Aquatic product processing production link temperature control method and system

Abstract

The invention discloses an aquatic product processing production link temperature control method, which comprises the following steps: dynamically adjusting temperature control parameters in a processing link according to temperature quality related parameters, and if the temperature quality related parameter of a certain link displays that the influence weight of temperature fluctuation on a quality index is higher than a preset threshold value, judging that the temperature control parameters are not the temperature control parameters; if yes, the temperature tolerance range of the link is reduced, and an optimized temperature control parameter set is obtained; and constructing a problem tracing path according to the abnormal feedback data monitored in real time, automatically tracing back to the corresponding data segment in the integrated temperature information base aiming at the time point and the link label of the alarm record, and extracting the temperature change sequence before and after the occurrence of the abnormality to obtain a detailed abnormality tracing report.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature control of production links, and in particular to a method and system for temperature control of aquatic product processing production links. Background Art

[0002] Aquatic product processing is a crucial area for food safety and economic profitability. Its core focus lies in ensuring product quality and extending shelf life. Temperature control, a key factor affecting product freshness and safety, directly determines the success or failure of the entire processing chain. Proper temperature management not only inhibits microbial growth but also maintains the product's taste and nutritional value, making it an irreplaceable factor in the entire production chain.

[0003] However, many aquatic product processing companies currently face significant deficiencies in temperature control, generally relying on manual record-keeping or simple equipment monitoring, and lacking efficient data integration and analysis methods. This approach often results in fragmented and incomplete temperature data, failing to fully reflect the true state of the production process and making it even more difficult to quickly identify the cause when problems occur. Against this backdrop, the industry faces several interrelated core challenges. The primary challenge is the real-time collection and complete recording of temperature data. Due to the lack of a unified data management platform, temperature information from various links in the processing chain is often stored in a decentralized manner, making it difficult to establish a continuous monitoring chain. This further complicates traceability. When product quality issues arise, it is impossible to quickly determine whether they are related to temperature anomalies, let alone accurately pinpoint the specific production link and time point.

[0004] The deeper problem is that simple temperature records cannot reveal the deep connection between temperature and product quality. The lack of regularity mining based on data analysis makes the optimization of temperature control parameters lack scientific basis, affecting the stability of product quality. Summary of the Invention

[0005] In order to solve the above-mentioned technical problems, the present invention provides a temperature control method for aquatic product processing production link.

[0006] The technical solution of the present invention is achieved as follows:

[0007] A temperature control method for an aquatic product processing production link, comprising:

[0008] Real-time acquisition of temperature data at each stage of aquatic product processing, high-frequency sampling of temperature changes at each node, and a preliminary temperature monitoring data set are obtained;

[0009] Preprocess the collected preliminary temperature monitoring data set to obtain a cleaned temperature data set;

[0010] Through the cleaned temperature data set, a unified data management platform is built to integrate the temperature information of all processing links into a single database to obtain an integrated temperature information database;

[0011] Based on the integrated temperature information database, feature extraction is performed on the temperature data stream, and the temperature change trend of each processing link is quantitatively analyzed to obtain a temperature feature data set containing abnormal markers;

[0012] By correlating the temperature feature dataset containing abnormal marks with product quality inspection records, we can obtain temperature quality correlation parameters.

[0013] Dynamically adjust the temperature control parameters in the processing process according to the temperature-quality correlation parameters;

[0014] By optimizing the temperature control parameter set and updating the control logic of IoT sensor devices, we can compare the real-time temperature monitoring data of each processing link and obtain real-time monitoring abnormal feedback data.

[0015] Based on the abnormal feedback data from real-time monitoring, a problem tracing path is constructed, which automatically traces back to the corresponding data segment in the integrated temperature information library, extracts the temperature change sequence before and after the abnormality occurs, and obtains a detailed abnormality tracing report.

[0016] Furthermore, obtaining a preliminary temperature monitoring data set includes:

[0017] Collect temperature data from all nodes in the entire link through IoT devices to form an initial temperature monitoring data set;

[0018] Based on the initial temperature monitoring data set, sort the data according to the timestamp identifier of each node to obtain an ordered temperature data sequence;

[0019] Among them, if there are missing values in the ordered temperature data sequence, the missing data are filled by linear interpolation to determine the complete temperature data stream;

[0020] Based on the complete temperature data stream, the support vector machine model is used to analyze the temperature change trend of each node to determine whether there is abnormal fluctuation;

[0021] If the analysis results show that the temperature change trend of a node exceeds the preset threshold, the temperature data stream of the node is segmented to obtain the data subset of the abnormal time period;

[0022] By using the data subset of the abnormal time period and combining the correspondence between the timestamp identifier and the full-link node, the specific raw material storage or finished product packaging link can be located to determine the specific location where the abnormality occurred.

[0023] Based on the positioning results, high-frequency sampling data within the relevant time period is extracted for the specific location where the anomaly occurs, and detailed temperature change records are obtained for subsequent processing.

[0024] Furthermore, obtaining the cleaned temperature data set includes:

[0025] Based on the preliminary temperature monitoring data set, a data cleaning algorithm is used to preprocess the collected temperature data stream to identify and repair missing values and outliers in the data;

[0026] If it is detected that the temperature value corresponding to a certain timestamp exceeds the preset reasonable range threshold, the temperature value of the previous and next time points is supplemented by linear interpolation to obtain a cleaned temperature dataset;

[0027] Specifically, by further processing the cleaned data set, the temperature data distribution characteristics under each timestamp are obtained and its fluctuation range is determined;

[0028] According to the distribution characteristics, statistical analysis tools are used to evaluate the stability of temperature data and obtain the stability index;

[0029] If the stability index exceeds the preset threshold, the temperature data segment corresponding to the timestamp is extracted based on the result of the outlier identification to determine whether there is a persistent deviation.

[0030] By segmenting the temperature data fragments with persistent deviations, the time interval of abnormal fluctuations is obtained and the critical time range is determined;

[0031] Based on the key time range, combined with the association between timestamp identifiers and temperature monitoring, the historical records of relevant temperature data are extracted to obtain detailed fluctuation information;

[0032] The support vector machine model is used to classify the detailed fluctuation information and determine whether the fluctuation is within the controllable range;

[0033] Among them, if the fluctuation exceeds the controllable range, the change pattern of the deviation trend is obtained through comparative analysis of historical records and current temperature data to determine the basis for adjustment.

[0034] Furthermore, the obtaining of the integrated temperature information database includes:

[0035] By using the temperature data after cleaning, a unified management platform is built to integrate data streams from raw material storage, processing, and finished product packaging into a single database. Classification labels are added for different links to obtain an integrated information database.

[0036] Obtaining a temperature feature dataset containing abnormal markers includes:

[0037] Through the integrated temperature information database, the temperature information of each processing link is preliminarily sorted out, and the time series analysis method is used to structure the data to obtain the sorted temperature data stream;

[0038] Based on the sorted temperature data stream, the temperature information of each processing link is tested for change trends, and the fluctuation amplitude in the trend is compared using a preset threshold range. If the fluctuation amplitude of a certain link exceeds the threshold range, it is marked as a potential anomaly point, and a data stream with potential anomaly marks is obtained.

[0039] Furthermore, the obtaining of temperature quality related parameters includes:

[0040] By associating the temperature feature dataset containing abnormality marks with quality inspection records, a preliminary match is performed between the temperature anomalies and the product batch data. The data alignment method is used to organize the correspondence between the temperature features and the quality data to obtain a preliminary matching dataset.

[0041] Based on the initially matched data set, feature extraction is performed on the correlation between temperature anomalies and quality data. Pre-set rules are used to compare temperature fluctuations with quality indicators. If the temperature fluctuation exceeds the preset threshold and the quality indicator deviates, it is marked as a key correlation point, and the key correlation point set is determined;

[0042] Through the set of key correlation points, the distribution characteristics of temperature fluctuations and quality indicators are modeled and processed. The regression analysis method is used to quantify the correlation between temperature characteristics and quality data, and the correlation parameters between temperature and quality are obtained.

[0043] Furthermore, the optimized temperature control parameter set includes:

[0044] In view of the relationship between temperature fluctuation and quality indicators, the temperature control records in the processing links were extracted from the historical data. The data cleaning method was used to remove outliers to obtain the sorted temperature fluctuation data set.

[0045] Based on the organized temperature fluctuation data set, the correlation between temperature fluctuation and quality indicators is analyzed, and the linear regression analysis method is used to quantify the influence weight and determine the temperature correlation parameters of each processing link;

[0046] If the influence weight of a certain processing link exceeds the preset threshold, the temperature control strategy of this link will be dynamically adjusted to tighten the tolerance range and obtain the optimized parameter set.

[0047] Furthermore, obtaining abnormal feedback data from real-time monitoring includes:

[0048] By adjusting the sensor logic of IoT devices through an optimized temperature control parameter set, temperature data during processing is continuously collected to obtain real-time monitoring of temperature fluctuations.

[0049] The collected temperature fluctuation information is compared with the preset tolerance range. If the temperature data is detected to be outside the tolerance range, the abnormal alarm mechanism is triggered and the triggering record of the abnormal alarm is obtained;

[0050] For the trigger records of abnormal alarms, the corresponding time records and processing link information are automatically extracted, the abnormal feedback data is saved through data storage, and the detailed archive of abnormal events is determined.

[0051] Furthermore, the detailed abnormality tracing report includes:

[0052] Through the abnormal feedback data obtained by real-time monitoring, the relevant data fragments in the temperature information library are automatically located according to the time point and processing link identification in the alarm record, and a preliminary data set before and after the abnormality occurs is obtained;

[0053] Based on the preliminary data set, the temperature change sequence before and after the anomaly occurs is extracted and compared using a preset threshold range. If the temperature change exceeds the threshold range, it is marked as a key fluctuation segment, and the distribution characteristics of the key fluctuation segment are determined;

[0054] Based on the distribution characteristics of key fluctuation segments, a time series analysis model of temperature changes is constructed, and the support vector machine model is used to classify the fluctuation segments to obtain the classified fluctuation type information;

[0055] Through the classified fluctuation type information, the corresponding processing link identification is automatically matched, the correlation between temperature changes and processing links is analyzed, and the fluctuation influencing factors related to the links are obtained;

[0056] Based on the fluctuation influencing factors related to each link, the core time points where the anomaly occurred are screened out. Combined with the historical data fragments in the temperature information database, a detailed path for anomaly tracing is constructed to obtain the associated data groups in the tracing path.

[0057] For the associated data groups in the traceability path, the correspondence between the abnormal feedback data and the processing links is automatically archived, and the archived information is saved through data storage to determine the complete information of abnormal traceability.

[0058] Furthermore, it also includes:

[0059] Through detailed anomaly tracing reports, the temperature-quality correlation model is updated, and the model parameters are iteratively optimized based on the newly collected anomaly data and quality data. The incremental learning method is used to adjust the weight of the impact of temperature fluctuations on quality, and the updated correlation model parameters are obtained for temperature control optimization in subsequent processing links.

[0060] A temperature control system for aquatic product processing production link, comprising:

[0061] The data acquisition and preprocessing module is used to collect and store temperature data at high frequency in real time at all nodes in the entire aquatic product processing chain, and then clean and repair the data to obtain a cleaned temperature data set;

[0062] The data integration and feature extraction module is used to integrate the cleaned temperature data and classify and label them, build a temperature information database, extract temperature change trend features, mark abnormal events, and obtain a temperature feature dataset;

[0063] The correlation analysis and parameter optimization module is used to correlate temperature feature data sets with product quality data, build correlation models, calculate influence weights, and dynamically adjust temperature control parameters based on these data to obtain an optimized parameter set.

[0064] Real-time monitoring and abnormality feedback module, used to update the control logic based on the optimized parameter set, monitor temperature data in real time, and alarm and record abnormal data if it exceeds the tolerance range;

[0065] The anomaly tracing and model updating module is used to trace the problem based on the anomaly feedback data, extract the temperature sequence before and after the anomaly, update the associated model, and optimize the model parameters for subsequent temperature control optimization.

[0066] Compared with the prior art, the present invention has the following beneficial effects:

[0067] The present invention deploys Internet of Things sensors to collect temperature data from all processing links in real time, builds a unified data management platform to integrate temperature information, applies time series analysis to extract temperature characteristics and mark anomalies, establishes a correlation model between temperature and product quality, dynamically optimizes temperature control parameters, realizes real-time monitoring and abnormal alarms, builds problem tracing paths and continuously updates the correlation model, realizes precise temperature control of the entire process of aquatic product processing, effectively reduces the adverse effects of temperature fluctuations on product quality, improves the automation and intelligence level of the processing process, and provides technical support for ensuring the quality and safety of aquatic products. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 This is a flow chart of a temperature control method for an aquatic product processing production link in Example 1;

[0069] Figure 2 This is a flow chart of a temperature control method for an aquatic product processing production link according to Example 2;

[0070] Figure 3 This is a framework diagram of a temperature control system for an aquatic product processing production link in Example 3. DETAILED DESCRIPTION

[0071] In order to make the purposes, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0072] Example 1

[0073] like Figure 1 As shown, this embodiment provides a method for controlling temperature in an aquatic product processing production link, including:

[0074] Real-time acquisition of temperature data at each stage of aquatic product processing, high-frequency sampling of temperature changes at each node, and a preliminary temperature monitoring data set are obtained;

[0075] Preprocess the collected preliminary temperature monitoring data set to obtain a cleaned temperature data set;

[0076] Through the cleaned temperature data set, a unified data management platform is built to integrate the temperature information of all processing links into a single database to obtain an integrated temperature information database;

[0077] Based on the integrated temperature information database, feature extraction is performed on the temperature data stream, and the temperature change trend of each processing link is quantitatively analyzed to obtain a temperature feature data set containing abnormal markers;

[0078] By correlating the temperature feature dataset containing abnormal marks with product quality inspection records, we can obtain temperature quality correlation parameters.

[0079] Dynamically adjust the temperature control parameters in the processing process according to the temperature-quality correlation parameters;

[0080] By optimizing the temperature control parameter set and updating the control logic of IoT sensor devices, we can compare the real-time temperature monitoring data of each processing link and obtain real-time monitoring abnormal feedback data.

[0081] Based on the abnormal feedback data from real-time monitoring, a problem tracing path is constructed, which automatically traces back to the corresponding data segment in the integrated temperature information library, extracts the temperature change sequence before and after the abnormality occurs, and obtains a detailed abnormality tracing report.

[0082] Furthermore, obtaining a preliminary temperature monitoring data set includes:

[0083] Collect temperature data from all nodes in the entire link through IoT devices to form an initial temperature monitoring data set;

[0084] Based on the initial temperature monitoring data set, sort the data according to the timestamp identifier of each node to obtain an ordered temperature data sequence;

[0085] Among them, if there are missing values in the ordered temperature data sequence, the missing data are filled by linear interpolation to determine the complete temperature data stream;

[0086] Based on the complete temperature data stream, the support vector machine model is used to analyze the temperature change trend of each node to determine whether there is abnormal fluctuation;

[0087] If the analysis results show that the temperature change trend of a node exceeds the preset threshold, the temperature data stream of the node is segmented to obtain the data subset of the abnormal time period;

[0088] By using the data subset of the abnormal time period and combining the correspondence between the timestamp identifier and the full-link node, the specific raw material storage or finished product packaging link can be located to determine the specific location where the abnormality occurred.

[0089] Based on the positioning results, high-frequency sampling data within the relevant time period is extracted for the specific location where the anomaly occurs, and detailed temperature change records are obtained for subsequent processing.

[0090] In some embodiments, the process may be as follows:

[0091] Furthermore, obtaining the cleaned temperature data set includes:

[0092] Based on the preliminary temperature monitoring data set, a data cleaning algorithm is used to preprocess the collected temperature data stream to identify and repair missing values and outliers in the data;

[0093] If it is detected that the temperature value corresponding to a certain timestamp exceeds the preset reasonable range threshold, the temperature value of the previous and next time points is supplemented by linear interpolation to obtain a cleaned temperature dataset;

[0094] Specifically, by further processing the cleaned data set, the temperature data distribution characteristics under each timestamp are obtained and its fluctuation range is determined;

[0095] According to the distribution characteristics, statistical analysis tools are used to evaluate the stability of temperature data and obtain the stability index;

[0096] If the stability index exceeds the preset threshold, the temperature data segment corresponding to the timestamp is extracted based on the result of the outlier identification to determine whether there is a persistent deviation.

[0097] By segmenting the temperature data fragments with persistent deviations, the time interval of abnormal fluctuations is obtained and the critical time range is determined;

[0098] Based on the key time range, combined with the association between timestamp identifiers and temperature monitoring, the historical records of relevant temperature data are extracted to obtain detailed fluctuation information;

[0099] The support vector machine model is used to classify the detailed fluctuation information and determine whether the fluctuation is within the controllable range;

[0100] Among them, if the fluctuation exceeds the controllable range, the change pattern of the deviation trend is obtained through comparative analysis of historical records and current temperature data to determine the basis for adjustment.

[0101] In some embodiments, the process may be as follows:

[0102] For full-chain temperature monitoring in aquatic product processing, data collection is first achieved by deploying IoT sensor devices. Specifically, TS-300 temperature and humidity sensors are installed at key locations, such as cold storage for raw materials, processing workshops, and finished product packaging areas. Each sensor covers an area of 10 square meters, ensuring comprehensive data. The sensors have built-in high-precision temperature measurement chips with a measurement range of -20°C to 50°C and an accuracy of ±0.5°C. Temperature data is automatically collected every minute, forming a high-frequency sampling stream.

[0103] Secondly, the data transmission link uses the 4G network module to upload the collected temperature data to the cloud in real time. The AES-128 encryption algorithm is used during the transmission process to ensure data security. Each data is accompanied by a timestamp (such as 2023-10-01 08:00:00) and device ID to ensure data traceability.

[0104] Next, store the data in a cloud database, such as Alibaba Cloud RDS MySQL. Design a table structure containing the following fields: timestamp, device ID, and temperature value (unit: °C). Set up an automatic partitioning strategy, partitioning the data by day to improve query efficiency. The daily data volume is expected to be 1,440 records per device (60 minutes x 24 hours).

[0105] Subsequently, a Python-based monitoring algorithm was developed to analyze temperature changes. Temperature thresholds were set (for example, raw material storage must be maintained at -18°C ± 2°C). The hourly temperature mean and standard deviation were calculated using a sliding window algorithm. If the mean deviated from the threshold or the standard deviation exceeded 0.8, an abnormal alarm was triggered and the alarm information was pushed to the enterprise management platform.

[0106] At the same time, combined with business associations, abnormal data can be linked to the cold chain equipment API to automatically adjust the temperature. For example, when the cold storage temperature rises to -16°C, refrigeration is automatically started to the target value.

[0107] Finally, data visualization processing uses ECharts to generate temperature trend charts, aggregate data by hour, and display the temperature fluctuation curve of each node within 24 hours, which is convenient for tracing problem links. For example, if the temperature in the packaging area is found to be continuously higher than 5°C for more than 30 minutes, an analysis report is automatically generated. Combined with historical data (the average value for the past 7 days is 4.2°C), possible causes can be inferred, such as equipment failure or doors not closed tightly, to assist in decision optimization.

[0108] Furthermore, the obtaining of the integrated temperature information database includes:

[0109] By using the cleaned temperature data, a unified management platform is built to integrate data streams from raw material storage, processing, and finished product packaging into a single database. Classification labels are added for different links to obtain an integrated information database.

[0110] Obtaining a temperature feature dataset containing abnormal markers includes:

[0111] Through the integrated temperature information database, the temperature information of each processing link is preliminarily sorted out, and the time series analysis method is used to structure the data to obtain the sorted temperature data stream;

[0112] Based on the sorted temperature data stream, the temperature information of each processing link is tested for change trends, and the fluctuation amplitude in the trend is compared using a preset threshold range. If the fluctuation amplitude of a certain link exceeds the threshold range, it is marked as a potential anomaly point, and a data stream with potential anomaly marks is obtained.

[0113] In some embodiments, the process may be as follows:

[0114] To extract features and flag anomalies in the temperature data stream, we first obtain the temperature data stream of the processing link from the integrated temperature information database. Assume that the temperature data of a processing link is collected once a minute, and 1440 data points are obtained for 24 consecutive hours. The data range is between 20.5 and 35.8 degrees Celsius.

[0115] Using a time series analysis method and a sliding window algorithm to extract temperature trend characteristics, the window size is set to 30 minutes, that is, 30 data points. The mean and standard deviation of the temperature within each window are calculated. For example, the mean in a certain window is 28.3 degrees Celsius and the standard deviation is 1.2, reflecting the temperature fluctuation.

[0116] Then, the temperature change trend of each processing link is quantitatively analyzed, and the difference between the mean values of adjacent windows is calculated as the trend indicator. If the mean value rises from 28.3 to 30.1 within a certain period of time, the trend indicator is 1.8, indicating a clear temperature increase trend.

[0117] The temperature fluctuation threshold is then set to twice the standard deviation. That is, if the standard deviation of a window exceeds 2.4, it is considered an abnormal fluctuation. Furthermore, based on business rules, if the mean temperature exceeds the process requirement range of 25 to 30 degrees Celsius, it is also marked as an abnormality. For example, if the mean temperature in a window reaches 31.5 degrees Celsius and the standard deviation is 2.6, exceeding the threshold of 2.4 and the mean value is out of limit, it is automatically marked as an abnormal event.

[0118] Finally, a temperature feature dataset containing anomaly markers is generated. Each record in the dataset includes a timestamp, temperature mean, standard deviation, trend indicator, and anomaly marker fields. For example, the record is "2023-10-01 08:30:00, 31.5, 2.6, 1.8, anomaly". The dataset is stored in the database for subsequent process optimization analysis.

[0119] Through the above method, the entire process from data extraction to abnormality marking can be automatically completed to ensure the accuracy of temperature monitoring in the processing link. At the same time, it is closely integrated with the process requirements to form a complete logical chain from data to decision-making.

[0120] Furthermore, the obtaining of temperature quality related parameters includes:

[0121] By associating the temperature feature dataset containing abnormality marks with quality inspection records, a preliminary match is performed between the temperature anomalies and the product batch data. The data alignment method is used to organize the correspondence between the temperature features and the quality data to obtain a preliminary matching dataset.

[0122] Based on the initially matched data set, feature extraction is performed on the correlation between temperature anomalies and quality data. Pre-set rules are used to compare temperature fluctuations with quality indicators. If the temperature fluctuation exceeds the preset threshold and the quality indicator deviates, it is marked as a key correlation point, and the key correlation point set is determined;

[0123] Through the set of key correlation points, the distribution characteristics of temperature fluctuations and quality indicators are modeled and processed. The regression analysis method is used to quantify the correlation between temperature characteristics and quality data, and the correlation parameters between temperature and quality are obtained.

[0124] In some embodiments, the process may be as follows:

[0125] Using a temperature feature dataset containing anomaly markers, we first extract temperature data with anomaly markers from the database. For example, suppose the data from a batch of processing on October 2, 2023, shows that the average temperature for a certain period of time is 32.8 degrees Celsius with a standard deviation of 2.9, which is marked as an anomaly.

[0126] Next, the product quality inspection records are automatically associated, and the temperature data of the time period is matched with the product quality data of the corresponding batch. For example, the hardness index of this batch of products is 75.5, which is lower than the standard value of 78. The matching result is recorded as "Batch A, temperature abnormality 32.8, hardness 75.5".

[0127] Subsequently, a temperature-quality correlation model was constructed based on the matching results. A linear regression analysis algorithm was used, and multiple sets of matching data were input. For example, the mean temperature of another batch was 29.2 degrees Celsius and the hardness was 79.3. The weight of the impact of temperature fluctuations on the hardness index was calculated, and the regression coefficient was -0.85, indicating that for every 1 degree Celsius increase in temperature, the hardness index decreased by 0.85 units. In further analysis, the data of temperature anomaly events were weighted. Combined with the severity of the anomaly mark, it was assumed that the weight of the temperature exceeding the standard range by more than 3 degrees Celsius was 1.5. The temperature-quality correlation parameters were calculated. For example, the comprehensive impact factor was -1.28, reflecting the significant negative impact of temperature anomalies on quality. To form a complete logical chain, the correlation parameters were also combined with the periodic fluctuation data of the production batches. Assuming that the frequency of temperature anomalies in a certain period was 5 times, the automatic adjustment parameter was -1.35 to reflect the periodic impact. Finally, all calculation results were stored in the analysis library for reference in subsequent production adjustments.

[0128] Furthermore, the optimized temperature control parameter set includes: extracting temperature control records in the processing link from historical data based on the relationship between temperature fluctuations and quality indicators, using a data cleaning method to remove outliers, and obtaining a sorted temperature fluctuation data set;

[0129] Based on the organized temperature fluctuation data set, the correlation between temperature fluctuation and quality indicators is analyzed, and the linear regression analysis method is used to quantify the influence weight and determine the temperature correlation parameters of each processing link;

[0130] If the influence weight of a certain processing link exceeds the preset threshold, the temperature control strategy of this link will be dynamically adjusted to tighten the tolerance range and obtain the optimized parameter set.

[0131] In some embodiments, the process may be as follows:

[0132] First, the temperature control of the processing link is dynamically adjusted based on the temperature-quality correlation parameter. For example, if the temperature-quality correlation parameter of a processing link shows an impact weight of -1.42, which is higher than the preset threshold of -1.0, indicating that temperature fluctuations have a significant impact on quality indicators. The adjustment mechanism is automatically triggered. Through historical data analysis, the temperature records of this link over the past 30 days are extracted and found to have fluctuated between 28.5 and 33.5 degrees Celsius, with an average of 31.0 degrees Celsius and a standard deviation of 1.8.

[0133] Next, a built-in optimization algorithm was invoked, employing a gradient descent-based control parameter adjustment model to tighten the temperature tolerance from the original ±2.5°C to ±1.5°C. The resulting upper temperature limit was calculated to be 32.5°C, with a lower limit of 29.5°C. Simultaneously, the system analyzed the temperature regulator's response time, assuming a 2-second response time, and further fine-tuned the control parameters to ensure that the frequency of temperature fluctuations did not exceed 0.5 times per hour, ultimately forming an optimized set of temperature control parameters.

[0134] In the logic chain, the adjusted parameters are associated with the energy consumption data of the processing link. Assuming that energy consumption increases by 3.2% after the temperature tolerance is tightened, the balance point between energy consumption and quality improvement is automatically calculated to obtain the optimal control strategy. The parameter set is stored in the control database for the equipment to automatically perform adjustments to ensure stable processing quality.

[0135] Furthermore, obtaining abnormal feedback data from real-time monitoring includes:

[0136] By adjusting the sensor logic of IoT devices through an optimized temperature control parameter set, temperature data during processing is continuously collected to obtain real-time monitoring of temperature fluctuations.

[0137] The collected temperature fluctuation information is compared with the preset tolerance range. If the temperature data is detected to be outside the tolerance range, the abnormal alarm mechanism is triggered and the triggering record of the abnormal alarm is obtained;

[0138] For the trigger records of abnormal alarms, the corresponding time records and processing link information are automatically extracted, the abnormal feedback data is saved through data storage, and the detailed archive of abnormal events is determined.

[0139] In some embodiments, the process may be as follows:

[0140] During the processing phase, the optimized temperature control parameter set is first distributed to each sensor device through the IoT platform. For example, if the optimized temperature tolerance range for a certain process is ±1.2 degrees Celsius for both the upper and lower limits, and the target temperature is set at 30.0 degrees Celsius, the sensor control logic is automatically updated, setting the monitoring threshold to the range of 28.8 to 31.2 degrees Celsius.

[0141] Next, the sensor device collects temperature data once a minute. For example, if a value collected is 31.5 degrees Celsius, exceeding the upper threshold of 31.2 degrees Celsius, the built-in anomaly detection algorithm immediately calculates a deviation of 0.3 degrees Celsius. Combined with the historical temperature trend analysis over the past five minutes, it finds that the temperature is continuously rising at a slope of 0.1 degrees Celsius per minute, indicating an abnormal fluctuation.

[0142] Subsequently, the alarm mechanism is automatically triggered, and an abnormal signal is sent to the central control platform through the IoT gateway. The time of the abnormality is recorded as 14:23:45, and the time point, device number, temperature value and other information are stored in the abnormality log database;

[0143] Next, the real-time monitoring module was called to extract feedback data within 10 seconds of the anomaly. Assuming the feedback showed that the temperature control device had initiated a cooling process, but the temperature remained at 31.4 degrees Celsius, further analysis of the control device's response delay revealed a calculated delay of 3.5 seconds, exceeding the standard value of 2.0 seconds, indicating a possible risk of equipment aging.

[0144] To form a complete logical chain, the abnormal data is associated with the batch records of the processing link. Assuming that the current batch is P20231001, the batch is automatically marked as possibly affected by temperature abnormalities, and an abnormality analysis report is generated and pushed to quality monitoring for subsequent traceability and optimization of equipment maintenance strategies to ensure the stability of the processing link.

[0145] Furthermore, the detailed abnormality tracing report includes:

[0146] Through the abnormal feedback data obtained by real-time monitoring, the relevant data fragments in the temperature information library are automatically located according to the time point and processing link identification in the alarm record, and a preliminary data set before and after the abnormality occurs is obtained;

[0147] Based on the preliminary data set, the temperature change sequence before and after the anomaly occurs is extracted and compared using a preset threshold range. If the temperature change exceeds the threshold range, it is marked as a key fluctuation segment, and the distribution characteristics of the key fluctuation segment are determined;

[0148] Based on the distribution characteristics of key fluctuation segments, a time series analysis model of temperature changes is constructed, and the support vector machine model is used to classify the fluctuation segments to obtain the classified fluctuation type information;

[0149] Through the classified fluctuation type information, the corresponding processing link identification is automatically matched, the correlation between temperature changes and processing links is analyzed, and the fluctuation influencing factors related to the links are obtained;

[0150] Based on the fluctuation influencing factors related to each link, the core time points where the anomaly occurred are screened out. Combined with the historical data fragments in the temperature information database, a detailed path for anomaly tracing is constructed to obtain the associated data groups in the tracing path.

[0151] For the associated data groups in the traceability path, the correspondence between the abnormal feedback data and the processing links is automatically archived, and the archived information is saved through data storage to determine the complete information of abnormal traceability.

[0152] In some embodiments, the process may be as follows:

[0153] In the temperature monitoring scenario of the processing link, a problem tracing path is established based on the abnormal feedback data of real-time monitoring, and a series of data processing and analysis processes are automatically executed;

[0154] First, lock the time point of the alarm record, for example, the abnormality occurred at 15:10:30, and combined with the link label "heat treatment segment", access the integrated temperature information database through the data interface, locate the data segment 30 minutes before and after this time point, and extract the temperature change sequence data. Assume that the temperature value in the 10 minutes before the abnormality gradually increased from 29.5 degrees Celsius to 30.8 degrees Celsius, and the temperature value fluctuated between 30.7 and 31.0 degrees Celsius in the 10 minutes after the abnormality;

[0155] Next, a time series analysis algorithm was used to calculate the acceleration of temperature change. It was found that the temperature rose at an acceleration of 0.05 degrees Celsius per minute squared in the five minutes before the anomaly, exceeding the normal fluctuation range of 0.02 degrees Celsius per minute squared, indicating a potential overheating risk.

[0156] Subsequently, an abnormality tracing report was automatically generated, integrating the extracted temperature series data with link labels, time points, and other information. The report marked the critical turning point of the temperature change as 15:05:00 and calculated the abnormal impact duration as 15 minutes.

[0157] At the same time, the traceability results were correlated and compared with the production environment parameter database, and the environmental humidity data for the same period was extracted. Assuming the humidity value was 65%, which is higher than the standard value of 60%, the analysis showed that humidity may exacerbate temperature fluctuations, forming a logical correlation;

[0158] Finally, the traceability report is stored in the exception analysis archive and automatically pushed to the production optimization module, marking the link that may require adjustment of environmental control parameters, forming a complete data processing chain to ensure the stability of subsequent production links.

[0159] Example 2

[0160] like Figure 2 As shown, this embodiment provides a method for controlling temperature in an aquatic product processing production link, including:

[0161] Real-time acquisition of temperature data at each stage of aquatic product processing, high-frequency sampling of temperature changes at each node, and a preliminary temperature monitoring data set are obtained;

[0162] Preprocess the collected preliminary temperature monitoring data set to obtain a cleaned temperature data set;

[0163] Through the cleaned temperature data set, a unified data management platform is built to integrate the temperature information of all processing links into a single database to obtain an integrated temperature information database;

[0164] Based on the integrated temperature information database, feature extraction is performed on the temperature data stream, and the temperature change trend of each processing link is quantitatively analyzed to obtain a temperature feature data set containing abnormal markers;

[0165] By correlating the temperature feature dataset containing abnormal marks with product quality inspection records, we can obtain temperature quality correlation parameters.

[0166] Dynamically adjust the temperature control parameters in the processing process according to the temperature-quality correlation parameters;

[0167] By optimizing the temperature control parameter set and updating the control logic of IoT sensor devices, we can compare the real-time temperature monitoring data of each processing link and obtain real-time monitoring abnormal feedback data.

[0168] Based on the abnormal feedback data from real-time monitoring, a problem tracing path is constructed, which automatically traces back to the corresponding data segment in the integrated temperature information library, extracts the temperature change sequence before and after the abnormality occurs, and obtains a detailed abnormality tracing report.

[0169] Furthermore, obtaining a preliminary temperature monitoring data set includes:

[0170] Collect temperature data from all nodes in the entire link through IoT devices to form an initial temperature monitoring data set;

[0171] Based on the initial temperature monitoring data set, sort the data according to the timestamp identifier of each node to obtain an ordered temperature data sequence;

[0172] Among them, if there are missing values in the ordered temperature data sequence, the missing data are filled by linear interpolation to determine the complete temperature data stream;

[0173] Based on the complete temperature data stream, the support vector machine model is used to analyze the temperature change trend of each node to determine whether there is abnormal fluctuation;

[0174] If the analysis results show that the temperature change trend of a node exceeds the preset threshold, the temperature data stream of the node is segmented to obtain the data subset of the abnormal time period;

[0175] By using the data subset of the abnormal time period and combining the correspondence between the timestamp identifier and the full-link node, the specific raw material storage or finished product packaging link can be located to determine the specific location where the abnormality occurred.

[0176] Based on the positioning results, high-frequency sampling data within the relevant time period is extracted for the specific location where the anomaly occurs, and detailed temperature change records are obtained for subsequent processing.

[0177] Furthermore, obtaining the cleaned temperature data set includes:

[0178] Based on the preliminary temperature monitoring data set, a data cleaning algorithm is used to preprocess the collected temperature data stream to identify and repair missing values and outliers in the data;

[0179] If it is detected that the temperature value corresponding to a certain timestamp exceeds the preset reasonable range threshold, the temperature value of the previous and next time points is supplemented by linear interpolation to obtain a cleaned temperature dataset;

[0180] Specifically, by further processing the cleaned data set, the temperature data distribution characteristics under each timestamp are obtained and its fluctuation range is determined;

[0181] According to the distribution characteristics, statistical analysis tools are used to evaluate the stability of temperature data and obtain the stability index;

[0182] If the stability index exceeds the preset threshold, the temperature data segment corresponding to the timestamp is extracted based on the result of the outlier identification to determine whether there is a persistent deviation.

[0183] By segmenting the temperature data fragments with persistent deviations, the time interval of abnormal fluctuations is obtained and the critical time range is determined;

[0184] Based on the key time range, combined with the association between timestamp identifiers and temperature monitoring, the historical records of relevant temperature data are extracted to obtain detailed fluctuation information;

[0185] The support vector machine model is used to classify the detailed fluctuation information and determine whether the fluctuation is within the controllable range;

[0186] Among them, if the fluctuation exceeds the controllable range, the change pattern of the deviation trend is obtained through comparative analysis of historical records and current temperature data to determine the basis for adjustment.

[0187] Furthermore, the obtaining of the integrated temperature information database includes:

[0188] By using the temperature data after cleaning, a unified management platform is built to integrate data streams from raw material storage, processing, and finished product packaging into a single database. Classification labels are added for different links to obtain an integrated information database.

[0189] Obtaining a temperature feature dataset containing abnormal markers includes:

[0190] Through the integrated temperature information database, the temperature information of each processing link is preliminarily sorted out, and the time series analysis method is used to structure the data to obtain the sorted temperature data stream;

[0191] Based on the organized temperature data stream, the temperature information of each processing link is tested for change trends. The fluctuation amplitude in the trend is compared using a preset threshold range. If the fluctuation amplitude of a link exceeds the threshold range, it is marked as a potential anomaly point, resulting in a data stream with potential anomaly marks.

[0192] Through the data stream with potential anomaly markers, the time window is divided for the marked potential anomaly points, the time period distribution of the anomaly points is obtained, and the time interval range of the anomaly points is determined;

[0193] Based on the time range of the abnormal point, the temperature information of the corresponding processing link is traced back to obtain relevant records in the historical data to determine whether there are periodic fluctuation characteristics. If there are periodic fluctuation characteristics, the abnormal point is classified and labeled to obtain the classified abnormal data group;

[0194] Through the classified abnormal data group, the correlation analysis is carried out on the abnormal points in each processing link, and the support vector machine model is used to classify the distribution characteristics of the abnormal points to obtain the category distribution of the abnormal points;

[0195] Based on the distribution of abnormal points, the temperature information of each processing link is screened for key periods of time to obtain the time intervals with the highest incidence of abnormal points and determine the time period range for key monitoring;

[0196] Through the key monitoring time period, real-time data collection is carried out on the temperature data stream of the corresponding processing link, and the real-time data is subjected to fluctuation detection using preset rules. If the fluctuation amplitude of the real-time data exceeds the preset threshold, it is marked as a real-time abnormal event, and a set of abnormal markers for real-time monitoring is obtained.

[0197] Based on the integrated information database, the temperature data streams of each link are stored in layers, and the data are grouped using preset classification labels to obtain layered data sets;

[0198] Through the stratified data set, the time series continuity test is performed on the temperature data stream under each classification label. If a time series interruption is detected in the data stream of a certain link, the timestamp is supplemented through historical data records to determine the supplemented data sequence;

[0199] Based on the completed data sequence, a preliminary analysis of the fluctuation range of the temperature data flow at each link is conducted. Statistical tools are used to calculate the distribution characteristics of the data and obtain the fluctuation characteristic values of each link.

[0200] Through the fluctuation characteristic value of each link, the data flow under the classification label is compared across links. If the fluctuation characteristic value of a link exceeds the preset threshold range, the link is marked as a key monitoring object, and the marked data group is obtained;

[0201] According to the marked data grouping, deep feature extraction is performed on the temperature data stream of key monitoring objects, and the support vector machine model is used to classify the fluctuation features and determine the category distribution of the fluctuation features;

[0202] Through the category distribution of fluctuation characteristics, the temperature data stream of key monitoring objects is divided into time intervals, the time period distribution of abnormal fluctuations is obtained, and the time interval range of abnormal fluctuations is determined.

[0203] Furthermore, the obtaining of temperature quality related parameters includes:

[0204] By associating the temperature feature dataset containing abnormality marks with quality inspection records, a preliminary match is performed between the temperature anomalies and the product batch data. The data alignment method is used to organize the correspondence between the temperature features and the quality data to obtain a preliminary matching dataset.

[0205] Based on the initially matched data set, feature extraction is performed on the correlation between temperature anomalies and quality data. Pre-set rules are used to compare temperature fluctuations with quality indicators. If the temperature fluctuation exceeds the preset threshold and the quality indicator deviates, it is marked as a key correlation point, and the key correlation point set is determined;

[0206] Through the collection of key correlation points, the distribution characteristics of temperature fluctuations and quality indicators are modeled and processed. The correlation between temperature characteristics and quality data is quantified using regression analysis methods to obtain the correlation parameters between temperature and quality.

[0207] Based on the correlation parameters between temperature and quality, the temperature anomaly data of each product batch is classified and processed. The time period with large temperature fluctuations is extracted through data screening methods to obtain the abnormal intervals of key concern.

[0208] Through the abnormal intervals of focus, the quality data of the corresponding product batches are retrospectively analyzed. If the quality data shows obvious deviations within the abnormal intervals, it is marked as a high-risk batch and a high-risk batch list is determined;

[0209] Based on the high-risk batch list, relevant temperature characteristic data is monitored in real time. Temperature fluctuations are continuously tracked using pre-set monitoring rules to obtain a real-time abnormality mark set.

[0210] Through the real-time abnormality mark collection, the quality data of high-risk batches are dynamically updated, and the real-time temperature abnormalities are correlated and compared with quality indicators using data integration methods to determine the final abnormality and quality association records.

[0211] Furthermore, the optimized temperature control parameter set includes: extracting temperature control records in the processing link from historical data based on the relationship between temperature fluctuations and quality indicators, using a data cleaning method to remove outliers, and obtaining a sorted temperature fluctuation data set;

[0212] Based on the organized temperature fluctuation data set, the correlation between temperature fluctuation and quality indicators is analyzed, and the linear regression analysis method is used to quantify the influence weight and determine the temperature correlation parameters of each processing link;

[0213] If the influence weight of a certain processing link exceeds the preset threshold, the temperature control strategy of that link is dynamically adjusted to tighten the tolerance range and obtain the optimized parameter set;

[0214] Based on the optimized parameter set, the temperature control equipment in the processing link is updated with parameters, and the adjusted temperature tolerance range is automatically issued to obtain the updated control records;

[0215] Based on the updated control records, the temperature fluctuation data in the processing link is continuously collected. The data comparison method is used to determine whether the temperature is stable within the adjusted tolerance range and determine the stability status of the temperature control;

[0216] If the stability of temperature control does not meet the preset standards, the temperature-related parameters are recalibrated, and data analysis tools are used to extract abnormal fluctuation points to obtain the calibrated adjustment strategy;

[0217] Based on the calibrated adjustment strategy, the temperature control of the processing link is further optimized, and the final temperature control plan is determined by automatically updating the tolerance range and monitoring frequency.

[0218] Furthermore, obtaining abnormal feedback data from real-time monitoring includes:

[0219] By adjusting the sensor logic of IoT devices through an optimized temperature control parameter set, temperature data during processing is continuously collected to obtain real-time monitoring of temperature fluctuations.

[0220] The collected temperature fluctuation information is compared with the preset tolerance range. If the temperature data is detected to be outside the tolerance range, the abnormal alarm mechanism is triggered and the triggering record of the abnormal alarm is obtained;

[0221] For the trigger records of abnormal alarms, the corresponding time records and processing link information are automatically extracted, abnormal feedback data is saved through data storage, and detailed files of abnormal events are determined;

[0222] Based on the detailed archive of abnormal feedback data, the fluctuation pattern of temperature data in specific processing links is analyzed, and the support vector machine model is used to classify the abnormal fluctuation points to obtain the classified abnormal feature set;

[0223] Based on the classified abnormal feature set, a targeted parameter adjustment strategy is automatically generated, and the control parameters of the sensor logic are updated through the IoT device to obtain the adjusted monitoring rules;

[0224] Furthermore, the adjusted monitoring rules include:

[0225] By recording the triggering of abnormal alarms, the corresponding time points and processing link information are automatically obtained, and the feedback data is persistently saved using data storage to build a detailed archive of abnormal events;

[0226] Based on the constructed detailed archive, the fluctuation anomaly information related to the processing link is extracted, and the support vector machine model is used to classify the fluctuation anomaly points to obtain the classified feature set;

[0227] For the classified feature set, analyze its distribution characteristics in the processing link. Through preset threshold comparison, if the distribution of abnormal points in the feature set exceeds the threshold range, generate the corresponding parameter adjustment strategy and determine the adjustment direction;

[0228] According to the generated parameter adjustment strategy, the control parameters of IoT devices are automatically updated, and the updated control parameter data is obtained to form new monitoring rules;

[0229] Through the updated monitoring rules, real-time data from the processing link is continuously collected. If the real-time data does not meet the preset threshold range, the data collection frequency is adjusted to obtain an optimized collection mode;

[0230] According to the optimized collection mode, the data is automatically sent to the sensor module of the IoT device, and the real-time data is continuously compared to determine whether it meets the requirements of the monitoring rules and determine the final control logic;

[0231] Through the final control logic, the data stability in the processing link is regularly verified to obtain long-term operation status information and determine whether the monitoring rules need to be further optimized;

[0232] For example, in a temperature monitoring scenario during a processing step, the system automatically extracts the time of the abnormality and the corresponding processing step information based on the triggering record of the abnormality alarm. For example, if the abnormality record time is 15:47:32 and the processing step is the heat treatment stage, the abnormality feedback data is saved in a distributed database through data storage to form a detailed file containing information such as the temperature value, equipment number, and ambient humidity. For example, if the temperature is 33.5 degrees Celsius, the equipment number is HT2023-001, and the humidity is 65%;

[0233] Subsequently, a support vector machine model was used to classify abnormal temperature data fluctuations. The temperature data series from the past 30 minutes was extracted, and the fluctuation frequency was calculated to be once every 10 minutes with a standard deviation of 0.8 degrees Celsius. Through model analysis, the abnormal points were classified as "periodic overheating" and a set of abnormal features was generated, including characteristic values such as a fluctuation amplitude of 1.5 degrees Celsius and a duration of 8 minutes.

[0234] Then, based on the classification results and feature set, a parameter adjustment strategy was automatically generated. It calculated that the temperature monitoring frequency needed to be increased from once per minute to once every 30 seconds, and the upper alarm threshold should be adjusted from 33.0 degrees Celsius to 32.5 degrees Celsius to improve sensitivity.

[0235] The adjusted control parameters are then distributed to the sensor devices via the IoT platform, and monitoring rules are updated to ensure that subsequent data collection complies with the new threshold settings. The adjustment record is also associated with the processing batch number HT20231002, and a parameter change log is automatically generated and stored in a cloud database, forming a complete data traceability chain.

[0236] If similar anomalies are detected subsequently, the historical adjustment strategy will be called for comparative analysis and the strategy matching degree will be calculated. Assuming the matching degree is 85%, the historical strategy will be reused directly to reduce the consumption of repeated computing resources and ensure the continuity of the processing link and the intelligence of data processing.

[0237] Through the adjusted monitoring rules, temperature data from the processing link is continuously collected and compared with the updated tolerance range. If the temperature data is still outside the range, the monitoring frequency is further adjusted to determine the final stable control solution;

[0238] According to the final stable control plan, it is automatically sent to the sensor logic module of the IoT device to continuously verify the real-time monitoring data and obtain long-term temperature control stability data.

[0239] Furthermore, the detailed abnormality tracing report includes:

[0240] Through the abnormal feedback data obtained by real-time monitoring, the relevant data fragments in the temperature information library are automatically located according to the time point and processing link identification in the alarm record, and a preliminary data set before and after the abnormality occurs is obtained;

[0241] Based on the preliminary data set, the temperature change sequence before and after the anomaly occurs is extracted and compared using a preset threshold range. If the temperature change exceeds the threshold range, it is marked as a key fluctuation segment, and the distribution characteristics of the key fluctuation segment are determined;

[0242] Based on the distribution characteristics of key fluctuation segments, a time series analysis model of temperature changes is constructed, and the support vector machine model is used to classify the fluctuation segments to obtain the classified fluctuation type information;

[0243] Through the classified fluctuation type information, the corresponding processing link identification is automatically matched, the correlation between temperature changes and processing links is analyzed, and the fluctuation influencing factors related to the links are obtained;

[0244] Based on the fluctuation influencing factors related to each link, the core time points where the anomaly occurred are screened out. Combined with the historical data fragments in the temperature information database, a detailed path for anomaly tracing is constructed to obtain the associated data groups in the tracing path.

[0245] For the associated data groups in the traceability path, the corresponding relationship between abnormal feedback data and processing links is automatically archived, and the archived information is saved through data storage to determine the complete information of abnormal traceability;

[0246] Furthermore, the complete information for determining abnormal tracing includes:

[0247] For key fluctuation segments, a preset threshold range is used for preliminary screening. Data segments of abnormal intervals are extracted from the temperature variation. The fluctuation characteristic distribution within the abnormal interval is obtained to determine the preliminary fluctuation type.

[0248] According to the fluctuation type, a time series analysis framework is constructed to analyze the continuity of temperature changes in the time dimension, extract distribution eigenvalues, and obtain the periodic law of fluctuation changes;

[0249] By combining the distribution characteristic values with the processing link labels, the corresponding relationship between the temperature change and the processing link is analyzed. If the temperature change exceeds the preset threshold range, it is marked as an abnormal link, and the fluctuation influencing factors related to the link are determined;

[0250] Based on the factors influencing the fluctuation, the abnormal tracing path is sorted out, the related data groups are automatically matched, and relevant data fragments are extracted from the historical temperature records to obtain the complete data sequence before and after the abnormality occurs;

[0251] Based on the complete data sequence, analyze the key time nodes in the abnormality traceability path. If there is a corresponding relationship between the key time nodes and the processing link labels, classify them as key monitoring objects and determine the distribution pattern of abnormal data;

[0252] By combining the distribution pattern of abnormal data with the automatic archiving system, the abnormal feedback data is associated with the processing link mark and stored in the data storage to obtain the data mapping relationship after archiving;

[0253] According to the mapping relationship of archived data, the basis for complete information judgment is analyzed. If the archived data covers all key nodes of the abnormal traceability path, the integrity of the traceability information is confirmed and the final abnormality analysis results are determined;

[0254] For example, in a temperature monitoring scenario during processing, a series of automated processes are used to analyze the correlation between temperature fluctuations and processing steps, and to build a detailed abnormality tracing path;

[0255] First, key fluctuation segments were extracted from historical temperature data. For example, between 2:20:00 PM and 2:40:00 PM, the temperature rapidly rose from 28.3°C to 29.9°C, with a fluctuation of 1.6°C, exceeding the normal range by 0.8°C. Based on this data, a time series analysis model was constructed. A support vector machine algorithm was used to classify the fluctuation segments. The feature vectors, including the fluctuation amplitude, duration, and rate of change, were calculated. The classification result was "rapid temperature rise," and the processing phase was automatically identified as the "preheating phase."

[0256] Subsequently, the correlation between temperature changes and processing steps was analyzed. By comparing historical data, it was found that temperature fluctuations during the preheating phase were related to the equipment power setting. Assuming a power setting of 75%, which is higher than the standard value of 70%, the temperature rise rate reached 0.08 degrees Celsius per minute, exceeding the safety threshold of 0.03 degrees Celsius per minute. The power setting was determined to be the primary influencing factor.

[0257] Next, an abnormal traceability path is generated, and the fluctuation type, link identifier, and influencing factors are integrated into a related data group. The key time point 14:25:00 is locked as the fluctuation starting point, and the production batch data covering the next 10 minutes of the impact range is calculated to form a complete traceability information.

[0258] At the same time, the correspondence between abnormal feedback data and processing links is automatically archived, and the information is saved to the abnormal archive database through data storage, generating a unique identification code such as "EXP202310-001" for subsequent inquiries

[0259] Finally, based on the archived information, the integrity of the abnormality traceability is judged, and the logical verification algorithm is called to verify that the matching degree of temperature fluctuations, link identification and influencing factors in the data group reaches more than 95%. The traceability information is confirmed to be complete, and the results are synchronized to the production monitoring platform to form a closed-loop analysis chain.

[0260] Furthermore, it also includes:

[0261] Through detailed anomaly tracing reports, the temperature-quality correlation model is updated, and the model parameters are iteratively optimized based on the newly collected anomaly data and quality data. The incremental learning method is used to adjust the weight of the impact of temperature fluctuations on quality, and the updated correlation model parameters are obtained for temperature control optimization in subsequent processing links.

[0262] The obtaining of updated association model parameters includes:

[0263] By collecting abnormal data and quality data, we build an initial data set, pre-process the collected data, clean the noise information, and obtain the processed data set;

[0264] Extract characteristic information related to temperature fluctuations from the processed data set and compare it with a preset threshold range. If the characteristic information exceeds the threshold range, it is marked as an abnormal fluctuation feature, and the distribution of the abnormal fluctuation feature is determined.

[0265] By analyzing the distribution of abnormal fluctuation characteristics and combining them with historical records in the traceability report, we can analyze the corresponding relationship between temperature fluctuations and quality impacts and obtain the corresponding impact weight information.

[0266] According to the influence weight information, the structure of the temperature correlation model is adjusted, and the model parameters are updated using the incremental learning method to obtain the adjusted model parameter set;

[0267] The correlation between temperature fluctuation and quality impact is verified using the adjusted model parameter set. If the verification result deviates significantly from the historical data, the parameters are fine-tuned to determine the final parameter configuration.

[0268] Based on the final parameter configuration, update the operating logic of the temperature-quality correlation model and obtain the updated model operating rules;

[0269] Through the updated model operation rules, real-time analysis is performed on the subsequently collected data to determine the dynamic changes in the impact of temperature fluctuations on quality and obtain an assessment of the fluctuation impact under real-time monitoring.

[0270] In some embodiments, the process may be as follows:

[0271] In the temperature monitoring scenario of the processing link, the temperature quality correlation model is updated based on the abnormal traceability report, and a series of data processing and analysis processes are automatically executed;

[0272] First, extract newly collected abnormal data from the abnormality tracing report. For example, during a certain processing, the temperature fluctuation range was between 31.2 and 32.5 degrees Celsius, exceeding the standard upper limit of 31.0 degrees Celsius. At the same time, obtain the corresponding quality data. Assume that the finished product defect rate is 3.2%, which is 2.0% higher than the normal value.

[0273] This data was then fed into a temperature-quality correlation model, and an incremental learning algorithm was used to iteratively optimize the model parameters. The weight adjustment for the impact of temperature fluctuation on quality was calculated. The original weight was 0.65, but after analysis, it was adjusted to 0.72, reflecting the greater impact of temperature fluctuation on defect rates.

[0274] Next, a regression analysis algorithm was used to further quantify the correlation between temperature and quality. It was found that for every 0.5°C increase in temperature, the defect rate increased by approximately 0.3%. This relationship was then updated into the model parameters to form a new prediction basis.

[0275] Then, the updated model parameters are compared with the historical processing data for verification. Assuming that the prediction error on the verification set is reduced from the original 1.8% to 1.5%, it is confirmed that the model optimization is effective.

[0276] Finally, the updated parameters are automatically synchronized to the temperature control module to generate new control thresholds, such as adjusting the upper temperature limit to 30.8 degrees Celsius, and linked to the production batch management database to ensure that subsequent processing links are adjusted in real time according to the new model, forming a complete closed loop from abnormal data to control optimization.

[0277] Example 3

[0278] like Figure 3 As shown, this embodiment provides a temperature control system for an aquatic product processing production link, including:

[0279] The data acquisition and preprocessing module is used to collect and store temperature data at high frequency in real time at all nodes in the entire aquatic product processing chain, and then clean and repair the data to obtain a cleaned temperature data set;

[0280] Specifically, the data acquisition and preprocessing module is used to acquire temperature data in real time at each stage of aquatic product processing by deploying IoT sensor devices, covering the entire chain of nodes from raw material storage to finished product packaging. High-frequency sampling is performed on the temperature changes of each node, with the sampling frequency set to once per minute. The acquired temperature data stream is stored in the cloud database with a timestamp as the identifier to obtain a preliminary temperature monitoring data set; and based on the preliminary temperature monitoring data set, a data cleaning algorithm is used to preprocess the collected temperature data stream, identify and repair missing values and outliers in the data, and if it is detected that the temperature value corresponding to a certain timestamp exceeds the preset reasonable range threshold, it is supplemented based on the temperature values of the previous and next time points through linear interpolation to obtain a cleaned temperature data set;

[0281] The data integration and feature extraction module is used to integrate the cleaned temperature data and classify and label them, build a temperature information database, extract temperature change trend features, mark abnormal events, and obtain a temperature feature dataset;

[0282] Specifically, the data integration and feature extraction module is used to build a unified data management platform based on the cleaned temperature data set, integrate the temperature information of all processing links into a single database, set classification labels for the temperature data streams of different links, including categories such as raw material storage, processing and finished product packaging, and obtain an integrated temperature information database; and based on the integrated temperature information database, apply time series analysis methods to extract features from the temperature data stream, and conduct quantitative analysis of the temperature change trend of each processing link. If the temperature fluctuation amplitude of a link exceeds the preset fluctuation threshold, it is marked as an abnormal event, and a temperature feature dataset containing abnormal labels is obtained;

[0283] The correlation analysis and parameter optimization module is used to correlate temperature feature data sets with product quality data, build correlation models, calculate influence weights, and dynamically adjust temperature control parameters based on these data to obtain an optimized parameter set.

[0284] Specifically, the association analysis and parameter optimization module is used to associate product quality inspection records with a temperature feature dataset containing abnormality markers, match temperature anomaly events with product quality data of the corresponding batch, construct a temperature-quality association model based on the matching results, and use a regression analysis algorithm to calculate the weight of the impact of temperature fluctuations on quality indicators to obtain temperature-quality association parameters. Furthermore, based on the temperature-quality association parameters, the module dynamically adjusts the temperature control parameters in the processing steps. If the temperature-quality association parameters of a certain step indicate that the weight of the impact of temperature fluctuations on quality indicators is higher than a preset threshold, the temperature tolerance range of that step is reduced to obtain an optimized temperature control parameter set.

[0285] Real-time monitoring and abnormality feedback module, used to update the control logic based on the optimized parameter set, monitor temperature data in real time, and alarm and record abnormal data if it exceeds the tolerance range;

[0286] Specifically, the real-time monitoring and abnormality feedback module is used to update the control logic of the IoT sensor equipment through the optimized temperature control parameter set, compare the real-time temperature monitoring data of each processing link, and trigger an automatic alarm and record the abnormal time point if the temperature value is detected to be outside the optimized tolerance range, and obtain real-time monitoring abnormality feedback data;

[0287] The anomaly tracing and model updating module is used to trace the problem based on the anomaly feedback data, extract the temperature sequence before and after the anomaly, update the correlation model, and optimize the model parameters for subsequent temperature control optimization;

[0288] Specifically, the anomaly tracing and model updating module is used to build a problem tracing path based on the anomaly feedback data monitored in real time. It automatically traces back to the corresponding data segment in the integrated temperature information library for the time point and link label of the alarm record, extracts the temperature change sequence before and after the anomaly occurs, and obtains a detailed anomaly tracing report; and through the detailed anomaly tracing report, it updates the temperature-quality association model, iteratively optimizes the model parameters for the newly collected anomaly data and quality data, uses the incremental learning method to adjust the weight of the impact of temperature fluctuations on quality, and obtains the updated association model parameters for temperature control optimization in subsequent processing links.

[0289] While the specific embodiments of the present invention have been described in detail above, they are intended only as examples, and the present invention is not limited to the specific embodiments described above. Those skilled in the art should understand that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and that various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A temperature control method for aquatic product processing production link, characterized in that: include: Real-time acquisition of temperature data at each stage of aquatic product processing, high-frequency sampling of temperature changes at each node, and a preliminary temperature monitoring data set are obtained; Preprocess the collected preliminary temperature monitoring data set to obtain a cleaned temperature data set; Through the cleaned temperature data set, a unified data management platform is built to integrate the temperature information of all processing links into a single database to obtain an integrated temperature information database; Based on the integrated temperature information database, feature extraction is performed on the temperature data stream, and the temperature change trend of each processing link is quantitatively analyzed to obtain a temperature feature data set containing abnormal markers; By correlating the temperature feature dataset containing abnormal marks with product quality inspection records, we can obtain temperature quality correlation parameters. Dynamically adjust the temperature control parameters in the processing process according to the temperature-quality correlation parameters; By optimizing the temperature control parameter set and updating the control logic of IoT sensor devices, we can compare the real-time temperature monitoring data of each processing link and obtain real-time monitoring abnormal feedback data. Based on the abnormal feedback data from real-time monitoring, a problem tracing path is constructed, which automatically traces back to the corresponding data segment in the integrated temperature information library, extracts the temperature change sequence before and after the abnormality occurs, and obtains a detailed abnormality tracing report.

2. A temperature control method for aquatic product processing production link according to claim 1, characterized in that: The obtaining of a preliminary temperature monitoring data set includes: Collect temperature data from all nodes in the entire link through IoT devices to form an initial temperature monitoring data set; Based on the initial temperature monitoring data set, sort the data according to the timestamp identifier of each node to obtain an ordered temperature data sequence; Among them, if there are missing values in the ordered temperature data sequence, the missing data are filled by linear interpolation to determine the complete temperature data stream.

3. The temperature control method for aquatic product processing production link according to claim 1, characterized in that: The cleaned temperature data set includes: Based on the preliminary temperature monitoring data set, a data cleaning algorithm is used to preprocess the collected temperature data stream to identify and repair missing values and outliers in the data; If it is detected that the temperature value corresponding to a certain timestamp exceeds the preset reasonable range threshold, the temperature value of the previous and next time points is supplemented by linear interpolation to obtain a cleaned temperature dataset.

4. The temperature control method for aquatic product processing production link according to claim 1, characterized in that: The obtaining of the integrated temperature information database includes: By using the temperature data after cleaning, a unified management platform is built to integrate data streams from raw material storage, processing, and finished product packaging into a single database. Classification labels are added for different links to obtain an integrated information database. Obtaining a temperature feature dataset containing abnormal markers includes: Through the integrated temperature information database, the temperature information of each processing link is preliminarily sorted out, and the time series analysis method is used to structure the data to obtain the sorted temperature data stream; Based on the sorted temperature data stream, the temperature information of each processing link is tested for change trends, and the fluctuation amplitude in the trend is compared using a preset threshold range. If the fluctuation amplitude of a certain link exceeds the threshold range, it is marked as a potential anomaly point, and a data stream with potential anomaly marks is obtained.

5. The temperature control method for aquatic product processing production link according to claim 1, characterized in that: The obtaining of temperature quality related parameters includes: By associating the temperature feature dataset containing abnormality marks with quality inspection records, a preliminary match is performed between the temperature anomalies and the product batch data. The data alignment method is used to organize the correspondence between the temperature features and the quality data to obtain a preliminary matching dataset. Based on the initially matched data set, feature extraction is performed on the correlation between temperature anomalies and quality data. Pre-set rules are used to compare temperature fluctuations with quality indicators. If the temperature fluctuation exceeds the preset threshold and the quality indicator deviates, it is marked as a key correlation point, and the key correlation point set is determined; Through the set of key correlation points, the distribution characteristics of temperature fluctuations and quality indicators are modeled and processed. The regression analysis method is used to quantify the correlation between temperature characteristics and quality data, and the correlation parameters between temperature and quality are obtained.

6. The temperature control method for aquatic product processing production link according to claim 1, characterized in that: The optimized temperature control parameter set includes: In view of the relationship between temperature fluctuation and quality indicators, the temperature control records in the processing links were extracted from the historical data. The data cleaning method was used to remove outliers to obtain the sorted temperature fluctuation data set. Based on the organized temperature fluctuation data set, the correlation between temperature fluctuation and quality indicators is analyzed, and the linear regression analysis method is used to quantify the influence weight and determine the temperature correlation parameters of each processing link; If the influence weight of a certain processing link exceeds the preset threshold, the temperature control strategy of this link will be dynamically adjusted to tighten the tolerance range and obtain the optimized parameter set.

7. The temperature control method for aquatic product processing production link according to claim 1, characterized in that: The acquisition of abnormal feedback data for real-time monitoring includes: By adjusting the sensor logic of IoT devices through an optimized temperature control parameter set, temperature data during processing is continuously collected to obtain real-time monitoring of temperature fluctuations. The collected temperature fluctuation information is compared with the preset tolerance range. If the temperature data is detected to be outside the tolerance range, the abnormal alarm mechanism is triggered and the triggering record of the abnormal alarm is obtained; For the trigger records of abnormal alarms, the corresponding time records and processing link information are automatically extracted, the abnormal feedback data is saved through data storage, and the detailed archive of abnormal events is determined.

8. The temperature control method for aquatic product processing production link according to claim 1, characterized in that: The detailed abnormal tracing report includes: Through the abnormal feedback data obtained by real-time monitoring, the relevant data fragments in the temperature information library are automatically located according to the time point and processing link identification in the alarm record, and a preliminary data set before and after the abnormality occurs is obtained; Based on the preliminary data set, the temperature change sequence before and after the anomaly occurs is extracted and compared using a preset threshold range. If the temperature change exceeds the threshold range, it is marked as a key fluctuation segment, and the distribution characteristics of the key fluctuation segment are determined; Based on the distribution characteristics of key fluctuation segments, a time series analysis model of temperature changes is constructed, and the support vector machine model is used to classify the fluctuation segments to obtain the classified fluctuation type information; Through the classified fluctuation type information, the corresponding processing link identification is automatically matched, the correlation between temperature changes and processing links is analyzed, and the fluctuation influencing factors related to the links are obtained; Based on the fluctuation influencing factors related to each link, the core time points where the anomaly occurred are screened out. Combined with the historical data fragments in the temperature information database, a detailed path for anomaly tracing is constructed to obtain the associated data groups in the tracing path. For the associated data groups in the traceability path, the correspondence between the abnormal feedback data and the processing links is automatically archived, and the archived information is saved through data storage to determine the complete information of abnormal traceability.

9. The temperature control method for aquatic product processing production link according to claim 1, characterized in that: Also includes: Through detailed anomaly tracing reports, the temperature-quality correlation model is updated, and the model parameters are iteratively optimized based on the newly collected anomaly data and quality data. The incremental learning method is used to adjust the weight of the impact of temperature fluctuations on quality, and the updated correlation model parameters are obtained for temperature control optimization in subsequent processing links.

10. A temperature control system for aquatic product processing production chain, characterized in that: include: The data acquisition and preprocessing module is used to collect and store temperature data at high frequency in real time at all nodes in the entire aquatic product processing chain, and then clean and repair the data to obtain a cleaned temperature data set; The data integration and feature extraction module is used to integrate the cleaned temperature data and classify and label them, build a temperature information database, extract temperature change trend features, mark abnormal events, and obtain a temperature feature dataset; The correlation analysis and parameter optimization module is used to correlate temperature feature data sets with product quality data, build correlation models, calculate influence weights, and dynamically adjust temperature control parameters based on these data to obtain an optimized parameter set. Real-time monitoring and abnormality feedback module, used to update the control logic based on the optimized parameter set, monitor temperature data in real time, and alarm and record abnormal data if it exceeds the tolerance range; The anomaly tracing and model updating module is used to trace the problem based on the anomaly feedback data, extract the temperature sequence before and after the anomaly, update the associated model, and optimize the model parameters for subsequent temperature control optimization.

Images