Remote monitoring method and system for food freshness

Abstract

The invention relates to a remote monitoring method and system for food freshness, and the method comprises the steps: obtaining the processing technology data and circulation transportation data of fresh food, and carrying out the shelf life calculation, thereby obtaining the shelf life of the food; performing process detection and abnormity identification on the processing process data to obtain a first detection result; detecting a temperature accumulation value and a transportation state of the circulation transportation data, and performing result integration to obtain a second detection result; and performing response regulation and control according to the food shelf life, the first detection result and the second detection result to obtain remote regulation and control information. According to the invention, fine management of a cross-link risk conduction mechanism can be realized, and on-demand regulation and control and timely intervention can be realized.

Description

Technical Field

[0001] This invention relates to the technical field of air source heat pumps, and in particular to a method and system for remotely monitoring the freshness of food. Background Technology

[0002] With increasingly demanding food quality from consumers and the growing complexity of global supply chains, achieving precise and intelligent status perception and decision-making control throughout the entire process from production and processing to transportation and sales has become a key direction for technological upgrading in the industry. Existing monitoring methods mostly focus on data collection and analysis at a single stage, such as offline detection of processing parameters or independent monitoring of temperature and humidity changes during distribution. They lack in-depth integration of multi-source heterogeneous data and dynamic correlation analysis across stages. This fragmented monitoring model struggles to effectively capture the quality deterioration mechanisms resulting from the coupling effect of processing anomalies and the distribution environment, leading to significant errors in food shelf-life predictions and delayed anomaly responses. This fails to meet the needs of collaborative management and precise risk intervention across multiple stages in the modern food supply chain. Summary of the Invention

[0003] The main objective of this invention is to provide a method and system for remote monitoring of food freshness, which enables refined management of risk transmission mechanisms across different stages, and facilitates on-demand regulation and timely intervention.

[0004] To achieve the above objectives, the present invention provides a method for remote monitoring of food freshness, comprising: Obtain processing technology data and distribution and transportation data of fresh food, calculate shelf life, and obtain the food's shelf life; The processing data is subjected to process inspection and anomaly identification to obtain a first inspection result; The temperature accumulation value and transportation status of the circulation and transportation data are detected, and the results are integrated to obtain a second detection result; Based on the food's shelf life, the first test result, and the second test result, response control is performed to obtain remote control information.

[0005] Preferably, the step of obtaining processing technology data and distribution and transportation data of fresh food, and calculating the shelf life to obtain the food shelf life includes: The cleaning intensity data and packaging barrier data are extracted from the processing technology data, and the quality decay rate is calculated to obtain the food baseline spoilage rate. Temperature extraction and environmental classification are performed on the aforementioned transportation data to obtain transportation temperature data and environmental type identifiers; A temperature impact analysis is performed based on the transport temperature data and the environmental type identifier to obtain the temperature impact coefficient. The shelf life of the food is obtained by dynamically fitting the food's baseline spoilage rate and the temperature influence coefficient.

[0006] Preferably, the step of extracting cleaning strength data and packaging barrier data from the processing technology data, calculating the quality degradation rate, and obtaining the food baseline spoilage rate includes: The cleaning intensity data is used to evaluate the cleaning effectiveness to obtain cleaning quality data; Calculate the barrier property coefficients of the packaging barrier data to obtain the gas barrier coefficient and humidity barrier coefficient; By combining the cleanliness quality data, the gas barrier coefficient, and the humidity barrier coefficient, a quality degradation factor is calculated to obtain the quality degradation coefficient. The quality decay process is simulated based on the quality decay coefficient, and the quality decay rate is calculated to obtain the food baseline spoilage rate.

[0007] Preferably, the step of performing process inspection and anomaly identification on the processing data to obtain a first detection result includes: The processing data is subjected to process parameter extraction and time alignment processing to obtain a process parameter sequence; Based on the process parameter sequence, state anomaly detection and process stage matching are performed to obtain process anomaly indicators. Based on the aforementioned process anomaly indicators, process execution anomaly analysis is performed to obtain process judgment results and anomaly data point set; The first detection result is obtained by classifying the abnormality level and assessing the severity of the abnormality using the abnormal data point set, the process abnormality index, and the process judgment result.

[0008] Preferably, the step of classifying and assessing the severity of anomalies by combining the abnormal data point set, the process anomaly indicators, and the process judgment results to obtain the first detection result includes: The abnormal data point set is subjected to food processing abnormal pattern recognition processing to obtain the processing abnormality type; Based on the processing anomaly type, the process anomaly indicators are analyzed for processing pollution to obtain pollution impact data; Based on the pollution impact data, a safety assessment is performed on the process determination results to obtain a safety level; The food processing is determined by combining the safety level and the type of processing abnormality to obtain the first detection result.

[0009] Preferably, the step of detecting the cumulative temperature value and transportation status of the circulation and transportation data, and integrating the results to obtain a second detection result, includes: The transportation data is segmented by time to obtain multiple transportation time periods; Temperature sampling and mean calculation are performed on the data for each of the aforementioned transportation periods to obtain the average temperature value for each transportation period. The temperature cumulative effect is calculated based on each average temperature value and the corresponding transportation period data to obtain the temperature cumulative value; Vibration frequency analysis and impact event detection are performed on the aforementioned transportation data to obtain transportation status data; A second detection result is obtained by performing state fusion detection based on the accumulated temperature value and the transportation status data.

[0010] Preferably, the step of responding and adjusting based on the food's shelf life, the first test result, and the second test result to obtain remote control information includes: A time sensitivity analysis was performed on the shelf life of the food to obtain the spoilage sensitive time. Anomaly correlation is performed between the first detection result and the second detection result to obtain anomaly correlation information; Based on the deterioration sensitivity time and the first detection result, process weight allocation is performed to obtain the process influence coefficient; The freshness is assessed based on the process influence coefficient and the second test result to obtain a freshness assessment value; Based on the freshness assessment value and the abnormal correlation information, monitoring and analysis are performed to obtain remote monitoring information.

[0011] Preferably, the step of abnormally correlating the first detection result and the second detection result to obtain abnormal correlation information includes: The first type of index analysis is performed on the first detection result to obtain the first set of analytical data; The second detection result is analyzed using a second type of index to obtain a second set of analytical data; The first set of parsed data and the second set of parsed data are cross-compared to generate a comparison result set; The comparison result set is classified into event levels according to preset rules, and the abnormal association information is output.

[0012] The present invention also provides a remote monitoring system for food freshness, applied to the remote monitoring method for food freshness described in any one of the above claims, comprising: The data acquisition module is used to acquire processing technology data and distribution and transportation data of fresh food, calculate the shelf life, and obtain the shelf life of the food. The analysis module is used to perform process detection and anomaly identification on the processing technology data to obtain a first detection result; The association module is used to detect the cumulative temperature value and transportation status of the circulation and transportation data, and integrate the results to obtain a second detection result; The processing module is used to respond and adjust based on the food's shelf life, the first test result, and the second test result to obtain remote control information.

[0013] The remote monitoring method and system for food freshness provided by this invention have the following beneficial effects: By integrating and analyzing processing technology data and distribution and transportation data across the entire supply chain, multi-dimensional collaborative monitoring of food freshness is achieved, effectively solving the assessment bias problem caused by isolated data in existing technologies. Through process detection and anomaly identification combined with temperature cumulative effect analysis, food shelf life can be calculated more accurately, providing a more reliable data foundation for freshness control. By mapping processing anomaly data with distribution and transportation status and dynamically assigning weights based on spoilage sensitivity characteristics, refined management of the risk transmission mechanism across links is achieved, facilitating on-demand regulation and timely intervention to prevent further quality deterioration. Through comprehensive analysis of shelf life data, process anomaly indicators, and transportation status parameters, a more scientific dynamic control strategy is formulated, effectively maintaining food freshness and reducing loss rates and food safety risks. Attached Figure Description

[0014] Figure 1 This is a flowchart of a remote monitoring method for food freshness provided by the present invention; Figure 2 This is a structural diagram of a remote monitoring system for food freshness provided by the present invention.

[0015] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0017] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments.

[0018] Reference Figure 1 As shown, the present invention provides a method for remote monitoring of food freshness, comprising: Step S10: Obtain processing technology data and distribution and transportation data of fresh food, calculate the shelf life, and obtain the food shelf life; Step S20: Perform process inspection and anomaly identification on the processing technology data to obtain the first inspection result; Step S30: Detect the cumulative temperature value and transportation status of the circulation and transportation data, and integrate the results to obtain the second detection result; Step S40: Based on the food's shelf life, the first test result, and the second test result, respond and adjust to obtain remote control information.

[0019] Based on the above steps, the detailed process is as follows: Step S10: Processing data includes, but is not limited to, key process parameters such as cleaning intensity, sterilization temperature, cutting precision, and modified atmosphere packaging ratio. These parameters are collected in real time through a high-precision sensor network deployed on the production line. Distribution and transportation data encompasses multi-dimensional information such as temperature and humidity records inside refrigerated vehicles, vibration frequency, number of door openings, geographical location trajectory, and transportation duration, and is continuously uploaded at minute-level frequencies via IoT devices. After data acquisition, time-series alignment technology is used to standardize data from different sources according to a unified time benchmark, eliminating data asynchrony issues caused by differences in device sampling frequencies.

[0020] The shelf-life calculation module dynamically calculates the theoretical shelf life of food under current conditions based on integrated multi-dimensional data and combined with the microbial growth kinetics model and chemical deterioration rate model of this type of food. The calculation process comprehensively considers the initial contamination level during processing, the historical cumulative effect of temperature during distribution, and the protective performance of packaging conditions, outputting a predicted shelf-life value with probability distribution characteristics. This predicted value is dynamically adjusted as real-time data is updated.

[0021] Step S20: The process detection module first extracts features from the raw data, including the distribution characteristics of process parameters such as mean, variance, and extreme values, as well as the periodicity and trend characteristics of parameter fluctuations. Anomaly identification employs a dual detection mechanism combining rule-based and machine learning approaches: the rule engine sets threshold ranges for process parameters according to food safety standards, triggering over-limit alarms in real time; the machine learning model is trained on historical normal data to obtain the dynamic change patterns of process parameters and identify hidden anomalies deviating from normal patterns. The detection process pays particular attention to the correlation between anomalies, such as the combined effect of insufficient sterilization temperature and subsequent packaging sealing defects. The first detection result includes structured information such as anomaly type identifiers (e.g., microbial contamination risk, physical damage, excessive chemical residues), the time point of anomaly occurrence, the duration of anomaly, and an anomaly severity score.

[0022] Step S30: Temperature accumulation detection employs a time-temperature integral algorithm to calculate the cumulative impact of temperature fluctuations experienced by food throughout its distribution process on its freshness, with differentiated attenuation coefficients configured for different temperature zones (frozen, refrigerated, and ambient). Transportation condition detection assesses the risk of mechanical damage and temperature stability levels by analyzing vibration frequency spectra, acceleration variation curves, and the spatiotemporal distribution of door opening events.

[0023] The results integration module establishes a correlation model between the cumulative effect of temperature and transportation state parameters, such as the synergistic destructive effect of temperature fluctuations on food texture under high-frequency vibration environments. The second detection result output is a standardized risk assessment index, including quantitative parameters such as the percentage of temperature violation duration, cumulative heat load value, mechanical damage index, and transportation environment stability score. These parameters complement the shelf-life prediction in step S10, jointly constructing a comprehensive view of the freshness degradation of food during the distribution process.

[0024] Step S40: The control engine constructs a three-layer analysis architecture: the first layer processes the remaining safety margin between the food's shelf life and the current time; the second layer integrates the coupling effect of processing anomalies (first detection result) and distribution risks (second detection result); and the third layer combines product value and supply chain constraints to generate optimization strategies. Response control employs a dynamic decision tree model, matching a pre-set control rule library based on the combination pattern of input parameters. For example, when insufficient sterilization and excessive cold chain temperature are detected, it automatically triggers downgraded sales or accelerated distribution instructions.

[0025] The remote control information includes structured control commands such as specific execution instructions (e.g., adjusting storage temperature, changing delivery routes, updating sales priorities), expected execution time, and suggestions for adjusting monitoring indicator thresholds. This step forms a closed-loop control with the previous steps, and the control effect is verified and optimized through a new round of data collection (returning to step S10), achieving continuous improvement in food freshness management.

[0026] This invention provides a remote monitoring method for food freshness. By integrating and analyzing processing technology data and distribution and transportation data across the entire supply chain, it achieves multi-dimensional collaborative monitoring of food freshness, effectively solving the evaluation bias problem caused by isolated data in existing technologies. Through process detection and anomaly identification combined with temperature cumulative effect analysis, the shelf life of food can be calculated more accurately, providing a more reliable data foundation for freshness control. By mapping processing anomaly data with distribution and transportation status and dynamically assigning weights based on spoilage sensitivity characteristics, it achieves refined management of the risk transmission mechanism across links, facilitating on-demand regulation and timely intervention to prevent further quality deterioration. By comprehensively analyzing shelf life data, process anomaly indicators, and transportation status parameters, a more scientific dynamic control strategy is formulated, thereby effectively maintaining food freshness and reducing loss rates and food safety risks.

[0027] In one embodiment, processing technology data and distribution and transportation data of fresh food are obtained, and shelf life is calculated to obtain the food's shelf life, including: Processing data is derived from records kept during the production of fresh food, including detailed parameters for cleaning operations and packaging. Cleaning intensity data pertains to indicators of the cleanliness of the production environment or food surfaces, such as cleaning solution concentration, cleaning duration, disinfection frequency, and microbial test results. This data is acquired through sensors on the production line or quality control logs and undergoes standardization to eliminate unit variations and outliers.

[0028] Packaging barrier data describes the barrier performance of packaging materials against external factors such as gases, moisture, or light. Common parameters include oxygen permeability, water vapor transport rate, and material thickness. Data extraction is performed through database queries or file parsing, where cleanliness strength data and packaging barrier data are separated and converted into numerical formats for easy subsequent calculations. Quality degradation rate calculations are based on this extracted data, referencing spoilage models in food science to correlate cleanliness strength with microbial growth and packaging barrier with oxidation or hydrolysis rates.

[0029] The calculation process involves weighted averaging or regression analysis, but does not rely on complex algorithms. Instead, it follows predefined rules or lookup tables; for example, high cleanliness corresponds to a lower microbial growth rate, and high-barrier packaging corresponds to slower chemical deterioration. The resulting baseline food spoilage rate represents the rate at which food quality declines under ideal environmental conditions, expressed as a percentage of spoilage per day. This rate serves as the basis for subsequent dynamic fitting, and its accuracy depends on the completeness and consistency of the processing data.

[0030] The distribution and transportation data comes from monitoring records of fresh food in the logistics chain, including temperature sensor readings, geographic location information, timestamps, and possibly additional environmental parameters. The temperature extraction process first identifies temperature-related fields in the data stream, such as parsing temperature values ​​from IoT devices or transportation management systems.

[0031] The data cleaning step removes invalid or outlier readings, such as temperature values ​​outside the reasonable range, and processes missing data using interpolation or smoothing techniques to ensure the continuity of the temperature series. Environmental classification, based on the extracted temperature data and other contextual information, such as the type of transport vehicle or route characteristics, categorizes the transport environment into predefined categories, such as refrigerated, frozen, ambient temperature, or high temperature and humidity.

[0032] The classification criteria are based on industry standards or food preservation requirements. For example, temperatures below a certain threshold are identified as refrigerated environments, while fluctuations and durations are considered to enhance robustness. Environment type identifiers are output in coded or labeled form, such as numeric codes or strings, for easy subsequent processing. Transport temperature data is organized into a structured format, including average temperature, extreme values, and trends, while environment type identifiers provide qualitative context.

[0033] Transportation temperature data includes temperature values ​​recorded continuously or intermittently during the logistics process, while environmental type identifiers provide a qualitative description of transportation conditions. Temperature impact analysis aims to quantify the accelerating or slowing effect of temperature fluctuations on the food spoilage process. Temperature data is first segmented, divided into uniform intervals based on changes in time series or environmental identifiers, such as by transportation stage or consistency of environmental type.

[0034] Representative temperature values, such as average temperature or weighted temperature, are calculated for each interval to capture the overall level of heat exposure. Environmental type identifiers are used to adjust the analysis strategy; different environmental categories correspond to different impact models. For example, refrigerated environments may emphasize temperature stability, while ambient temperature environments focus on the cumulative effects of high-temperature events.

[0035] The analysis process follows the principles of the Arrhenius equation, but avoids direct use of mathematical formulas, instead employing a rule-based or experience-based lookup table approach. Predefined rules compare temperature values ​​with standard reference temperatures to calculate the magnitude of changes in the relative rate of deterioration. Environmental indicators introduce correction factors, such as the potential for increased microbial risk in high-humidity environments.

[0036] The temperature effect coefficient is ultimately output as a dimensionless value, representing the rate of acceleration of spoilage under actual transportation conditions relative to baseline conditions. This coefficient works in conjunction with the baseline food spoilage rate to provide a key input for dynamic fitting, ensuring that the predictive model reflects the temperature variability and environmental complexity of the real logistics chain.

[0037] The baseline spoilage rate of food represents the rate of quality decline under ideal conditions, while the temperature effect coefficient quantifies the environmental intensification effects during distribution. Dynamic fitting calculations integrate these two parameters to generate shelf-life estimates for specific food batches and logistics routes.

[0038] A baseline spoilage rate is used as the initial spoilage rate value, and a temperature influence coefficient is applied as a multiplicative factor to this rate to obtain the actual average spoilage rate. The calculation process is based on the concept of time integration, but does not involve complex calculus; instead, it uses discrete-time step simulation or a pre-calculated rate adjustment table. For example, the expected total transportation time is divided into small time segments, and a corresponding temperature influence coefficient is applied to each segment (if the coefficient changes over time). The quality degradation of each segment is accumulated until a preset spoilage threshold is reached. The spoilage threshold is predefined based on the food type and safety standards, such as sensory acceptance limits or microbiological limits. The final output of the food shelf life is in time units, such as the remaining edible days or a specific expiration date. This result dynamically responds to changes in processing and distribution data, significantly improving prediction accuracy and reliability compared to fixed shelf-life labels.

[0039] This embodiment achieves multi-dimensional and accurate prediction of food shelf life by extracting cleaning intensity and packaging barrier data from the processing technology and combining them with temperature and environmental information during distribution and transportation. This method comprehensively considers key factors in both production and logistics, significantly improving the scientific rigor and reliability of shelf-life calculations. Utilizing dynamic fitting calculations, it can reflect real-time temperature fluctuations and changes in transportation conditions, making the prediction results more closely match actual quality degradation and effectively avoiding the conservative or risky estimations caused by traditional fixed shelf-life methods.

[0040] In one embodiment, sterilization intensity data and packaging barrier data are extracted from the processing data to calculate the spoilage rate of fresh food, including: Cleaning intensity data is derived from hygiene operation parameters recorded during food processing, including but not limited to cleaning agent concentration, contact time, mechanical action intensity, and microbial disinfection indicators. Cleaning effectiveness assessment aims to transform this raw data into quantitative indicators reflecting the ability of hygiene conditions to inhibit the initial microbial load on food.

[0041] Cleaning intensity data is normalized to eliminate dimensional differences between parameters; for example, concentration and time values ​​are converted into uniform dimensionless fractions using a predefined scale. The assessment process references mapping relationships established from industry hygiene standards or historical experimental data, correlating normalized data with microbial residue rates or contamination risk levels. For instance, higher detergent concentrations combined with sufficient contact time correspond to lower initial total bacterial counts; this relationship is achieved through lookup tables or empirical models, avoiding complex calculations. The final output of cleaning quality data is a comprehensive score or grade code, characterizing the initial hygienic quality of processed food.

[0042] This data serves as the foundation for integrating quality degradation factors, and its accuracy directly impacts the reliability of subsequent degradation rate calculations. The evaluation process emphasizes data objectivity and traceability; all input parameters are derived from production logs or direct sensor readings to ensure that the evaluation results accurately reflect processing conditions.

[0043] Packaging barrier data describes the barrier performance of packaging materials against external gases (such as oxygen) and water vapor. Common parameters include oxygen transmission rate (OTR), water vapor transmission rate (WVTR), and material thickness and composition information. The calculation of barrier property coefficients aims to convert these physical parameters into dimensionless correction factors that reflect the degree of influence of packaging on food spoilage.

[0044] The gas barrier coefficient is calculated based on the permeability of gases such as oxygen. It generates a relative barrier efficiency index by comparing the measured oxygen permeability value with a reference standard value (such as the oxygen permeability of high-standard barrier materials). The lower the value, the worse the barrier effect. The humidity barrier coefficient is calculated similarly based on water vapor permeability, reflecting the packaging's ability to control internal humidity. The calculation process uses linear or piecewise linear transformation models; for example, mapping oxygen permeability to a coefficient in the range of 0-1, where 0 represents no barrier and 1 represents complete barrier.

[0045] The coefficient calculations rely on material databases or predefined transformation rules to ensure consistency and repeatability. The obtained gas barrier coefficients and humidity barrier coefficients are used in subsequent fusion calculations to quantify the role of packaging in delaying oxidative deterioration and water activity-related deterioration. The output of this step is structured data, which facilitates collaborative processing with cleanliness quality data, forming a key input for quality degradation modeling.

[0046] Cleanliness quality data characterizes the initial microbial state of food, while gas barrier coefficient and humidity barrier coefficient reflect the packaging's ability to inhibit oxidation and moisture-induced spoilage, respectively. The fusion calculation aims to harmonize these heterogeneous parameters to form a unified composite index characterizing the overall quality degradation tendency.

[0047] Each input parameter is standardized to ensure it falls within the same numerical range; for example, it is converted to a scale value between 0 and 1, where lower values ​​represent lower spoilage risk. The fusion mechanism is based on predefined weighting rules, with weights determined according to food type and dominant spoilage mode; for example, gas barrier coefficients have a higher weight for easily oxidized foods, while humidity barrier coefficients have a higher weight for moisture-sensitive foods. Cleanliness quality data serves as a foundational factor, and its value directly affects the baseline level of the fusion results.

[0048] The calculations are performed using a weighted summation or multiplication model, but without involving complex algorithms. Instead, parameter integration is achieved through empirical rules or lookup tables. The quality degradation coefficient is output as a dimensionless value that comprehensively reflects the combined effects of processing hygiene conditions and packaging performance on food shelf life. A higher coefficient value indicates a stronger overall driving force for quality degradation. This coefficient provides the core input for subsequent simulations, ensuring that the degradation model simultaneously covers both the initial quality state and external barrier effects.

[0049] The quality degradation coefficient is used as a key driving variable to simulate the dynamic process of food quality degradation over time under standard reference conditions. The simulation process employs a time-stepping method, dividing the total time into discrete intervals. Within each interval, the quality degradation coefficient is combined with a predefined baseline degradation constant to calculate the amount of quality decline in that interval.

[0050] The baseline spoilage constant is preset according to the food category and represents the inherent spoilage tendency of that type of food without any protection. The simulation continues, accumulating the quality decline until a set quality failure threshold is reached, which is determined based on sensory, microbiological, or chemical safety standards. The ratio of the total time of the entire simulation cycle to the initial quality value is the food baseline spoilage rate, typically expressed as a percentage of quality loss per day or a similar measure.

[0051] This rate represents the theoretical spoilage rate under a controlled baseline environment (such as constant low temperature and specific humidity), excluding dynamic interference factors in actual distribution. The calculation process emphasizes the transparency and repeatability of the simulation, with all parameters derived from objective data or recognized standards. The final food baseline spoilage rate serves as the core output, providing an indispensable foundational value for subsequent dynamic fitting of the temperature influence coefficient under actual transportation conditions.

[0052] This embodiment achieves refined modeling of the food quality degradation process by integrating multi-source data such as cleaning intensity and packaging barrier properties. Based on cleaning effectiveness assessment and barrier property coefficient calculation, the impact of processing and packaging on the spoilage rate is accurately quantified, significantly improving the scientific rigor of the baseline spoilage rate calculation. A quality degradation coefficient is generated through a multi-factor fusion mechanism, effectively coordinating the synergistic effects of hygiene conditions and packaging performance, enabling the model to more comprehensively reflect actual degradation drivers. A dynamic simulation method is used to calculate the baseline spoilage rate, providing a reliable basis for subsequently introducing distribution environment variables, ultimately resulting in a more realistic shelf-life prediction, effectively reducing the risk of food quality misjudgment and improving the accuracy of supply chain management.

[0053] In one embodiment, process inspection and anomaly identification are performed on the processing technology data to obtain a first inspection result, including: Processing of process data begins with the acquisition and integration of multi-source heterogeneous data. Sensors deployed at various stages of the production line capture raw data in real time, including key physicochemical parameters such as temperature, pressure, flow rate, and pH value, as well as equipment operating status and operation timestamps. The process parameter extraction module identifies and separates core parameters directly related to food quality from the raw data stream, such as the temperature maintenance value during sterilization, the water pressure fluctuation range during washing, and the gas replacement rate during packaging.

[0054] The extraction process employs feature engineering methods, using a sliding window to statistically calculate the mean, variance, extreme values, and trends of parameters to generate high-dimensional feature vectors. For time alignment processing, a unified time coordinate axis is established for device data with different sampling frequencies, and interpolation algorithms are used to reconstruct asynchronous data sequences into synchronous time series. For event-based data (such as valve opening actions), it is converted into state interval data with duration.

[0055] The processed process parameter sequence forms a standardized time-parameter matrix, with each time point corresponding to a complete set of process state descriptors, providing structured input for subsequent anomaly detection.

[0056] Anomaly detection is based on multi-modal analysis of process parameter sequences. The detection process employs a hybrid detection mechanism combining rule-based and machine learning approaches. The rule engine presets reasonable threshold ranges for various parameters and identifies numerical deviations exceeding the allowable ranges in real time.

[0057] A dynamic correlation model between parameters is established by training on historical normal data to detect abnormal patterns that violate normal correlations, such as unexpected decoupling of temperature and pressure values. Anomaly detection also focuses on temporal characteristics, identifying dynamic features such as the frequency, duration, and gradient of abnormal fluctuations in parameter values.

[0058] The process stage matching module divides the detected time-series data into continuous process stages, such as pre-cleaning, main sterilization, rapid cooling, and vacuum packaging, based on a preset process flow diagram and time logic. Each stage corresponds to a specific normal range of parameters and association rules. The generation of process anomaly indicators comprehensively considers the importance weight of the anomaly parameters, the proportion of the anomaly duration to the total duration of the stage, and the potential transmission effect of the anomaly on subsequent process stages, outputting a quantitative anomaly severity score and anomaly type identifier.

[0059] Process execution anomaly analysis is based on process anomaly indicators to conduct in-depth diagnosis and attribution analysis. The analysis process establishes a multi-level diagnostic framework: the first level identifies local anomalies of single parameters, the second level detects complex anomalies caused by the synergistic effect of multiple parameters, and the third level assesses the propagation effect of anomalies over time.

[0060] Anomaly diagnosis employs a combination of case-based reasoning and causal graph models. It matches current anomaly patterns with a historical anomaly case database based on similarity, and infers the root cause and potential consequences of the anomaly through causal networks. Process judgment results prioritize food safety, assigning the highest priority to anomalies that may affect microbial safety or chemical contamination. The construction of the anomaly data point set not only includes the original anomaly data points but also expands to include relevant data from the time periods preceding and following the anomaly, forming an anomaly data cluster with contextual information.

[0061] Each set of anomalous data points is labeled with structured information such as anomaly type code, occurrence time range, affected process steps, and possible cause analysis. The process judgment result output is a combination of qualitative conclusions and quantitative scores, including process compliance status determination, anomaly impact scope assessment, and recommended handling measures level.

[0062] Anomaly classification and severity assessment are conducted through a multi-dimensional comprehensive analysis. The classification model establishes anomaly level classification standards based on the spatiotemporal distribution characteristics of the abnormal data point set, the quantitative intensity of process anomaly indicators, and the safety risk assessment of process judgment results.

[0063] The severity classification considers three core dimensions: the immediate impact of the anomaly on food safety, the potential damage to product quality, and the disruption to the stability of the production process. The severity assessment employs a multi-factor weighted scoring method, assigning different weight coefficients based on the anomaly type; for example, microbiological anomalies have a higher weight than equipment parameter anomalies.

[0064] The evaluation process incorporates a dynamic adjustment mechanism, increasing the severity rating for consecutive occurrences of similar anomalies or concurrent occurrences of multiple anomalies. The initial detection results are generated using a standardized output format, including anomaly level code, severity score, description of impact scope, emergency handling priority recommendations, and a detailed anomaly analysis report index. These results not only serve for immediate quality control decisions but also act as historical data input into the process optimization system, providing data support for continuous improvement of production processes.

[0065] This embodiment achieves accurate identification and quantitative assessment of quality risks in the production process by performing multi-level anomaly detection and process stage matching on processing technology data. Through time alignment processing of process parameters and anomaly pattern analysis, cross-parameter collaborative anomalies and time-series anomalies can be effectively captured, significantly improving the coverage and accuracy of anomaly detection. Severity assessment and classification based on multi-dimensional anomaly indicators provide a scientific basis for quality control decisions and support the formulation of differentiated handling strategies. By dynamically linking anomaly data with process stage characteristics, precise location and source tracing of anomaly impacts are achieved, providing data support for production process optimization.

[0066] In one embodiment, anomaly level classification and severity assessment are performed using abnormal data point sets, process anomaly indicators, and process judgment results to obtain a first detection result, including: The abnormal pattern recognition processing of food processing based on abnormal data point sets is achieved through multi-dimensional feature analysis and classification models. The process first extracts key feature vectors from the abnormal data point set, including the deviation of abnormal values, duration of occurrence, spatial distribution range, and correlation strength with other parameters. Feature extraction employs a combination of time and frequency domain methods to capture the fluctuation patterns and periodic characteristics of abnormal signals.

[0067] The pattern recognition module, based on a pre-established anomaly pattern classification system, maps extracted feature vectors to specific anomaly type categories. The classification system covers three major categories: microbial contamination anomalies, physical contamination anomalies, and chemical contamination anomalies. Each major category is further subdivided into specific anomaly patterns. For example, microbial contamination anomalies include subcategories such as insufficient sterilization and cross-contamination; physical contamination anomalies include subcategories such as foreign matter contamination and mechanical damage; and chemical contamination anomalies involve subcategories such as cleaning agent residue and lubricant leakage.

[0068] The identification process employs a multi-classifier ensemble strategy, combining machine learning algorithms such as decision trees and support vector machines, and determines the final anomaly type through a voting mechanism. The output of processed anomaly types uses a standardized encoding format, with each type corresponding to a specific processing priority and impact assessment rule.

[0069] The process of analyzing processing contamination establishes a mapping model between anomaly types and process anomaly indicators. The analysis module employs differentiated analysis strategies and calculation rules for different types of processing anomalies. For microbial contamination anomalies, the focus is on analyzing the duration and degree of deviation of temperature anomaly indicators, and calculating the potential growth rate and order-of-magnitude changes of microorganisms.

[0070] For physical contamination anomalies, the analysis primarily focuses on indicators such as abnormal equipment vibration and abnormal metal detection signals to assess the probability and possible size range of foreign matter contamination. For chemical contamination anomalies, the analysis emphasizes indicators such as cleaning agent concentration and pH fluctuations to calculate the possible concentration and distribution of chemical residues. The generation of contamination impact data includes three core dimensions: contamination concentration level, contamination duration, and contamination diffusion range.

[0071] Concentration levels are calculated based on the deviation of abnormal indicator values ​​from standard limits. The duration is directly taken from the abnormal time span data, and the diffusion range is inferred based on the location of the equipment where the abnormality occurred and the product flow direction. Pollution impact data is normalized to output standardized impact scores, facilitating comparison and integration between different anomaly types.

[0072] The safety assessment process establishes a multi-dimensional risk assessment matrix, which deeply integrates and analyzes pollution impact data with process judgment results. The assessment matrix uses pollution concentration level, duration, and diffusion range as basic dimensions, with each dimension divided into several risk levels and assigned corresponding weight coefficients.

[0073] The assessment process employs a forward reasoning mechanism, automatically matching corresponding clauses from a pre-defined risk assessment rule base based on the quantitative indicators of pollution impact data. The rule base is established according to food safety standards and industry norms, covering different risk assessment standards for microbial, physical, and chemical contamination. Safety levels are output in a tiered format, typically divided into four main levels: emergency risk, high risk, medium risk, and low risk.

[0074] Each level corresponds to a clear description of the risk characteristics and the required timeframe for handling them. For example, an emergency risk level indicates an immediate food safety hazard, requiring the immediate initiation of product isolation and recall procedures. The safety assessment module also provides risk uncertainty analysis; when input data is missing or conflicting, it outputs corresponding confidence indicators as supplementary references for level determination.

[0075] The food processing judgment process implements rule-based comprehensive decision analysis. The judgment engine establishes a two-dimensional decision matrix, with safety level as the vertical axis and processing anomaly type as the horizontal axis. Each intersection of the matrix is ​​pre-set with corresponding judgment conclusions and handling suggestions.

[0076] The decision-making rules follow the principle of prioritizing food safety, and high-level safety risks trigger the highest priority response measures regardless of the type of anomaly. For medium- and low-risk situations, the judgment process further considers the characteristics of the anomaly type; for example, physical foreign object contamination and microbial contamination may have different response strategies even if they have the same safety level.

[0077] The judgment output includes information from multiple dimensions: overall compliance status (pass / fail / pending), specific anomaly description, impact assessment, recommended remedial measures, and subsequent monitoring requirements. The first test result is output using a standardized structure.

[0078] This embodiment achieves precise quantitative assessment of quality and safety risks in food production by performing multi-level anomaly pattern recognition and contamination impact analysis on processing technology data. Pattern recognition based on anomaly data point sets can accurately distinguish different anomaly types such as microbial contamination, physical contamination, and chemical contamination, providing a basis for targeted treatment. Through the fusion analysis of contamination impact data and process judgment results, a scientific safety level assessment system is established, significantly improving the accuracy and reliability of risk assessment. The comprehensive judgment mechanism combining safety level and anomaly type can output differentiated disposal strategies, ensuring the bottom line of food safety while avoiding resource waste caused by over-processing.

[0079] In one embodiment, the accumulated temperature value and transportation status of the circulation and transportation data are detected, and the results are integrated to obtain a second detection result, including: The time-segmented processing of logistics and transportation data is based on the phased changes in environmental conditions and operational status during transportation. The processing employs an adaptive time window algorithm, which automatically identifies key time nodes in the transportation process based on temperature fluctuation frequency, equipment operating status change records, and geographical location change information.

[0080] The determination of time points comprehensively considers changes in the external environment, such as alternating day and night temperature differences and regional climate changes, as well as internal operational events, such as loading and unloading, transit stops, and switching of transportation methods. The segmented algorithm prioritizes setting segment boundaries at locations where environmental parameters undergo significant transitions, ensuring that relatively stable temperature and environmental conditions are maintained within each time period.

[0081] For long-distance transportation, segmented processing further considers the division of transportation stages, such as the characteristics of different stages like the initial segment, trunk transportation segment, and last-mile delivery segment, and adopts differentiated time window strategies. The processed data for each transportation period includes complete time-series records, environmental parameter sets, and associated metadata information, providing a structured data foundation for subsequent refined analysis.

[0082] Temperature sampling involves standardizing the temperature time series for each transportation period. The sampling process employs an equal-interval resampling method to transform potentially non-uniform raw temperature data into a regularized sequence with a unified time base.

[0083] For the raw data acquired at high frequencies, a sliding window smoothing filter technique is applied to eliminate instantaneous noise interference while preserving important temperature change trends. A weighted average algorithm is used for mean calculation, assigning differentiated weights based on the non-linear characteristics of the impact of temperature values ​​at different time points on food quality. High-temperature periods are given higher weighting coefficients than low-temperature periods, reflecting the accelerating effect of temperature on the spoilage rate.

[0084] The calculation process pays special attention to the amplitude of temperature fluctuations. For periods with large temperature oscillations, extreme temperature values ​​and their durations are recorded in addition to the average value calculation. The average temperature value for each transportation period is output in a standardized format, along with temperature stability indicators, including auxiliary parameters such as temperature variance and maximum fluctuation range, providing comprehensive input data for calculating the cumulative temperature effect.

[0085] The calculation of the cumulative effect of temperature employs the time-temperature integral principle to quantify the cumulative impact on food quality under different temperature conditions. The calculation process establishes a nonlinear relationship model between temperature and the rate of quality decay. The average temperature value for each transportation period is first converted into a relative spoilage rate coefficient, which represents the ratio of the food spoilage rate at the current temperature to that at the reference temperature.

[0086] The conversion process employs a piecewise linear approximation method, applying different conversion coefficients across different temperature ranges to accurately reflect the nonlinear influence of temperature on microbial growth and chemical reactions. The time-temperature integral calculation multiplies the relative spoilage rate coefficient for each time period by the corresponding duration to obtain the equivalent temperature influence for that period.

[0087] The cumulative calculation sums the temperature impact equivalents across all transportation periods and incorporates a temperature sequence effect correction factor to account for the additional impact of temperature fluctuations on quality compared to constant temperatures. The cumulative temperature value is output in standardized spoilage equivalent units, which comprehensively reflects the overall impact of temperature conditions on food freshness throughout the distribution process, providing a key input parameter for subsequent quality assessment.

[0088] Vibration frequency analysis extracts time-frequency domain features from raw vibration data acquired by a triaxial accelerometer. The analysis process uses a Fast Fourier Transform (FFT) to convert the time-domain vibration signal into a frequency-domain energy distribution spectrum, identifying vibration energy intensity within characteristic frequency ranges. Special attention is paid to frequency bands matching the resonant frequency of the transport vehicle, as well as specific frequency intervals that may cause mechanical damage to food processing equipment.

[0089] Impact event detection is based on an acceleration rate of change threshold. A dynamically adjustable impact detection threshold is set, and the sensitivity is adaptively adjusted according to the characteristics of different transportation stages. The detection algorithm distinguishes between continuous vibration and discrete impact events, extracting multi-dimensional features such as duration, peak acceleration, and impact direction for impact signals exceeding the threshold.

[0090] The generation of transportation status data includes comprehensive indicators such as vibration intensity index, impact event statistical distribution, and transportation stability score. The vibration intensity index quantifies the overall vibration energy level, the impact event statistics record the frequency and intensity distribution of major impacts, and the transportation stability score comprehensively assesses the mechanical environmental stability of the transportation process. These indicators together constitute a multi-dimensional descriptive system for transportation status data.

[0091] The state fusion detection employs multi-source information fusion technology to establish a correlation model between cumulative temperature values ​​and transportation status data. The fusion process is based on quality damage mechanism analysis, quantifying the synergistic impact of temperature and mechanical effects on food quality.

[0092] Different fusion weight coefficients are configured according to the characteristics of different food types. For example, fragile foods are given a higher weight for mechanical effects, while temperature-sensitive foods are emphasized for temperature accumulation effects. The fusion algorithm uses a fuzzy logic reasoning mechanism to convert temperature accumulation values ​​into quality degradation indicators, and transportation status data into mechanical damage risk indicators. The two indicators are comprehensively reasoned through a pre-established rule base to output a comprehensive quality assessment result.

[0093] The second test result includes the current quality status rating, remaining shelf life correction value, risk warning level, and recommended handling measures. The quality status rating uses a five-level classification system, indicating the level of food freshness from best to worst. The remaining shelf life correction value dynamically adjusts the theoretical shelf life based on the temperature and mechanical conditions during actual distribution. The risk warning level indicates quality issues requiring immediate attention, and the recommended handling measures provide specific operational guidance, forming a complete distribution quality test conclusion.

[0094] This embodiment, through time-segmented processing and temperature sampling analysis of distribution and transportation data, can accurately calculate the cumulative temperature effect and precisely reflect the degree of influence of temperature conditions on food quality during distribution. Vibration frequency analysis and impact event detection can comprehensively assess the mechanical stability of the transportation environment and identify risk factors that may cause physical damage to food. The fusion detection based on cumulative temperature values ​​and transportation status data achieves a comprehensive assessment of the impact on quality during distribution, overcoming the limitations of single-factor assessments. This method can dynamically adjust quality assessment parameters to adapt to changes in different transportation conditions and food characteristics, improving the accuracy of assessment results. Through collaborative analysis of multi-dimensional data, it provides a scientific basis for quality control in the food distribution process, helping to reduce transportation losses and ensure food quality.

[0095] In one embodiment, response control is performed based on the food's shelf life, the first test result, and the second test result to obtain remote control information, including: Time sensitivity analysis is conducted to deeply analyze the dynamic changes in food shelf life. The process establishes a shelf-life decay model to analyze the differences in the rate of change in food quality across different time stages. The analysis identifies key time points within the shelf life, including the boundary points between the quality stabilization period, the accelerated spoilage period, and the rapid deterioration period.

[0096] The analysis module calculates the duration percentage and degradation rate gradient of each stage, highlighting the start time and duration of the accelerated degradation period. The determination of degradation-sensitive times uses a relative time identification method, calculating the remaining time until the next quality stage transition point based on the current time.

[0097] The analysis process also considers the characteristics of different food types, employing differentiated criteria for determining the sensitive period for different food types. For example, the sensitive period for short-shelf-life foods appears earlier and lasts for a shorter duration. The output of the spoilage sensitivity time includes multiple time parameters: the start time of the most recent sensitive period, the expected duration of the sensitive period, and the time of the quality inflection point. These parameters provide a time dimension for subsequent process weight allocation.

[0098] Anomaly correlation analysis involves in-depth mining and pattern recognition of anomaly data across different stages. The process first involves standardizing and preprocessing the processing anomaly data from the first detection result and the circulation anomaly data from the second detection result, unifying the data format and quantification standards.

[0099] The correlation analysis employs a combination of rule-based and machine learning approaches to establish a causal relationship model between processing anomalies and circulation anomalies. The analysis process identifies the temporal patterns of anomalous events, detects spatial propagation paths, and uncovers synergistic effects between different types of anomalies.

[0100] Association rule mining focuses on the correlation strength between microbial contamination anomalies in the processing stage and temperature anomalies in the distribution stage, as well as the correspondence between physical damage anomalies and transportation vibration anomalies. The generation of anomaly association information includes multiple dimensions of output, such as a set of association rules, association strength indicators, and influence transmission paths.

[0101] The association rule set describes the causal relationship patterns between different anomaly types, the association strength index quantifies the reliability of this relationship, and the influence transmission path shows the propagation trajectory of anomalies from processing to distribution. This information provides cross-stage anomaly association evidence for subsequent freshness assessment and monitoring decisions.

[0102] The process weight allocation process establishes a dynamic correlation model between time sensitivity and process anomalies. The allocation mechanism determines the importance level of different process anomalies at the current time point based on the spoilage sensitivity time. For foods that are in or near the spoilage sensitivity time window, the weight coefficients of certain key process anomalies are increased accordingly. The weight allocation uses a multi-dimensional decision matrix, where the row dimensions correspond to different spoilage sensitivity stages, and the column dimensions correspond to various types of process anomalies.

[0103] Each matrix element has a pre-set basic weight value, which is then dynamically adjusted based on the specific characteristics of the anomalies in the first detection result. Adjustment factors include parameters such as the severity of the anomaly, its duration, and the scope of its impact. The process influence coefficient is calculated using a weighted summation algorithm, which multiplies each anomaly indicator in the first detection result by its corresponding weight coefficient, sums the results, and then normalizes them to obtain a standardized coefficient value.

[0104] Freshness assessment employs multi-source data fusion calculations, collaboratively analyzing the process influence coefficient with data from the distribution process in the second test results. The assessment process establishes a quality degradation model, with inputs including multiple dimensions of parameters such as the process influence coefficient, cumulative temperature, and transportation status indicators.

[0105] The model calculations employ a piecewise function, using different calculation rules for different quality stages. For the quality stabilization period, the linear superposition of the process influence coefficient and the cumulative effect of temperature is primarily considered; for the accelerated spoilage period, a nonlinear amplification factor is introduced to amplify the synergistic effect of process defects and deteriorating distribution environment. The freshness assessment value is generated by including two sub-indicators: the current quality status score and the remaining quality retention time.

[0106] The quality status score is expressed on a 100-point scale, comprehensively reflecting the real-time quality level of the food; the remaining quality retention time predicts the duration for which the food will maintain acceptable quality under current conditions. A confidence level index is incorporated into the calculation of the assessment value to reflect the reliability of the assessment results and provide a reference for subsequent monitoring and decision-making.

[0107] The monitoring and analysis process develops differentiated monitoring strategies based on freshness assessment results and anomaly correlation information. The analysis module establishes a tiered response mechanism, automatically triggering corresponding levels of monitoring measures based on the range of the freshness assessment value.

[0108] For excellent quality conditions, standard monitoring frequencies and parameters are used; for critical quality conditions, monitoring frequencies are increased and monitoring indicators are added; for deteriorating quality conditions, emergency response procedures are initiated. Anomaly correlation information is used to determine the key areas and specific content of monitoring. For example, when a specific anomaly correlation pattern is detected between the processing and distribution stages, the monitoring of relevant indicators is strengthened in a targeted manner.

[0109] The output of remote monitoring information includes a set of instructions covering various aspects such as monitoring parameter settings, monitoring frequency adjustments, early warning threshold revisions, and suggestions for handling measures. These instructions are sent to each monitoring node through standard communication protocols, enabling precise and differentiated monitoring and management of food quality status, forming a complete closed-loop control from assessment to execution.

[0110] This embodiment, through time-sensitivity analysis of food shelf life, accurately identifies spoilage-sensitive time periods, providing a key time window for subsequent control. Anomaly correlation analysis between processing and distribution stages enables collaborative assessment of risk factors across stages, overcoming the limitations of single-stage monitoring. Dynamically allocating process weights based on spoilage-sensitive times makes the process influence coefficient more reflective of the actual risk level at the current time point. The use of a multi-source data fusion-based freshness assessment method significantly improves the accuracy and reliability of quality status judgment. Through comprehensive analysis of anomaly correlation information and freshness assessment values, differentiated monitoring strategies can be formulated, enabling precise remote control and effectively improving the adaptability and scientific basis of food quality monitoring and decision-making.

[0111] In one embodiment, anomaly correlation is performed between the first detection result and the second detection result to obtain anomaly correlation information, including: The first type of indicator analysis involves structured processing of anomaly detection results during the processing stage. The analysis process extracts three core indicators from the initial detection results: anomaly type feature values, anomaly time distribution parameters, and anomaly impact quantification values. The anomaly type feature values ​​encode detected anomalies into three main categories: microbial contamination, physical contamination, and chemical contamination. Each category is further subdivided into specific anomaly patterns and assigned standardized identification codes.

[0112] The analysis of anomaly time distribution parameters reveals the temporal characteristics of anomalies, including the anomaly's onset time, duration, frequency, and temporal distribution patterns. The quantitative value of anomaly impact severity calculates the actual intensity of the anomaly's influence on food quality, employing a multi-dimensional scoring system that includes anomaly severity score, impact range coefficient, and propagation probability index.

[0113] The analysis process employs data standardization techniques to convert raw indicators with different dimensions into a unified scaling range, eliminating analytical biases caused by differences in data scale. The output of the first set of analyzed data adopts a hierarchical data structure: the upper layer contains aggregated information on anomaly types, the middle layer contains temporal distribution characteristics, and the lower layer contains detailed parameters of the degree of influence, forming a complete descriptive system of processing anomaly characteristics.

[0114] The second type of indicator analysis focuses on in-depth feature extraction of detection data from the distribution and transportation process. The analysis process focuses on three main dimensions: temperature cumulative effect characteristics, mechanical environment characteristics, and time-sensitivity characteristics. The temperature cumulative effect characteristic analysis uses the time-temperature integral principle to calculate the equivalent heat load value, while simultaneously extracting detailed parameters such as temperature fluctuation frequency, extreme temperature duration, and temperature change gradient.

[0115] Mechanical environmental characteristics extract key indicators from vibration and shock data, including parameters such as vibration dominant frequency energy distribution, shock event intensity statistics, and transportation stability index, to quantify the risk of mechanical damage during transportation. Timeliness characteristics analyze circulation time parameters, including total transportation time, time allocation for each stage, and time efficiency indicators, to assess the degree of impact of time factors on quality.

[0116] An environmentally adaptive correction algorithm is introduced during the analysis process to weight various indicators according to different food characteristics. For example, the weight of temperature-related indicators is increased for temperature-sensitive foods. The output of the second set of analyzed data adopts a multi-level index structure to establish a correlation mapping relationship between temperature, mechanical properties, and time, providing structured circulation process characteristic data for subsequent data cross-comparison.

[0117] The data cross-comparison process establishes a multi-dimensional correlation analysis model between the processing and distribution stages. The comparison is implemented using time series alignment technology to accurately match the time of processing anomalies with the time of changes in the distribution environment, identifying causal relationships in the time domain.

[0118] The comparison comprises three levels: eigenvalue correlation analysis, time-series pattern matching, and influence strength coupling calculation. Eigenvalue correlation analysis calculates the statistical correlation coefficients of similar indicators across different stages, such as the correlation strength between microbial contamination indicators in the processing stage and temperature fluctuation indicators in the distribution stage. Time-series pattern matching identifies abnormal propagation patterns across stages, detecting whether processing abnormalities trigger chain reactions in the distribution stage.

[0119] The influence strength coupling degree calculation quantifies the synergistic effect between processing anomalies and distribution anomalies, and assesses the combined impact of the two on food quality. The comparison result set is generated using a standardized record format, with each record containing a matching index pair, correlation strength value, time difference parameter, and coupling influence coefficient, forming a complete chain of evidence for cross-stage anomaly associations.

[0120] Event level classification employs intelligent decision analysis based on a pre-established rule base. The rule base contains multi-level classification standards: the first level determines basic association levels based on association strength values; the second level adjusts the levels by incorporating time proximity parameters; and the third level considers the degree of influence coupling for final level determination. The classification process utilizes a fuzzy reasoning mechanism to handle level determination in boundary cases.

[0121] Each rule in the rule base corresponds to a specific anomaly association scenario, such as a combined association rule between insufficient processing sterilization and excessive flow temperature, or a direct association rule between physical damage and transportation vibration. Each rule includes three parts: preconditions, reasoning logic, and conclusion output, forming a complete basis for judgment. The output of anomaly association information adopts a structured data format, including associated event codes, association level identifiers, impact scope assessments, and confidence indices.

[0122] The association event coding adopts standardized naming rules to accurately describe the type characteristics of abnormal associations; the association level identification adopts a five-level classification system, arranged from weak association to strong association; the impact scope assessment predicts the scope of quality impact that the associated anomaly may cause; and the confidence index reflects the reliability of the level determination results.

[0123] The regulation strategy generation module automatically generates targeted processing suggestions based on the level and type characteristics of abnormal correlation information. The strategy generation adopts a combination of case-based reasoning and rule-based reasoning. First, it searches for processing solutions for similar correlation patterns in the historical case library, and then adjusts and optimizes them according to the current specific situation through the rule engine.

[0124] The strategy recommendations encompass multiple dimensions: monitoring frequency adjustment suggestions, detection parameter optimization schemes, priority ranking of handling measures, and prevention and improvement measures. Monitoring frequency adjustment is dynamically set based on the correlation level; strong correlation anomalies receive increased monitoring frequency, while weak correlation anomalies maintain regular monitoring. Detection parameter optimization targets the characteristics of correlated anomalies, increasing the monitoring intensity of relevant indicators.

[0125] The priority of response measures is ranked according to the correlation level and the degree of impact, ensuring that key issues are addressed first. Preventive and improvement measures provide long-term optimization suggestions to reduce the probability of abnormal correlations occurring at the source. The final output of remote monitoring information integrates all strategic recommendations to form a complete quality control plan, providing decision support for food freshness monitoring.

[0126] This embodiment achieves standardized extraction and structured description of cross-stage anomaly features through multi-dimensional index analysis of detection data from processing and distribution stages. Through cross-data comparison and analysis, the inherent correlation between processing anomalies and changes in the distribution environment can be accurately identified, revealing the cross-stage risk transmission mechanism. An event level classification method based on preset rules enables scientific quantitative assessment of anomaly correlation information, providing a reliable basis for quality control decisions. In-depth mining and analysis of anomaly correlation information provides an effective technical means for early warning and prevention of quality risks in the food supply chain.

[0127] Reference Figure 2 The present invention also provides a remote monitoring system for food freshness, applicable to the remote monitoring method for food freshness described in any of the above-mentioned embodiments, comprising: The data acquisition module is used to acquire processing technology data and distribution and transportation data of fresh food, calculate the shelf life, and obtain the food's shelf life. The analysis module is used to perform process inspection and anomaly identification on the processing technology data to obtain the first inspection result; The correlation module is used to detect the cumulative temperature value and transportation status of circulation and transportation data, and integrate the results to obtain a second detection result; The processing module is used to respond and adjust based on the food's shelf life, the first test result, and the second test result to obtain remote control information.

[0128] The remote monitoring system for food freshness provided by this invention achieves multi-dimensional collaborative monitoring of food freshness by integrating and analyzing processing technology data and distribution and transportation data across the entire supply chain. This effectively solves the assessment bias problem caused by isolated data in existing technologies. By combining process detection and anomaly identification with temperature cumulative effect analysis, the system can more accurately calculate the shelf life of food, providing a more reliable data foundation for freshness control. By mapping processing anomaly data with distribution and transportation status and dynamically assigning weights based on spoilage sensitivity characteristics, the system achieves refined management of the risk transmission mechanism across links, facilitating on-demand regulation and timely intervention to prevent further deterioration of quality. By comprehensively analyzing shelf life data, process anomaly indicators, and transportation status parameters, a more scientific dynamic control strategy is formulated, thereby effectively maintaining food freshness and reducing loss rates and food safety risks.

[0129] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the system and each module described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0130] In this embodiment, the processor and memory can be connected via a bus or other means. The memory may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as read-only memory, flash memory, hard disk, or solid-state drive. The processor may be a general-purpose processor, such as a central processing unit, digital signal processor, application-specific integrated circuit, or one or more integrated circuits configured to implement embodiments of the present invention.

[0131] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for remotely monitoring the freshness of food, characterized in that, include: Obtain processing technology data and distribution and transportation data of fresh food, calculate shelf life, and obtain the food's shelf life; The processing data is subjected to process inspection and anomaly identification to obtain a first inspection result; The temperature accumulation value and transportation status of the circulation and transportation data are detected, and the results are integrated to obtain a second detection result; Based on the food's shelf life, the first test result, and the second test result, response control is performed to obtain remote control information.

2. The remote monitoring method for food freshness according to claim 1, characterized in that, The process of obtaining processing technology data and distribution and transportation data for fresh food, and calculating the shelf life to obtain the food's shelf life includes: The cleaning intensity data and packaging barrier data are extracted from the processing technology data, and the quality decay rate is calculated to obtain the food baseline spoilage rate. Temperature extraction and environmental classification are performed on the aforementioned transportation data to obtain transportation temperature data and environmental type identifiers; Based on the transportation temperature data and the environmental type identifier, a temperature impact analysis is performed to obtain the temperature impact coefficient; The shelf life of the food is obtained by dynamically fitting the food's baseline spoilage rate and the temperature influence coefficient.

3. The remote monitoring method for food freshness according to claim 2, characterized in that, The step of extracting cleaning strength data and packaging barrier data from the processing technology data, calculating the quality degradation rate, and obtaining the food baseline spoilage rate includes: The cleaning intensity data is used to evaluate the cleaning effectiveness to obtain cleaning quality data; Calculate the barrier property coefficients of the packaging barrier data to obtain the gas barrier coefficient and humidity barrier coefficient; By combining the cleanliness quality data, the gas barrier coefficient, and the humidity barrier coefficient, a quality degradation factor is calculated to obtain the quality degradation coefficient. The quality decay process is simulated based on the quality decay coefficient, and the quality decay rate is calculated to obtain the food baseline spoilage rate.

4. The remote monitoring method for food freshness according to claim 1, characterized in that, The process inspection and anomaly identification of the processing data to obtain a first detection result includes: The processing data is subjected to process parameter extraction and time alignment processing to obtain a process parameter sequence; Based on the process parameter sequence, state anomaly detection and process stage matching are performed to obtain process anomaly indicators. Based on the aforementioned process anomaly indicators, process execution anomaly analysis is performed to obtain process judgment results and anomaly data point set; The first detection result is obtained by classifying the abnormality level and assessing the severity of the abnormality using the abnormal data point set, the process abnormality index, and the process judgment result.

5. The remote monitoring method for food freshness according to claim 4, characterized in that, The step of classifying and assessing the severity of anomalies by using the abnormal data point set, the process anomaly indicators, and the process judgment results to obtain the first detection result includes: The abnormal data point set is subjected to food processing abnormal pattern recognition processing to obtain the processing abnormality type; Based on the processing anomaly type, the process anomaly indicators are analyzed for processing pollution to obtain pollution impact data; Based on the pollution impact data, a safety assessment is performed on the process determination results to obtain a safety level; The food processing is determined by combining the safety level and the type of processing abnormality to obtain the first detection result.

6. The method for remote monitoring of food freshness according to claim 1, characterized in that, The detection process involves measuring the cumulative temperature and transportation status of the data, integrating the results to obtain a second detection result, including: The transportation data is segmented by time to obtain multiple transportation time periods; Temperature sampling and mean calculation are performed on the data for each of the aforementioned transportation periods to obtain the average temperature value for each transportation period. The temperature cumulative effect is calculated based on each average temperature value and the corresponding transportation period data to obtain the temperature cumulative value; Vibration frequency analysis and impact event detection are performed on the aforementioned transportation data to obtain transportation status data; A second detection result is obtained by performing state fusion detection based on the accumulated temperature value and the transportation status data.

7. The remote monitoring method for food freshness according to claim 1, characterized in that, The step of responding and adjusting based on the food's shelf life, the first test result, and the second test result to obtain remote control information includes: A time sensitivity analysis was performed on the shelf life of the food to obtain the spoilage sensitive time. Anomaly correlation is performed between the first detection result and the second detection result to obtain anomaly correlation information; Based on the deterioration sensitivity time and the first detection result, process weight allocation is performed to obtain the process influence coefficient; The freshness is assessed based on the process influence coefficient and the second test result to obtain a freshness assessment value; Based on the freshness assessment value and the abnormal correlation information, monitoring and analysis are performed to obtain remote monitoring information.

8. The method for remote monitoring of food freshness according to claim 7, characterized in that, The step of performing anomaly correlation on the first detection result and the second detection result to obtain anomaly correlation information includes: The first type of index analysis is performed on the first detection result to obtain the first set of analytical data; The second detection result is analyzed using a second type of index to obtain a second set of analytical data; The first set of parsed data and the second set of parsed data are cross-compared to generate a comparison result set; The comparison result set is classified into event levels according to preset rules, and the abnormal association information is output.

9. A remote monitoring system for food freshness, characterized in that, The remote monitoring method for food freshness according to any one of claims 1-8 includes: The data acquisition module is used to acquire processing technology data and distribution and transportation data of fresh food, calculate the shelf life, and obtain the shelf life of the food. The analysis module is used to perform process detection and anomaly identification on the processing technology data to obtain a first detection result; The association module is used to detect the cumulative temperature value and transportation status of the circulation and transportation data, and integrate the results to obtain a second detection result; The processing module is used to respond and adjust based on the food's shelf life, the first test result, and the second test result to obtain remote control information.

Images