Prediction method for refrigeration storage quality attenuation of shrimp meat

Abstract

The invention discloses a shrimp meat refrigeration storage quality attenuation prediction method, and relates to the technical field of refrigeration storage, and the method comprises the steps: connecting a production line product input end, carrying out sampling and continuous bimodal detection on a shrimp meat product, and obtaining input detection data; and fitting analysis data, generating an equivalent input detection data column, and constructing a storage state sequence. And performing initial quality calibration based on the warehousing state sequence, and establishing an initial quality state. And initializing a quality attenuation mapping model, and determining a quality attenuation factor in combination with the refrigerated storage parameter correction model. And finally, predicting the refrigerated storage quality change of the shrimp meat product according to the attenuation factor. Therefore, the technical effects of realizing more flexible and finer refrigeration storage quality attenuation prediction of the shrimp meat and improving the prediction accuracy and reliability are achieved.

Description

Technical Field

[0001] The invention relates to the technical field of refrigerated storage, and in particular to a method for predicting quality attenuation of shrimp during refrigerated storage. Background Art

[0002] In the field of food cold chain management, especially for the storage of perishable products such as shrimp, the existing cold storage quality monitoring methods mostly rely on a single detection method, such as regular sampling or simple sensor monitoring. These methods are insufficient in the comprehensiveness and continuity of data, and it is difficult to accurately reflect the actual quality changes of shrimp during storage. In addition, traditional methods also have limitations in initial quality assessment and quality attenuation prediction. They are often unable to accurately determine the initial quality state of shrimp and the quality attenuation pattern under different cold storage conditions, thus affecting the effective management and risk control of shrimp quality. Summary of the invention

[0003] The present invention provides a method for predicting the quality attenuation of shrimp during refrigerated storage, so as to solve the technical problems in the prior art of poor detection flexibility and affecting the accuracy and reliability of storage quality prediction, achieve more flexible and precise prediction of the quality attenuation of shrimp during refrigerated storage, and improve the technical effect of improving the accuracy and reliability of the prediction.

[0004] The present invention provides a method for predicting the quality attenuation of shrimp during refrigerated storage, comprising: At the product input end of the interactive target production line, dual-modal detection is implemented on the received shrimp products to obtain input detection data, where the dual-modal detection includes sampling detection and continuous detection.

[0005] A fitting analysis is performed on the input detection data, and the fitting analysis result is output as an equivalent input detection data column, and a storage state sequence is constructed according to the equivalent input detection data column.

[0006] The storage status sequence is used to perform initial quality calibration on the received shrimp products to establish an initial quality status.

[0007] According to the initial mass state, the pre-constructed mass attenuation mapping model is initialized, and in combination with the preset refrigerated storage parameters, the mass attenuation mapping model is corrected to determine the mass attenuation factor.

[0008] The quality attenuation of shrimp products during refrigerated storage is predicted based on the quality attenuation factor.

[0009] In a feasible implementation, dual-mode detection is performed on the received shrimp product to obtain input detection data, including: Determine the sampling strategy based on the preset sampling detection set.

[0010] According to the sampling strategy, random sampling is performed at the product input end to obtain shrimp product samples.

[0011] Based on the sampling test set, a sampling test is performed on the shrimp product sample to obtain a sampling test result.

[0012] In a feasible implementation, performing dual-mode detection on the received shrimp product to obtain input detection data also includes: According to the preset continuous detection set, activate the online detection equipment and configure the online detection parameters.

[0013] Perform continuous testing based on online sensors or rapid testing equipment to obtain continuous testing results.

[0014] In a feasible implementation, before implementing the double detection, the following steps are performed: Obtain a test capability form of a target production line, wherein the test capability form includes a plurality of test items.

[0015] The detection delay characteristics and the detection cost characteristics of the detection capability form are analyzed.

[0016] Threshold screening is performed according to the detection delay feature and the detection cost feature. If any one of the detection delay feature and the detection cost feature exceeds a preset threshold, the corresponding detection item is stored in the sample detection set.

[0017] If both the detection delay feature and the detection cost feature do not exceed a preset threshold, the corresponding detection item is stored in the continuous detection set.

[0018] In a feasible implementation, the sampling detection set includes at least microbial indicators and chemical indicators, and the continuous detection set includes at least sensory indicators, temperature, and moisture.

[0019] In a feasible implementation, a fitting analysis is performed on the input detection data, and a fitting analysis result is output as an equivalent input detection data column, including: Perform correlation analysis between the detection items in the sample detection set and the continuous detection set to generate a correlation coefficient matrix.

[0020] The sampling detection results and the continuous detection results are preprocessed, and time sequence alignment and dimensionless processing are performed.

[0021] The correlation coefficient matrix is ​​used as fitting weights, and the sampling detection results and the continuous detection results after time series alignment and dimensionless processing are fitted to generate the input detection data column.

[0022] In a feasible implementation, the received shrimp products are initially calibrated using the storage status sequence to establish the initial quality status, including: Through statistical analysis methods, a calibration curve of storage status-quality status coefficient is established.

[0023] The first state group of the storage state sequence is obtained.

[0024] The first state group is mapped to the storage state-quality state coefficient calibration curve for state matching, and the point on the storage state-quality state coefficient calibration curve that is closest to the mapping result is determined as the initial quality of the first state group.

[0025] The storage state sequence is traversed to determine the initial quality state.

[0026] In a feasible implementation, according to the initial mass state, a pre-constructed mass attenuation mapping model is initialized, and in combination with preset cold storage parameters, the mass attenuation mapping model is corrected to determine the mass attenuation factor, before that, the method includes: Based on predefined decay variables, wherein the decay variables include an initial mass term and a plurality of refrigerated storage parameter terms.

[0027] Historical shrimp refrigerated storage data is obtained, and numerical extraction is performed on the historical shrimp refrigerated storage data based on the decay variable to generate a training data set.

[0028] The mass attenuation mapping model based on regression analysis is trained according to the training data set until a correlation coefficient of the mass attenuation mapping model is greater than or equal to a qualified constraint.

[0029] The present invention discloses a method for predicting the quality attenuation of shrimp during refrigerated storage, comprising: docking with a product input end of an interactive target production line, performing dual-modal detection on the received shrimp product, and obtaining input detection data, wherein the dual-modal detection includes sampling detection and continuous detection; performing fitting analysis on the input detection data, generating an equivalent input detection data column, and constructing a storage status sequence based on the data column; utilizing the storage status sequence to perform initial quality calibration on the received shrimp product, and establishing an initial quality state; initializing a pre-constructed quality attenuation mapping model based on the initial quality state, and correcting the model in combination with preset refrigerated storage parameters to determine a quality attenuation factor; predicting the quality attenuation trend of the shrimp product during refrigerated storage according to the quality attenuation factor. The method for predicting the quality attenuation of shrimp during refrigerated storage disclosed by the present invention solves the technical problems of poor detection flexibility and affecting the accuracy and reliability of storage quality prediction, realizes a more flexible and precise prediction of the quality attenuation of shrimp during refrigerated storage, and improves the technical effect of improving the accuracy and reliability of the prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic flow chart of a method for predicting the quality attenuation of shrimp during refrigerated storage according to the present invention; Figure 2 The present invention is a schematic diagram of a process for establishing an initial quality state in a method for predicting quality attenuation during refrigerated storage of shrimp. DETAILED DESCRIPTION

[0031] The above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods of the specification to better understand the above technical solution. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments used only to explain the present invention. 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. In addition, it should be noted that, for the convenience of description, only the parts related to the present invention are shown in the drawings, rather than all of them.

[0032] Example, Figure 1 The present invention is a flow chart of a method for predicting the quality attenuation of shrimp during refrigerated storage, wherein the method for predicting the quality attenuation of shrimp during refrigerated storage comprises: At the product input end of the interactive target production line, dual-modal detection is implemented on the received shrimp products to obtain input detection data, where the dual-modal detection includes sampling detection and continuous detection.

[0033] Specifically, the product input end of the target production line refers to the starting part of the production line related to the refrigerated storage of shrimps. This is the key node where the shrimps enter the refrigerated storage system and begin quality inspection. Dual-modal inspection is a method that combines two different inspection methods, namely sampling inspection and continuous inspection. Sampling inspection refers to randomly selecting a certain number of samples from a batch of shrimp products for detailed quality inspection. Continuous inspection uses online sensors or rapid detection equipment to conduct real-time and continuous monitoring of key quality indicators of shrimp products.

[0034] Specifically, for example, when the shrimp enters the cold storage system, the received shrimp products are first subjected to dual-modal testing at the product input end. On the one hand, through sampling testing, a certain proportion of shrimp samples are randomly selected for comprehensive testing to obtain detailed quality parameters including microbial indicators, chemical indicators (such as TVB-N, pH value, etc.); on the other hand, through continuous testing, online sensors or rapid testing equipment are used to monitor key quality indicators such as temperature, humidity, packaging integrity, and surface microbial activity in real time.

[0035] The above-mentioned dual-modal detection method can take into account both the accuracy and continuity of the data. It not only obtains comprehensive quality parameters through sampling detection, but also ensures real-time grasp of changes in shrimp quality through continuous detection, providing a reliable data basis for subsequent quality assessment and prediction.

[0036] By implementing dual-modal detection, the shortcomings of traditional single detection methods in data comprehensiveness and continuity are effectively solved, the flexibility of detection is improved, and it can more comprehensively and truly reflect the quality status of shrimp at the time of receipt, providing high-quality data support for subsequent initial quality calibration and quality attenuation prediction, thereby improving the accuracy and reliability of the entire prediction process.

[0037] In some embodiments, dual-mode detection is performed on the received shrimp product to obtain input detection data, including: Determine a sampling strategy based on a preset sampling test set; randomly draw at the product input end according to the sampling strategy to obtain shrimp product samples; perform sampling tests on the shrimp product samples based on the sampling test set to obtain sampling test results.

[0038] Specifically, the sampling test set is a set of pre-set test items and standards used to guide the sampling test process, and optionally includes the test indicators, the sampling-testing methods corresponding to each indicator, the corresponding sampling intervals, etc.; the sampling strategy is the specific method and plan for extracting samples from shrimp products determined based on the sampling test set, including the number of samples, extraction method, etc.

[0039] Specifically, before the shrimp products enter the cold storage system, first, the sampling strategy is determined according to the preset sampling test set, where the sampling test set is used to clarify the indicators and methods that need to be tested, such as microbial indicators, chemical indicators, etc.; then, according to the requirements of the sampling test set, a sampling strategy is formulated to determine the number and method of samples to be drawn from each batch of shrimp to ensure that the samples are representative; then, shrimp product samples are randomly drawn at the product input end according to the sampling strategy. Random sampling can avoid human bias and ensure that the samples can truly reflect the quality of the entire batch of shrimp; Optionally, the sampling strategy also includes: increasing the sampling ratio for batches with historical quality abnormalities or products from specific suppliers; classifying shrimps of different specifications, origins or batches, and then sampling proportionally from each category; selecting samples at fixed intervals (such as selecting 1 box for every 10 boxes) to improve efficiency.

[0040] Furthermore, based on the sampling test set, the selected shrimp product samples are sampled and tested to obtain the sampling test results, ensuring that the detailed quality parameters of the shrimp products are comprehensively obtained, providing reliable data support for subsequent quality assessment and prediction, thereby improving the accuracy and reliability of the entire prediction process.

[0041] In some embodiments, performing dual-mode detection on the received shrimp product to obtain input detection data further includes: According to the preset continuous detection set, activate the online detection equipment and configure the online detection parameters; perform continuous detection according to the online sensor or rapid detection equipment to obtain continuous detection results.

[0042] Specifically, the continuous detection set is a pre-set set of items and standards suitable for continuous detection, usually including indicators with low detection cost, fast detection speed and little interference with the detection process, such as temperature, humidity, integrity, etc.; the continuous detection set is collected through online detection equipment installed at the shrimp refrigerated storage site. The online detection equipment can obtain the quality data of shrimp products in real time or near real time, such as online temperature sensors, humidity sensors, near-infrared spectroscopy scanners, electronic noses, etc.

[0043] Among them, online detection parameters are specific parameters that need to be set for online detection equipment, such as detection frequency, detection range, alarm threshold, etc., to ensure that the equipment can accurately and effectively obtain quality data. Optionally, if online monitoring detects that the indicators exceed the standard (such as color abnormality, foreign body detection, temperature abnormality, etc.), an early warning will be triggered immediately, and further verification will be carried out in combination with sampling inspection, or relevant personnel / devices will be notified to take measures.

[0044] Specifically, after the shrimp products enter the cold storage system, according to the preset continuous detection set, activate the corresponding online detection equipment and configure the online detection parameters. For example, if the continuous detection set includes temperature detection, activate the temperature sensor installed in the storage area, and set the detection frequency (such as detection every 5 seconds) and alarm threshold (such as alarm when the temperature exceeds 4°C); then, use these online sensors or rapid detection equipment to continuously detect the shrimp products. For example, the temperature sensor will continuously monitor the temperature changes in the storage environment, and the rapid detection equipment will regularly conduct rapid detection of microbial activity on the surface of the shrimp.

[0045] Through the above method, continuous detection is combined with sampling detection to form a dual-modal detection system. For those indicators with low detection cost and fast detection speed, continuous detection can achieve real-time monitoring of shrimp product quality, timely capture small changes in quality indicators, and improve the sensitivity and timeliness of detection. At the same time, the accumulation of continuous detection data also provides a rich data foundation for subsequent quality trend analysis and the establishment of prediction models, so that the prediction model can better fit the actual quality change law and improve the accuracy and reliability of the prediction.

[0046] In some implementations, before performing the dual detection, the following steps are performed: Obtain an inspection capability form for a target production line, the inspection capability form including a plurality of inspection items; analyze the inspection delay characteristics and the inspection cost characteristics of the inspection capability form; perform threshold screening according to the inspection delay characteristics and the inspection cost characteristics, and if any one of the inspection delay characteristics and the inspection cost characteristics exceeds a preset threshold, store the corresponding inspection item in the sample inspection set; if both the inspection delay characteristics and the inspection cost characteristics do not exceed the preset threshold, store the corresponding inspection item in the continuous inspection set.

[0047] Specifically, in order to improve detection efficiency and optimize detection costs, before implementing dual-modal detection (sampling detection + continuous detection), it is necessary to evaluate and classify the detection capabilities of the target production line; the detection capability form of the target production line refers to a list of all detection items that the shrimp refrigerated storage production line can perform. For example, each detection item contains information such as detection content, detection method, and detection equipment; the detection delay feature refers to the time required to complete a certain test, including the time delay of the entire process from sample collection to obtaining the test results, reflecting the timeliness of the test; the detection cost feature refers to the cost required to conduct a certain test, including equipment cost, reagent cost, labor cost, etc., reflecting the economy of the test.

[0048] Specifically, before implementing double testing, first, obtain the testing capability form of the target production line to understand all the testing items that the production line can perform. For example, the form contains multiple testing items including but not limited to: appearance testing (color, black spots, damage, ice coating coverage), weight testing (weight deviation of a single shrimp), physical and chemical indicators (moisture content, pH value, TVB-N, salinity, etc.), microbial testing (total colony count, Escherichia coli, etc.), foreign matter detection (metal, plastic, broken shells, etc.), etc.

[0049] Then, the detection delay characteristics and detection cost characteristics of these detection items are analyzed to evaluate the feasibility of each detection item in terms of time and cost; for example: real-time detection (such as spectral analysis) has low delay and low cost, while laboratory analysis (such as microbial detection) has high delay and high cost.

[0050] Next, set the detection delay threshold and cost threshold, and perform threshold screening based on the analysis results to measure whether each detection item is suitable for continuous detection or sampling detection; if any of the detection delay characteristics or detection cost characteristics of a detection item exceeds the preset threshold, it means that the detection item is not suitable for frequent detection, and it is stored in the sampling detection set and performed through sampling detection; if both the detection delay characteristic and the detection cost characteristic of a detection item do not exceed the preset threshold, it means that the detection item can be conveniently and economically tested frequently, and it is stored in the continuous detection set and performed through continuous detection; through the above process, detection resources can be reasonably allocated to ensure the efficiency and economy of detection.

[0051] Table 1 Exemplary threshold screening correspondence

[0052] In some implementations, the sampling detection set includes at least microbial indicators and chemical indicators; the continuous detection set includes at least sensory indicators, temperature, and moisture.

[0053] Specifically, according to the detection delay characteristics and detection cost characteristics of the detection items, the detection items are divided into sampling detection sets and continuous detection sets. Among them, the sampling detection set is suitable for detection items that require laboratory analysis, culture or complex chemical treatment. The results cannot be obtained in real time and the detection cost is relatively high. It at least includes: microbial indicators, such as total colony count, Escherichia coli, Salmonella, etc., and the detection methods include microbial culture, PCR amplification, etc.; chemical indicators, such as volatile basic nitrogen (TVB-N), pH value, histamine content, heavy metal residues, etc., and the detection methods include chemical titration, spectral analysis, chromatographic analysis, etc.

[0054] Specifically, the continuous detection set is suitable for indicators that can be detected in real time by sensors or rapid detection equipment. The results can be quickly fed back and the detection cost is low. It at least includes: sensory indicators, such as color, black spots, glossiness, breakage, morphological integrity, ice coating coverage, etc., and detection methods include optical sensors and computer vision analysis; temperature, such as product core temperature and cold chain storage temperature, and detection methods include temperature sensors and infrared temperature measurement; moisture content, such as the ratio of free water to bound water, and detection methods include near-infrared spectroscopy and microwave moisture meter.

[0055] The above classification method can optimize detection efficiency and reduce costs while ensuring detection coverage.

[0056] A fitting analysis is performed on the input detection data, and the fitting analysis result is output as an equivalent input detection data column, and a storage state sequence is constructed according to the equivalent input detection data column.

[0057] Specifically, the input test data obtained comes from different detection methods (sampling detection and continuous detection). Due to the different detection methods, the granularity, frequency and accuracy of the data may vary. Therefore, fitting analysis is needed to unify the data format, improve the data continuity of the sampling detection data, and construct a storage status sequence. Among them, the equivalent input test data column is a new data sequence that is generated by fitting analysis of the original input test data and can represent the characteristics of the original data. It fills the missing part of some data and is consistent with the original data in reflecting the overall trend and characteristics of the data.

[0058] Exemplarily, a fitting analysis is performed on the input test data, which involves data cleaning and preprocessing, data alignment and interpolation fitting, etc. The equivalent input test data column after fitting is the product's warehousing status sequence, which is used for subsequent quality tracking and management.

[0059] In some embodiments, fitting analysis is performed on the input detection data, and the fitting analysis result is output as an equivalent input detection data column, including: Perform correlation analysis on the detection items in the sampling detection set and the continuous detection set to generate a correlation coefficient matrix; preprocess the sampling detection results and the continuous detection results, and perform time alignment and dimensionless processing; use the correlation coefficient matrix as fitting weights, fit the sampling detection results and the continuous detection results after time alignment and dimensionless processing, and generate the input detection data column.

[0060] Specifically, the sampling test set refers to a set of test items formed by detailed quality testing of shrimp samples obtained by random sampling, usually including microbial indicators, chemical indicators and other items that require a long time or high cost to test; the continuous test set refers to a set of test items formed by real-time and continuous monitoring of key quality indicators of shrimp products using online sensors or rapid testing equipment, usually including indicators such as temperature and humidity with low testing cost and fast testing speed.

[0061] Specifically, the correlation coefficient matrix is ​​a matrix generated by analyzing the correlation between the detection items in the sampling detection set and the continuous detection set. It is used to represent the degree of correlation between different detection items and provide a weight basis for subsequent data fitting. For example, through analysis, it is found that there is a certain correlation between the microbial indicators in the sampling detection and the temperature indicators in the continuous detection. This correlation will be reflected in the correlation coefficient matrix with corresponding values.

[0062] Specifically, first, the data of different detection items are unified and matched in chronological order to ensure the consistency of the data in the time dimension. At the same time, data of different dimensions and orders of magnitude are converted into dimensionless relative values ​​through dimensionless processing to eliminate the influence of dimension and order of magnitude differences on data fitting.

[0063] Specifically, in the data fitting process, the correlation coefficient matrix is ​​used as the fitting weight to guide the fitting algorithm to attach importance to the data of different detection items, and a suitable fitting algorithm (such as linear regression, nonlinear regression, etc.) is used to fit the sampling detection results and continuous detection results after time series alignment and dimensionless processing to generate input detection data columns, so that the fitting results can better reflect the true characteristics of the data.

[0064] Specifically, during the fitting process, the data needs to be preprocessed, such as removing outliers and performing normalization, to improve the accuracy and reliability of the fitting.

[0065] Specifically, after completing the fitting analysis, the fitting analysis results are output as the equivalent input test data column. Next, the storage state sequence is constructed based on the equivalent input test data column. Each data point in the equivalent input test data column can be regarded as the comprehensive quality state of the shrimp product at a certain moment or a certain batch. These data points are arranged in chronological order or batch order to form a storage state sequence. This sequence can clearly show the quality changes of shrimp products at different time points or different batches when entering the warehouse, providing an orderly and standardized data basis for subsequent initial quality calibration and quality attenuation prediction.

[0066] By fitting the input test data and constructing an equivalent input test data column, the data of the sampling test set is enhanced, which effectively improves the data processing efficiency. At the same time, the core characteristics and trends of the original data are retained, making the data smoother and more regular, which is convenient for subsequent analysis and modeling. The storage status sequence constructed based on the equivalent input test data column can more accurately and comprehensively reflect the quality status of shrimp products when they are stored, providing high-quality data support for initial quality calibration, thereby improving the accuracy and reliability of the entire prediction method.

[0067] The storage status sequence is used to perform initial quality calibration on the received shrimp products to establish an initial quality status.

[0068] Specifically, the storage status sequence is a time series or a status sequence generated by fitting analysis, representing the quality status of the shrimp product at the time of receipt, and reflects the quality changes of the shrimp at different time points or different batches when entering the warehouse.

[0069] Specifically, initial quality calibration is a process of evaluating and determining the initial quality of shrimp products based on the storage status sequence, and establishing a relative quality value that can comprehensively reflect the quality level of shrimp when entering storage, which is used to provide a basis for subsequent quality attenuation prediction; the initial quality status is the relative quality status of shrimp products when entering storage determined after calibration, which is a unified data expression of various test indicators (such as sensory indicators, temperature, moisture, microorganisms and chemical indicators, etc.) after fitting processing, represented by a scalar value, such as 90% means that the quality status of the shrimp product when entering storage is equivalent to 90% of the optimal state.

[0070] In some embodiments, Figure 2 As shown, the received shrimp products are initially calibrated using the storage status sequence to establish the initial quality status, including: Through the statistical analysis method, a warehousing state-quality state coefficient calibration curve is established; the first state group of the warehousing state sequence is obtained; the first state group is mapped to the warehousing state-quality state coefficient calibration curve for state matching, and the point on the warehousing state-quality state coefficient calibration curve that is closest to the mapping result is determined as the initial quality of the first state group; the warehousing state sequence is traversed to determine the initial quality state.

[0071] Specifically, the warehousing status-quality status coefficient calibration curve is established through a statistical analysis method, wherein the horizontal axis represents the fused characteristic value of multiple detection data in the warehousing status sequence, and the vertical axis represents the corresponding quality status coefficient. The curve is used to map the warehousing status data to the quality status coefficient.

[0072] Optionally, the storage status-quality status coefficient calibration curve can also be expressed as a set of multiple space curves in a multi-dimensional space, wherein each dimension corresponds to a detection item.

[0073] Specifically, an initial set of data (such as the first batch of detection data or data within an initial time window) is selected from the warehousing state sequence as the first state group; then, the acquired first state group data is subjected to necessary data preprocessing (such as normalization, dimensionless conversion, etc.), and then mapped to the coordinate space of a pre-established warehousing state-quality state calibration curve. Exemplarily, the mapping result is a point, which reflects the position of the state group relative to the calibration curve.

[0074] Furthermore, by comparing the similarity or distance between the mapping result point and other points on the curve, the closest point is determined, and the vertical axis value corresponding to the point represents the initial quality value of the first state group, which is output as the quality state coefficient; finally, the entire storage state sequence is traversed, and the above mapping and matching operations are repeated for each state group, and the initial quality state of each time point or batch is determined in turn, and finally a complete initial quality state sequence is constructed. This process converts the abstract storage state data into intuitive relative quality values, providing a clear and quantitative benchmark for subsequent quality decay prediction.

[0075] By establishing a calibration curve of the incoming state-quality state coefficient and mapping and matching each state group in the incoming state sequence, the initial quality state of the shrimp product when it enters the warehouse can be accurately determined. This method not only takes into account multiple quality indicators, but also simplifies complex quality state data into a scalar value, which is easy to understand and apply. For example, the initial quality state is expressed as 90%, which means that the quality of the shrimp when it enters the warehouse is equivalent to 90% of the optimal state. This intuitive representation method helps managers quickly grasp the product quality situation. In addition, traversing the entire incoming state sequence for calibration ensures the comprehensiveness and accuracy of the initial quality state, providing a reliable starting point for subsequent quality attenuation predictions, thereby improving the scientificity and practicality of the entire prediction process.

[0076] According to the initial mass state, the pre-constructed mass attenuation mapping model is initialized, and in combination with the preset refrigerated storage parameters, the mass attenuation mapping model is corrected to determine the mass attenuation factor.

[0077] Specifically, after determining the initial quality state of the shrimp product, the pre-built quality decay mapping model is first initialized. During initialization, the initial quality state is used as the starting point of the model and the initial parameter values ​​of the model are set; next, the quality decay mapping model is corrected in combination with the preset refrigerated storage parameters, such as the temperature is set to -18°C and the humidity is maintained at 85%. In this step, the model will adjust its internal coefficients and structures according to these parameters to adapt to the current storage conditions.

[0078] For example, if the storage temperature is low, the model may reduce the rate coefficient of mass decay; if the humidity is high, it may increase the mass decay term related to moisture. In this way, the model can more accurately reflect the decay of shrimp quality under specific refrigerated storage conditions, and ultimately determine the mass decay factor that meets the actual storage conditions.

[0079] Specifically, the quality decay factor is determined by the modified mass decay mapping model, which is a mathematical model or dynamic parameter used to quantify the degree of quality decay of shrimp under specific storage conditions, reflecting the rate and trend of shrimp quality change over time.

[0080] Exemplary. The pre-built models are based on historical data and domain knowledge, combining mathematical forms such as Arrhenius formula, exponential decay model or logistic model to describe the quality decay process.

[0081] By initializing the pre-built quality decay mapping model and modifying it in combination with the cold storage parameters, the model can be closely adapted to the actual storage environment and the initial quality state of the shrimp. This targeted modification improves the accuracy and reliability of the model's quality decay prediction, ensuring that the prediction results can truly reflect the quality change trend of shrimp under specific conditions.

[0082] In some embodiments, according to the initial mass state, a pre-constructed mass attenuation mapping model is initialized, and in combination with preset cold storage parameters, the mass attenuation mapping model is modified to determine the mass attenuation factor, before which the method includes: Based on a predefined attenuation variable, wherein the attenuation variable includes an initial mass item and a plurality of refrigerated storage parameter items; obtaining historical shrimp refrigerated storage data, and performing numerical extraction on the historical shrimp refrigerated storage data based on the attenuation variable to generate a training data set; according to the training data set, training the mass attenuation mapping model based on regression analysis until the correlation coefficient of the mass attenuation mapping model is greater than or equal to a qualified constraint.

[0083] Specifically, the attenuation variables include an initial quality item and multiple cold storage parameter items. The initial quality item represents the relative quality status of the product when it enters the warehouse (for example, the initial quality percentage, such as 90%); the cold storage parameter items include cold storage temperature, humidity, storage time, gas composition (such as oxygen and carbon dioxide concentrations) and other key factors that affect product quality attenuation.

[0084] Specifically, the historical inspection data of normal shrimp products under different refrigeration conditions are collected. The data should include the quality status of each batch of products when they enter the warehouse and the key environmental parameters during storage. Then, the historical data is preprocessed, and the values ​​of each inspection item are extracted based on the predefined attenuation variables to generate a training data set.

[0085] Table 2 Data table of an exemplary training data set

[0086] Specifically, the quality decay mapping model is constructed using the training data set. The optional models include: linear regression model, nonlinear regression model or exponential decay model; model training includes: fitting the model parameters using the training data set, optimizing the parameters using the least squares method, gradient descent and other methods until the correlation coefficient of the model is greater than or equal to the preset qualified constraint value (such as R 2 ≥0.8R).

[0087] The quality attenuation of shrimp products during refrigerated storage is predicted based on the quality attenuation factor.

[0088] Specifically, the determined quality attenuation factor provides a key quantitative basis for quality prediction and management decision-making, which helps to achieve intelligent and refined management of the shrimp refrigerated storage process.

[0089] For example, in a low temperature and low humidity environment, the quality decay factor will predict a slower quality decay rate, thereby guiding managers to adjust the storage temperature in a timely manner, reasonably arrange the storage time and shipping order, and reduce losses and food safety risks caused by quality decay.

[0090] Predicting the quality decay of shrimp products during cold storage through the quality decay factor can achieve a forward-looking grasp of product quality changes. This predictive ability allows managers to take appropriate measures before the quality of shrimp drops to an unsafe level, such as timely adjusting the cold storage temperature, optimizing the storage time, or arranging shipments in advance, thereby ensuring that the shrimp remains within a safe and qualified quality range throughout the storage period. In addition, accurate quality decay prediction can also help companies rationally plan production, storage and sales plans, reduce inventory costs, improve operational efficiency, and enhance market competitiveness.

[0091] In summary, the method for predicting the quality attenuation of shrimp during refrigerated storage provided by the present invention has the following technical effects: By docking with the product input end of the interactive target production line, the received shrimp products are subjected to dual-modal detection to obtain input detection data, wherein the dual-modal detection includes sampling detection and continuous detection; fitting analysis is performed on the input detection data to generate an equivalent input detection data column, and a storage status sequence is constructed based on the data column; the storage status sequence is used to perform initial quality calibration on the received shrimp products to establish an initial quality state; based on the initial quality state, a pre-constructed quality attenuation mapping model is initialized, and the model is corrected in combination with preset refrigerated storage parameters to determine the quality attenuation factor; based on the quality attenuation factor, the quality attenuation trend of the shrimp product during refrigerated storage is predicted, thereby achieving a more flexible and precise prediction of shrimp refrigerated storage quality attenuation, and improving the accuracy and reliability of the prediction.

[0092] It should be understood that the embodiments disclosed in the present invention and the above description can enable those skilled in the art to use the present invention to implement the present invention. At the same time, the present invention is not limited to the above-mentioned embodiments. It should be understood that those skilled in the art can still modify the technical solutions recorded in the above-mentioned embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A method for predicting the quality attenuation of shrimp during refrigerated storage, characterized in that: include: At the product input end of the interactive target production line, dual-mode detection is performed on the received shrimp products to obtain input detection data, wherein the dual-mode detection includes sampling detection and continuous detection; Perform fitting analysis on the input detection data, output the fitting analysis result as an equivalent input detection data column, and construct a storage state sequence according to the equivalent input detection data column; Using the storage status sequence to perform initial quality calibration on the received shrimp products to establish an initial quality status; Initializing a pre-constructed mass attenuation mapping model according to the initial mass state, and correcting the mass attenuation mapping model in combination with preset cold storage parameters to determine a mass attenuation factor; The quality attenuation of shrimp products during refrigerated storage is predicted based on the quality attenuation factor.

2. A method for predicting the quality attenuation of shrimp during refrigerated storage as claimed in claim 1, characterized in that: Perform dual-mode testing on the received shrimp products to obtain input test data, including: Determine the sampling strategy based on the preset sampling detection set; According to the sampling strategy, randomly sampling is performed at the product input end to obtain shrimp product samples; Based on the sampling test set, a sampling test is performed on the shrimp product sample to obtain a sampling test result.

3. A method for predicting the quality attenuation of shrimp during refrigerated storage as claimed in claim 2, characterized in that: Implement dual-mode testing on the received shrimp products to obtain input test data, including: According to the preset continuous detection set, activate the online detection equipment and configure the online detection parameters; Perform continuous testing based on online sensors or rapid testing equipment to obtain continuous testing results.

4. A method for predicting the quality attenuation of shrimp during cold storage as claimed in claim 3, characterized in that: Before implementing dual testing, include: Obtaining a test capability form of a target production line, wherein the test capability form includes a plurality of test items; Analyzing the detection delay characteristics and detection cost characteristics of the detection capability form; Perform threshold screening according to the detection delay feature and the detection cost feature, and if any one of the detection delay feature and the detection cost feature exceeds a preset threshold, store the corresponding detection item in the sample detection set; If both the detection delay feature and the detection cost feature do not exceed a preset threshold, the corresponding detection item is stored in the continuous detection set.

5. A method for predicting the quality attenuation of shrimp during refrigerated storage as claimed in claim 4, characterized in that: The sampling detection set includes at least microbial indicators and chemical indicators; the continuous detection set includes at least sensory indicators, temperature, and moisture.

6. A method for predicting the quality attenuation of shrimp during cold storage as claimed in claim 4, characterized in that: Perform fitting analysis on the input test data, and output the fitting analysis result as an equivalent input test data column, including: Performing correlation analysis between the sample detection set and the detection items in the continuous detection set to generate a correlation coefficient matrix; Preprocessing the sampling detection results and the continuous detection results, and performing time sequence alignment and dimensionless processing; The correlation coefficient matrix is ​​used as fitting weights, and the sampling detection results and the continuous detection results after time series alignment and dimensionless processing are fitted to generate the input detection data column.

7. A method for predicting the quality attenuation of shrimp during refrigerated storage as claimed in claim 6, characterized in that: The received shrimp products are initially calibrated using the storage status sequence to establish an initial quality status, including: Through statistical analysis methods, a calibration curve of storage status-quality status coefficient is established; Acquire the first state group of the storage state sequence; Mapping the first state group to the storage state-quality state coefficient calibration curve for state matching, and determining the point on the storage state-quality state coefficient calibration curve that is closest to the mapping result as the initial quality of the first state group; The storage state sequence is traversed to determine the initial quality state.

8. A method for predicting the quality attenuation of shrimp during cold storage as claimed in claim 7, characterized in that: Initializing a pre-constructed mass attenuation mapping model according to the initial mass state, and correcting the mass attenuation mapping model in combination with preset cold storage parameters to determine a mass attenuation factor, before that, comprising: Based on a predefined attenuation variable, wherein the attenuation variable includes an initial mass term and a plurality of refrigerated storage parameter terms; Acquire historical shrimp refrigerated storage data, and perform numerical extraction on the historical shrimp refrigerated storage data based on the attenuation variable to generate a training data set; The mass attenuation mapping model based on regression analysis is trained according to the training data set until a correlation coefficient of the mass attenuation mapping model is greater than or equal to a qualified constraint.

Images