in the hydrogenation reaction process, and the column top temperature ±0.5℃ and vacuum degree ±0.1kPa in the distillation separation process. All standard ranges have been verified by industrial mass production, providing unified standard support for subsequent risk calculation. Subsequently, this module collects production source data from each process node in real time, calculates the deviation of the real-time data from the standard range for each process, and uses the deviation formula: Deviation = |Actual Value - Standard Median| / Standard Range Width. The calculation process is synchronized with data collection; the corresponding deviation is calculated immediately after each data collection, quantifying the degree and risk of each process parameter deviating from the standard. Simultaneously, this module combines the predicted values ​​of core quality indicators with national and enterprise quality standards to calculate the deviation risk value of core quality indicators. The deviation risk value is calculated using the formula: Deviation Risk Value = |Predicted Freezing Point Value - Internal Control Standard Value| / Internal Control Standard Value, quantifying the excess of core quality indicators of the finished product. The standard risk level is established. Based on this, the module combines the overall quality assessment value, the deviation of quality fluctuations in each process, the risk value of deviations in core quality indicators, and the weight of each process's impact on the core quality indicators. It calculates the quality risk discrimination value through a multi-factor weighted comprehensive algorithm. The calculation formula is: Quality Risk Discrimination Value = 0.4 × Risk Value of Deviation in Core Quality Indicators + 0.3 × (1 - Overall Quality Assessment Value / 100) + 0.3 × max (Deviation of Quality Fluctuations in Each Process). The weights of each factor are scientifically set according to the actual impact on product quality. The calculation logic is fixed, the results are reproducible, and it can reflect the actual quality risk level of the batch. At any stage of the entire production process, based on the production source data of that stage and all preceding stages, the predicted values ​​of core quality indicators and the judgment values ​​of quality risks can be updated in real time. The update frequency is synchronized with the data collection frequency, and quality risks can be identified in advance without waiting for the final inspection of the finished product. The early warning information is pushed to the corresponding execution entity according to the preset path, with a push delay of ≤10s. At the same time, an early warning cycle confirmation mechanism is set up. After the early warning is triggered, the corresponding person in charge needs to confirm the handling result. After the handling is completed, the system verifies the risk elimination status, forming an early warning handling cycle.

[0030] The quality anomaly early warning module is also equipped with an adaptive iteration unit for the control system, which ensures the accuracy of early warnings and the adaptability of control during long-term system operation. This unit acquires various dynamic change data in the low-freezing-point oleic acid production process in real time, covering all dimensions of data, including production equipment status data, raw material composition change data, production environment change data, production process adjustment data, and product quality standard update data. Among them, equipment status data includes data on equipment operating accuracy decay, performance changes after equipment maintenance, and equipment service life data; raw material composition data includes data on differences in fatty acid composition between different batches of raw materials and composition changes after changes in raw material suppliers; environmental change data includes seasonal temperature and humidity changes, workshop environment adjustment data, and storage environment change data; process adjustment data includes parameter changes after production process optimization and production process adjustment data; and standard update data includes updates to national and industry standards and adjustments to enterprise internal control quality standards. The data acquisition method is real-time connection with relevant modules and equipment, and the acquisition frequency is synchronized with the data collection frequency, providing comprehensive and realistic evidence for the iteration of the control system. Based on the acquired real-time, multi-dimensional data, this unit uses a random forest machine learning algorithm to continuously and adaptively update the full-process quality risk transmission coefficient matrix, the influence weights of each process on core quality indicators, and the hierarchical control threshold ranges. It dynamically adjusts the matrix coefficients, weights, and threshold ranges according to actual changes in production conditions. Iteration triggers are divided into periodic iterations and event-triggered iterations. Periodic iterations are completed after every 100 batches of qualified data. Event-triggered iterations are initiated immediately when significant changes occur, such as major equipment maintenance, changes in raw material suppliers, or updates to quality standards. After each iteration, the data is processed. Only models with a prediction accuracy of ≥99% after independent validation set data verification can be deployed online, ensuring that quality risk assessment results, early warning responses, and actual production scenarios and quality control requirements are continuously adapted. At the same time, this unit updates the knowledge base of root causes of quality anomalies and process optimization solutions, and promptly adds new quality anomaly cases, root cause analysis logic, and handling plans to ensure that the knowledge base is synchronized with actual production. All iteration processes, parameter changes, and verification results are fully logged, making them traceable and auditable. A system rollback mechanism is also set up so that if the model becomes abnormal after iteration, it can be rolled back to the previous stable version with one click, ensuring the stability of system operation.

[0031] Fifth, the end-to-end cyclical control and iteration module enables the cyclical handling of quality anomalies through hierarchical control and the continuous optimization of production processes through data accumulation. First, a hierarchical control threshold range with multiple levels is pre-set. The range is based on a comprehensive set of national product quality standards, enterprise internal control standards, and historical production big data. The thresholds are clearly defined as follows: Level 1 warning threshold 0.6, Level 2 control threshold 0.8, and Level 3 shutdown threshold 0.95. Each threshold range corresponds to a different quality risk level. The Level 1 threshold corresponds to general quality fluctuations, the Level 2 threshold corresponds to relatively serious quality anomalies, and the Level 3 threshold corresponds to major quality anomalies, providing a clear standard basis for triggering control actions. The module then compares the quality risk assessment value with the graded control threshold range in real time. Based on the range in which the assessment value falls, it determines the level of quality anomaly and triggers corresponding control actions. These actions cover early warning prompts, cyclical adjustments to process parameters, suspension of material feeding for high-risk processes, and full-process emergency shutdown at all levels. All control actions comply with chemical safety production interlocking standards. Level 1 early warning corresponds to general quality fluctuations, pushing early warning prompts and parameter optimization suggestions to on-site operators within 5 minutes, continuously tracking parameter changes, and updating the risk assessment value every 30 seconds. Level 2 control corresponds to more serious quality anomalies. Within one minute, a parameter adjustment command is issued to the PLC controller of the abnormal process, and the process parameters are cyclically corrected. Simultaneously, the feeding of subsequent high-risk processes is suspended, and medium-to-high level warnings, abnormal data details, and adjustment plans are pushed to workshop management personnel. Level 3 shutdown corresponds to major quality abnormalities. An emergency shutdown command for the entire process is immediately issued to cut off the raw material feeding, lock the batch's full-process traceability data, and push the highest level warning, abnormal root cause analysis results, and emergency response plan to the production control person in charge via SMS, platform messages, and mobile app. All control actions are fully logged and traceable. For production batches deemed qualified, this module stores comprehensive information, including full-process production source data, traceability information, quality analysis results, and finished product full-item inspection data, into the core production database. The database is deployed on a private cloud, configured with dedicated storage servers and off-site backup mechanisms, with a data backup frequency of at least once per day to ensure data security and accessibility, providing a high-quality big data foundation for process optimization. Finally, the module uses machine learning algorithms to deeply mine and analyze the data in the core production database, iteratively optimizing the quality risk transmission coefficient matrix, the standard ranges of process parameters for each process step, and the threshold ranges for hierarchical control. Iterative optimization is achieved using a random forest machine learning algorithm, with a full iteration completed every 100 batches of qualified production data. After the entry of abnormal data for a single batch, local optimization is performed. After iteration, the model is validated using 20% ​​independent validation set data, and can only be deployed online if the model prediction accuracy is at least 99%. The optimized process parameters must undergo three batches of pilot testing, and only after the finished product qualification rate reaches 100% can they be updated to the production system's process standard library, achieving continuous self-optimization of the production process.

[0032] Finally, it should be noted that the above embodiments are merely examples for clearly illustrating the present invention and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A quality traceability and early warning system for low-freezing-point oleic acid production under an industrial internet architecture, characterized in that: include: The dynamic acquisition module is used to deploy edge acquisition terminals at each process node of the entire low-freezing-point oleic acid production process through an industrial internet architecture. It pre-constructs the influence weight matrix of each process on the core quality indicators of the product, and sets dynamic differentiated acquisition rules for the production source data of each process node according to the influence weight matrix, and synchronously collects multi-source heterogeneous production source data of the entire low-freezing-point oleic acid production process. The traceability link construction module is used to assign a globally unique batch identification code to each independent production batch of low freezing point oleic acid. The batch identification code is irreversibly and strongly bound to the production source data of each process node in the entire production process of that batch and the influence weight of each process on the core quality indicators of the product. A full-process traceability index is established with the batch identification code as the retrieval subject and the influence factor of the core quality indicators of the product as the correlation dimension, and traceability link information is generated. The coupling analysis module is used to perform coupling analysis on the quality fluctuations of each process node based on the production source data corresponding to the full-process traceability information and the pre-constructed quality risk transmission coefficient matrix of the entire process of low-freezing-point oleic acid production, and to calculate the predicted value of the core quality indicators of the product and the quality assessment value of the entire process. The early warning module is used to calculate the quality risk judgment value based on the predicted value of the product's core quality indicators, the quality assessment value of the whole process, and the deviation of the quality fluctuation at each process node. The control iteration module is used to determine whether the quality risk judgment value exceeds the preset hierarchical control threshold range; If the quality risk assessment value exceeds the preset graded control threshold range, it is determined to be a quality abnormality. Based on the risk transmission level of the abnormal process node, the corresponding level of early warning response and production process control action will be triggered. Conversely, if the quality is not satisfactory, the entire production process data is stored in the core production database for iterative optimization of the low-freezing-point oleic acid production process.

2. The low-freezing-point oleic acid production quality traceability and early warning system under the industrial internet architecture according to claim 1, characterized in that, The dynamic acquisition module includes: The raw material data acquisition unit is used to collect information on the source, batch, core component testing data, basic quality index data, and warehousing and storage environment data of the raw materials used in the production of low-freezing-point oleic acid, and to generate raw material attribute and component data. The key process data acquisition unit is used to collect real-time operating parameters of each key process in the entire production process of low-freezing-point oleic acid that has a direct impact on the core quality indicators of the product, and generate full-process production process parameter data. The equipment status acquisition unit is used to collect the operating status parameters, fault alarm information, maintenance records and operating accuracy data of the production equipment corresponding to the above key processes, and generate key equipment operating status data. The environmental data acquisition unit is used to collect real-time environmental parameters in the core process workshops of low-freezing-point oleic acid production and the finished product storage area, and generate full-scenario environmental monitoring data. The quality inspection and data acquisition unit is used to collect the intermediate control semi-finished product inspection data and the finished product full quality index inspection data during the production process of low freezing point oleic acid, and to generate process and finished product quality inspection data. The logistics data acquisition unit is used to collect outbound information, warehousing and transportation environment data, delivery information and terminal feedback data of low-freezing-point oleic acid products, and generate warehousing and logistics data. The dynamic data acquisition and control unit is used to construct an impact weight matrix of each process on the core quality indicators of the product based on historical production big data. The initial acquisition cycle and acquisition accuracy are set for each process node according to the impact weight matrix. At the same time, the fluctuation range of process parameters of each process node is monitored in real time. When the fluctuation of process parameters of a certain process node exceeds the preset stability threshold, the acquisition frequency and acquisition accuracy of the production source data corresponding to that process node are increased.

3. The low-freezing-point oleic acid production quality traceability and early warning system under the industrial internet architecture according to claim 2, characterized in that, The dynamic acquisition module also includes a full-link data standardization processing unit; The end-to-end data standardization processing unit is used to perform format standardization, outlier removal, and missing value completion processing on all types of production source data collected. According to the process nodes and material flow sequence of the entire production process of low freezing point oleic acid, all production source data are aligned with the process timestamp and material batch dual benchmarks to ensure that all production source data within the same process node are at the same time benchmark and material batch benchmark. The various production source data that have undergone standardization are classified and encrypted for storage according to batch identification codes and process nodes, providing a standardized data source for the construction of a full-process traceability chain and coupled analysis of quality risks.

4. The low-freezing-point oleic acid production quality traceability and early warning system under the industrial internet architecture according to claim 1, characterized in that, The traceability link construction module specifically includes: This is used to assign a globally unique batch identification code that conforms to the Industrial Internet Identifier Resolution System to each independent production batch of low freezing point oleic acid. The batch identification code is then irreversibly encrypted and bound to the corresponding batch's raw material batch information, production source data of each node in the entire process, the influence weight of each node on the core quality indicators of the product, and the measured data of the core quality indicators of the finished product through an encryption algorithm. Using batch identification codes as the core, and following the sequence of production processes, all production source data for the corresponding batch are sequentially associated according to process nodes, establishing a full-process traceability index with batch identification codes as the retrieval subject and product core quality indicator influencing factors as the association dimension. Based on the full-process traceability index, traceability link information covering the entire product lifecycle is generated. The traceability link information supports forward full-process traceability and reverse root cause localization. Forward traceability can fully display the full lifecycle information of the corresponding batch of products, while reverse root cause localization can directly locate the specific process node and single set of production source data that caused the quality abnormality based on the abnormal results of the core quality indicators of the finished product.

5. The low-freezing-point oleic acid production quality traceability and early warning system under the industrial internet architecture according to claim 4, characterized in that, The traceability link construction module also includes a multi-scenario batch traceability query unit; The multi-scenario batch traceability query unit supports hierarchical query services for entities with different permissions. It opens up traceability data query permissions according to the query needs of different entities. At the same time, it can generate standardized traceability reports and product quality compliance certificates that meet the corresponding requirements according to the needs of the querying entity.

6. The low-freezing-point oleic acid production quality traceability and early warning system under the industrial internet architecture according to claim 1, characterized in that, The coupling analysis module specifically includes: Used to extract standardized production source data for each node of the entire process of a corresponding production batch based on the traceability information of the entire process chain; Based on the quality risk transmission coefficient matrix of the entire process of low-freezing-point oleic acid production, which is pre-constructed using historical big data of production, the quality risk transmission coefficient matrix includes the quality fluctuation transmission coefficient between each process node and the direct impact coefficient of each process node on the core quality indicators of the product. By using a coupled analysis algorithm, the transmission effect of process parameter fluctuations at each process node on subsequent processes and the comprehensive impact on the core quality indicators of the final product are calculated, and the predicted values ​​of the core quality indicators of the corresponding production batch are generated. At the same time, based on the production source data of each node in the entire process, multi-dimensional quality assessment indicators for the corresponding production batch are calculated. By combining the predicted values ​​of the product's core quality indicators with multi-dimensional quality assessment indicators, the full-process quality assessment value of the corresponding production batch is obtained through weighted fusion calculation.

7. The low-freezing-point oleic acid production quality traceability and early warning system under the industrial internet architecture according to claim 6, characterized in that, The coupling analysis module also includes an intelligent identification unit for the root causes of quality anomalies; The intelligent identification unit for the root causes of quality abnormalities is used to pre-build a knowledge base for the root causes of quality abnormalities in low-freezing-point oleic acid. The knowledge base includes abnormal processes, abnormal parameters, root cause analysis logic and optimization solutions corresponding to different quality abnormalities. The multi-dimensional quality assessment indicators of the corresponding production batch are compared with the preset standard indicator range to identify abnormal assessment indicators that exceed the standard range and locate the corresponding abnormal process nodes and production source data. By combining the quality risk transmission coefficient matrix, we analyze the impact path and degree of quality fluctuations at abnormal process nodes on the core quality indicators of the product. We then match the knowledge base of root causes of quality anomalies to identify the core root causes of quality anomalies and generate process optimization solutions simultaneously.

8. The low-freezing-point oleic acid production quality traceability and early warning system under the industrial internet architecture according to claim 1, characterized in that, The early warning module specifically includes: Used to obtain the preset standard range of process parameters for each node in the entire process of low freezing point oleic acid production, and to calculate the deviation of the quality fluctuation between the real-time production source data of each process node and the standard range. Based on the predicted values ​​of the product's core quality indicators and the national quality standards and internal control quality standards of low-freezing-point oleic acid products, the deviation risk values ​​of the core quality indicators are calculated. By combining the overall quality assessment value, the quality fluctuation deviation of each process node, the deviation risk value of the core quality indicators, and the influence weight of each process node on the core quality indicators of the product, the quality risk discrimination value of the corresponding production batch is calculated. Simultaneously, at any process node in the production process, based on the production source data of that node and the preceding process, the predicted values ​​of the product's core quality indicators and the judgment values ​​of quality risks can be updated in real time.

9. The low-freezing-point oleic acid production quality traceability and early warning system under the industrial internet architecture according to claim 8, characterized in that, The early warning module also includes an adaptive iteration unit for the control system; The adaptive iteration unit of the control system is used to acquire real-time data on the status of production equipment for low-freezing-point oleic acid, data on changes in raw material composition, data on changes in the production environment, data on adjustments to the production process, and data on updates to product quality standards. Based on the real-time data obtained above, machine learning algorithms are used to adaptively and iteratively update the whole process quality risk transmission coefficient matrix, the influence weight of each process node on the core quality indicators of the product, and the hierarchical control threshold range.

10. The low-freezing-point oleic acid production quality traceability and early warning system under the industrial internet architecture according to claim 1, characterized in that, The control and iteration module specifically includes: This is used to pre-set a hierarchical control threshold range containing multiple levels of thresholds. The hierarchical control threshold range is set based on the national product quality standard for low freezing point oleic acid, the enterprise's internal control quality standard, and historical production big data. The real-time calculated quality risk discrimination value is compared with the graded control threshold range. Based on the threshold range where the quality risk discrimination value is located, the quality abnormality level is determined and the corresponding control action is triggered. The control actions include early warning prompts, process parameter adjustments, suspension of high-risk processes, and emergency shutdown of the entire process. For production batches that meet quality standards, the full-process production source data, traceability information, and quality analysis results of the corresponding batch are stored in the core production database. The quality risk transmission coefficient matrix, process parameter standard range, and hierarchical control threshold range are iteratively optimized through machine learning algorithms.

Images