A Method and System for Preservation Control of Jujubes Based on IoT and AI

Abstract

The present application relates to the field of fruit preservation technology, and specifically discloses a winter jujube preservation control method and system based on IoT and AI, which uses artificial intelligence technology to perform data analysis on near-infrared spectral images and multiple preservation parameters of winter jujube, respectively capturing the quality characteristics of winter jujube and the time domain correlation pattern of multimodal preservation parameters, and through fine-grained interactive response analysis of the two, to dig out the intrinsic relationship between winter jujube quality and preservation parameters, so as to intelligently predict the preservation time of winter jujube, and issue a fresh-keeping quality warning prompt when the predicted preservation time is lower than expected. In this way, precise control of the winter jujube preservation process can be achieved, effectively improving the preservation efficiency, reducing unnecessary waste, and reducing labor costs.

Description

Technical Field

[0001] The present application relates to the technical field of fruit preservation, and more specifically, to a winter jujube preservation control method and system based on IoT and AI. Background Art

[0002] With the improvement of people's living standards and the enhancement of health awareness, the demand for fresh fruits is growing. However, fruits are easily affected by environmental factors during storage after picking and their quality deteriorates. Especially for fruits like winter jujube, which have high requirements for storage conditions, how to effectively maintain their freshness and nutritional value has become an important research topic.

[0003] Traditional preservation methods mainly rely on manual experience, and try to maintain the freshness of the fruit by regularly checking and manually adjusting the storage environment (such as temperature, humidity, etc.). However, this method has obvious limitations, such as inaccurate monitoring and inability to provide quantitative analysis of changes in the quality of winter jujubes, which may lead to a decline in the quality of winter jujubes due to delayed response, resulting in economic losses.

[0004] In recent years, the development of Internet of Things (IoT) and artificial intelligence (AI) technologies has provided new ideas for solving the above problems. IoT technology can monitor changes in the storage environment in real time by deploying various types of sensors and transmit these data to the cloud or local servers for processing. AI technology can realize intelligent management and control of winter jujube preservation based on a large amount of data.

[0005] Therefore, we look forward to a winter jujube preservation control method and system based on IoT and AI. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a winter jujube preservation control method and system based on IoT and AI, which uses artificial intelligence technology to perform data analysis on the near-infrared spectral image and multiple preservation parameters of winter jujube, respectively capturing the quality characteristics of winter jujube and the time domain correlation pattern of multimodal preservation parameters, and through fine-grained interactive response analysis of the two, to dig out the intrinsic connection between winter jujube quality and preservation parameters, so as to intelligently predict the preservation time of winter jujube, and issue a fresh-keeping quality warning prompt when the predicted preservation time is lower than expected. In this way, precise control of the winter jujube preservation process can be achieved, the preservation efficiency can be effectively improved, unnecessary waste can be reduced, and labor costs can be reduced. The warehouse refers to the inside of the AI ​​preservation warehouse.

[0007] Accordingly, according to one aspect of the present application, a winter jujube preservation control method based on IoT and AI is provided, which includes:

[0008] The time queue of the preservation parameters in the warehouse is collected by the sensor matrix, and the near-infrared spectrum image of the winter jujube in the warehouse is obtained by the near-infrared spectrometer, wherein the preservation parameters in the warehouse include temperature value, humidity value, oxygen concentration value and carbon dioxide concentration value;

[0009] Performing multi-modal parameter time series association encoding on the time queue of the in-store fresh-keeping parameter to obtain a multi-modal time series association feature vector of the in-store fresh-keeping parameter;

[0010] Extracting winter jujube quality features from the near infrared spectrum image of the winter jujube in the warehouse to obtain a near infrared spectrum characterization feature vector for the winter jujube quality;

[0011] Performing interactive response analysis on the near infrared spectrum characterization feature vector of the winter jujube quality and the multimodal time series correlation feature vector of the storage fresh-keeping parameters to obtain the winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling feature vector;

[0012] Based on the significant interactive coupling eigenvector of the principal component granularity of the winter jujube quality-preservation parameter, it is determined whether to generate a fresh-keeping quality early warning prompt.

[0013] In the above-mentioned winter jujube preservation control method based on IoT and AI, an interactive response analysis is performed on the winter jujube quality near-infrared spectrum characterization feature vector and the multimodal time series association feature vector of the in-warehouse preservation parameters, including: performing principal component analysis on the winter jujube quality near-infrared spectrum characterization feature vector and the multimodal time series association feature vector of the in-warehouse preservation parameters respectively to obtain a set of principal component feature components of the winter jujube quality near-infrared spectrum and a set of principal component feature components of the in-warehouse preservation parameters; performing optimal pairing multi-scale interactive coupling on the set of principal component feature components of the winter jujube quality near-infrared spectrum and the set of principal component feature components of the in-warehouse preservation parameters to obtain the winter jujube quality-preservation parameter main component granularity significant interactive coupling feature vector.

[0014] In the above-mentioned winter jujube preservation control method based on IoT and AI, the time queue of the in-warehouse preservation parameters is subjected to multimodal parameter temporal association encoding to obtain a multimodal temporal association feature vector of the in-warehouse preservation parameters, including: data structuring the time queue of the in-warehouse preservation parameters to obtain a multimodal time domain joint matrix of the in-warehouse preservation parameters; inputting the multimodal time domain joint matrix of the in-warehouse preservation parameters into a preservation parameter temporal association feature extractor based on a void convolutional neural network model to obtain the multimodal time series association feature vector of the in-warehouse preservation parameters.

[0015] In the above-mentioned IoT and AI-based winter jujube preservation control method, winter jujube quality feature extraction is performed on the near-infrared spectral image of the winter jujube in the warehouse to obtain a near-infrared spectral characterization feature vector of the winter jujube quality, including: inputting the near-infrared spectral image of the winter jujube in the warehouse into a winter jujube quality feature extractor based on a fully connected convolutional neural network model to obtain the near-infrared spectral characterization feature vector of the winter jujube quality.

[0016] In the above-mentioned winter jujube preservation control method based on IoT and AI, principal component analysis is performed on the winter jujube quality near-infrared spectrum characterization feature vector to obtain a set of principal component characteristic components of the winter jujube quality near-infrared spectrum, including: calculating the covariance matrix of the winter jujube quality near-infrared spectrum characterization feature vector to obtain the winter jujube quality near-infrared spectrum characteristic covariance matrix, and performing eigenvalue decomposition on the winter jujube quality near-infrared spectrum characteristic covariance matrix to obtain a set of principal component eigenvalues ​​of the winter jujube quality near-infrared spectrum and a corresponding set of principal component characteristic components of the winter jujube quality near-infrared spectrum.

[0017] In the above-mentioned IoT and AI-based winter jujube preservation control method, the set of principal component characteristic components of the near-infrared spectrum of the winter jujube quality and the set of principal component characteristic components of the in-warehouse preservation parameters are optimally paired and multi-scale interactively coupled to obtain the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector, including: taking each winter jujube quality near-infrared spectrum principal component characteristic component in the set of principal component characteristic components of the winter jujube quality near-infrared spectrum as a query vector, and taking the set of principal component characteristic components of the in-warehouse preservation parameters as a query library, matching the in-warehouse preservation parameter principal component characteristic components that best match each of the query vectors from the query library to obtain a set of best matching pairs of {winter jujube quality near-infrared spectrum principal component characteristic components; in-warehouse preservation parameter principal component characteristic components}; performing multi-scale interactive fusion on the set of the best matching pairs of {winter jujube quality near-infrared spectrum principal component characteristic components; in-warehouse preservation parameter principal component characteristic components} to obtain the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector.

[0018] In the above-mentioned IoT and AI-based winter jujube preservation control method, the in-store preservation parameter principal component characteristic components that best match each of the query vectors are matched from the query library to obtain a set of best matching pairs of {winter jujube quality near-infrared spectrum principal component characteristic components; in-store preservation parameter principal component characteristic components}, including: calculating the Mahalanobis distance between the query vector and each in-store preservation parameter principal component characteristic component in the query library to obtain a set of winter jujube quality-preservation parameter principal component matching difference coefficients; selecting the in-store preservation parameter principal component characteristic component corresponding to the minimum value in the set of winter jujube quality-preservation parameter principal component matching difference coefficients as the optimal matching result of the query vector to obtain the {winter jujube quality near-infrared spectrum principal component characteristic components; in-store preservation parameter principal component characteristic components} best matching pair.

[0019] In the above-mentioned IoT and AI-based winter jujube preservation control method, the set of the best matching pairs of {principal component characteristic components of near-infrared spectrum of winter jujube quality; principal component characteristic components of in-warehouse preservation parameters} is subjected to multi-scale interactive fusion to obtain the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector, including: each of the {principal component characteristic components of near-infrared spectrum of winter jujube quality; principal component characteristic components of in-warehouse preservation parameters} best matching pairs in the set of the best matching pairs of {principal component characteristic components of near-infrared spectrum of winter jujube quality; principal component characteristic components of in-warehouse preservation parameters} best matching pairs is respectively input into a multi-scale interactive response coupling module to obtain a set of multi-scale interactive coupling representation vectors of the best matching pairs of winter jujube quality-preservation parameters; the set of multi-scale interactive coupling representation vectors of the best matching pairs of winter jujube quality-preservation parameters is cascaded to obtain the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector.

[0020] In the above-mentioned IoT and AI-based winter jujube preservation control method, each of the {winter jujube quality near infrared spectrum principal component characteristic components; in-warehouse preservation parameter principal component characteristic components} best matching pairs in the set of the {winter jujube quality near infrared spectrum principal component characteristic components; in-warehouse preservation parameter principal component characteristic components} best matching pairs is respectively input into a multi-scale interactive response coupling module to obtain a set of multi-scale interactive coupling representation vectors of the winter jujube quality-preservation parameter best matching pairs, including: respectively calculating the position point addition, position point subtraction and position point multiplication between the winter jujube quality near infrared spectrum principal component characteristic components and the in-warehouse preservation parameter principal component characteristic components in the {winter jujube quality near infrared spectrum principal component characteristic components; in-warehouse preservation parameter principal component characteristic components} best matching pairs to obtain a first best matching interactive representation vector The first best matching interaction representation vector, the second best matching interaction representation vector and the third best matching interaction representation vector are cascaded and fused, and then input into a one-dimensional convolutional layer including a maximum pooling layer to obtain a multi-scale interaction representation vector of the best matching of the principal component of the winter jujube quality-fresh-keeping parameter; the maximum value among the bi-norm of the principal component characteristic component of the near-infrared spectrum of the winter jujube quality, the bi-norm of the principal component characteristic component of the fresh-keeping parameter in the warehouse and the preset scale adjustment parameter is selected as a scale scaling factor, and each eigenvalue in the multi-scale interaction representation vector of the best matching of the principal component of the winter jujube quality-fresh-keeping parameter is divided by the scale scaling factor to obtain the multi-scale interaction coupling representation vector of the best matching pair of the winter jujube quality-fresh-keeping parameter.

[0021] In the above-mentioned winter jujube preservation control method based on IoT and AI, whether to generate a fresh-keeping quality warning prompt is determined based on the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector, including: inputting the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector into a decoder-based fresh-keeping time prediction module to obtain a fresh-keeping time prediction decoding value; based on the comparison between the fresh-keeping time prediction decoding value and a preset threshold, determining whether to generate the fresh-keeping quality warning prompt.

[0022] According to another aspect of the present application, a winter jujube preservation control system based on IoT and AI is provided, which includes:

[0023] A data acquisition module is used to collect the time queue of the preservation parameters in the warehouse through the sensor matrix, and simultaneously obtain the near-infrared spectrum image of the winter jujube in the warehouse collected by the near-infrared spectrometer, wherein the preservation parameters in the warehouse include temperature value, humidity value, oxygen concentration value and carbon dioxide concentration value;

[0024] A multimodal parameter time series association coding module is used to perform multimodal parameter time series association coding on the time queue of the storage storage parameter to obtain a multimodal time series association feature vector of the storage storage parameter;

[0025] A winter jujube quality feature extraction module is used to extract the winter jujube quality features from the near infrared spectrum image of the winter jujube in the warehouse to obtain a winter jujube quality near infrared spectrum characterization feature vector;

[0026] An interactive response analysis module is used to perform interactive response analysis on the near-infrared spectrum characterization feature vector of the winter jujube quality and the multimodal time series correlation feature vector of the storage fresh-keeping parameters to obtain a significant interactive coupling feature vector of the winter jujube quality-fresh-keeping parameter principal component granularity;

[0027] The fresh-keeping quality warning module is used to determine whether to generate a fresh-keeping quality warning prompt based on the significant interactive coupling eigenvector of the winter jujube quality-fresh-keeping parameter principal component granularity.

[0028] Compared with the prior art, the IoT and AI-based winter jujube preservation control method and system provided by this application use artificial intelligence technology to perform data analysis on the near-infrared spectral image and multiple preservation parameters of winter jujube, respectively capturing the quality characteristics of winter jujube and the time domain correlation pattern of multimodal preservation parameters, and conducting fine-grained interactive response analysis on the two to dig out the intrinsic connection between winter jujube quality and preservation parameters, so as to intelligently predict the preservation time of winter jujube, and issue a fresh-keeping quality warning when the predicted preservation time is lower than expected. In this way, precise control of the winter jujube preservation process can be achieved, effectively improving the preservation efficiency, reducing unnecessary waste, and reducing labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0030] Figure 1 This is a flow chart of a winter jujube preservation control method based on IoT and AI according to an embodiment of the present application.

[0031] Figure 2 This is a data flow diagram of the winter jujube preservation control method based on IoT and AI according to an embodiment of the present application.

[0032] Figure 3 This is a flowchart of step S120 in the winter jujube preservation control method based on IoT and AI according to an embodiment of the present application.

[0033] Figure 4This is a flowchart of step S140 in the winter jujube preservation control method based on IoT and AI according to an embodiment of the present application.

[0034] Figure 5 This is a block diagram of a winter jujube preservation control system based on IoT and AI according to an embodiment of the present application.

[0035] Figure 6 This is a schematic diagram comparing the surface reddening of winter jujube in AI fresh-keeping storage and ordinary cold storage.

[0036] Figure 7 This is a schematic diagram comparing the cross-section of winter jujube in AI fresh-keeping storage and ordinary cold storage.

[0037] The AI ​​fresh-keeping warehouse is an AI fresh-keeping warehouse constructed based on the winter jujube preservation control method based on IoT and AI in this application. DETAILED DESCRIPTION

[0038] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0039] Figure 1 This is a flow chart of a winter jujube preservation control method based on IoT and AI according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of the winter jujube preservation control method based on IoT and AI according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the winter jujube preservation control method based on IoT and AI includes the following steps: S110, collecting the time queue of the preservation parameters in the warehouse through the sensor matrix, and simultaneously obtaining the near-infrared spectrum image of the winter jujube in the warehouse collected by the near-infrared spectrometer, wherein the preservation parameters in the warehouse include temperature value, humidity value, oxygen concentration value and carbon dioxide concentration value; S120, performing multi-modal parameter temporal association encoding on the time queue of the preservation parameters in the warehouse to obtain the multi-modal temporal association feature vector of the preservation parameters in the warehouse; S130, performing winter jujube quality feature extraction on the near-infrared spectrum image of the winter jujube in the warehouse to obtain the near-infrared spectrum characterization feature vector of the winter jujube quality; S140, performing interactive response analysis on the near-infrared spectrum characterization feature vector of the winter jujube quality and the multi-modal temporal association feature vector of the preservation parameters in the warehouse to obtain the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector; S150, determining whether to generate a fresh-keeping quality warning prompt based on the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector.

[0040] In the above-mentioned winter jujube preservation control method based on IoT and AI, the step S110 collects the time queue of the preservation parameters in the warehouse through the sensor matrix, and simultaneously obtains the near-infrared spectrum image of the winter jujube in the warehouse collected by the near-infrared spectrometer, wherein the preservation parameters in the warehouse include temperature value, humidity value, oxygen concentration value and carbon dioxide concentration value. It should be understood that temperature, humidity, oxygen concentration and carbon dioxide concentration are important factors affecting the preservation effect of winter jujube. By real-time monitoring of the changes in these preservation parameters through the sensor matrix, it can be ensured that the environmental conditions are adjusted in time during the preservation process to maintain the best preservation state, thereby extending the shelf life of the product. Near-infrared spectral imaging technology is a non-destructive detection method that can capture the quality characteristics of winter jujube inside, such as moisture content, sugar content, cellulose structure, etc., which is of great significance for evaluating the freshness of winter jujube and predicting its preservation time. In the technical solution of the present application, by combining the preservation environment parameters with the product quality indicators, the correlation and influence relationship between the two can be used to realize the prediction and analysis of the quality changes of winter jujube, thereby providing an important decision-making basis for the preservation control of winter jujube.

[0041] In the above-mentioned winter jujube preservation control method based on IoT and AI, the step S120 performs multi-modal parameter temporal association encoding on the time queue of the in-store preservation parameter to obtain a multi-modal temporal association feature vector of the in-store preservation parameter. Figure 3 FIG. 1 is a flow chart of step S120 in the winter jujube preservation control method based on IoT and AI according to an embodiment of the present application. Figure 3 As shown, the step S120 includes: S121, structuring the time queue of the in-warehouse preservation parameters to obtain a multimodal time-domain joint matrix of the in-warehouse preservation parameters; S122, inputting the multimodal time-domain joint matrix of the in-warehouse preservation parameters into a preservation parameter time series association feature extractor based on a void convolutional neural network model to obtain a multimodal time series association feature vector of the in-warehouse preservation parameters.

[0042] Specifically, the step S121 performs data structuring on the time queue of the preservation parameters in the warehouse to obtain a multimodal time-domain joint matrix of the preservation parameters in the warehouse. It should be understood that since the preservation parameters in the warehouse change dynamically over time, they have obvious time series characteristics, and there are certain interactions between different preservation parameters, such as changes in temperature and humidity may affect the concentrations of oxygen and carbon dioxide. Therefore, in order to fully capture the interrelated action mode of the preservation parameters in time and temperature, the present application first performs data structuring on the time queue of the preservation parameters in the warehouse, and regularizes the multimodal parameters into a matrix form according to the time and parameter sample dimensions, thereby maintaining the time alignment and correlation structure between the multimodal parameter data, thereby obtaining a multimodal time-domain joint matrix of the preservation parameters in the warehouse, so as to facilitate subsequent data analysis and processing.

[0043] Specifically, in step S122, the multimodal time-domain joint matrix of the in-store fresh-keeping parameters is input into the fresh-keeping parameter temporal correlation feature extractor based on the hole convolutional neural network model to obtain the multimodal time-series correlation feature vector of the in-store fresh-keeping parameters. That is, in order to effectively extract the temporal correlation feature pattern between the multimodal parameters from the multimodal time-domain joint matrix of the in-store fresh-keeping parameters, the present application adopts a hole convolutional neural network model as a fresh-keeping parameter temporal correlation feature extractor, so as to utilize the hole convolution characteristics, and perform hole convolution operations on the spatial domain of the multimodal time-domain joint matrix of the in-store fresh-keeping parameters, so as to capture a wider range of contextual correlation information while maintaining a high time resolution, and fully explore the interaction and synergy between different parameters, so as to obtain the multimodal time-series correlation feature vector of the in-store fresh-keeping parameters.

[0044] In the above-mentioned winter jujube preservation control method based on IoT and AI, the step S130 extracts the winter jujube quality features from the near-infrared spectral image of the winter jujube in the warehouse to obtain the near-infrared spectral characterization feature vector of the winter jujube quality. In a specific example of the present application, the step S130 includes: inputting the near-infrared spectral image of the winter jujube in the warehouse into a winter jujube quality feature extractor based on a fully connected convolutional neural network model to obtain the near-infrared spectral characterization feature vector of the winter jujube quality. It should be understood that since the near-infrared spectral image of the winter jujube in the warehouse contains rich spectral information and has high dimensionality and complexity, in order to effectively extract the chemical composition information of the winter jujube revealed in the near-infrared spectral image to reflect the quality characteristics of the winter jujube, the present application uses a fully connected convolutional neural network model to construct a winter jujube quality feature extractor. The fully connected convolutional neural network combines the local perception ability of the convolutional neural network and the global perception ability of the fully connected layer. It can deeply learn and understand the complex patterns in the near-infrared spectral images. By extracting and abstracting the spectral features in the near-infrared spectral images layer by layer, the intrinsic chemical composition information of the winter jujube, such as key indicators such as sugar content, acidity, and moisture content, is mined, thereby obtaining the near-infrared spectral characterization feature vector of the winter jujube quality, which can reflect the maturity, freshness, and possible diseases of the winter jujube, thereby providing a more accurate quality assessment basis for the preservation control of the winter jujube.

[0045] In the above-mentioned winter jujube preservation control method based on IoT and AI, the step S140 performs interactive response analysis on the winter jujube quality near infrared spectrum characterization feature vector and the multimodal time series correlation feature vector of the storage fresh-keeping parameters to obtain the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector. It should be understood that the quality of winter jujube is affected by the fresh-keeping environment and changes over time. Therefore, in order to accurately predict the fresh-keeping time of winter jujube under the current fresh-keeping environment, provide timely fresh-keeping quality warning prompts for management personnel, so as to adjust the fresh-keeping strategy in time and avoid further decline in the quality of winter jujube, the present application further performs interactive response analysis on the winter jujube quality near infrared spectrum characterization feature vector and the multimodal time series correlation feature vector of the storage fresh-keeping parameters to capture the intrinsic connection between the winter jujube quality and the fresh-keeping environment, reveal the influence mechanism of the fresh-keeping environment on the winter jujube quality, and thus realize intelligent prediction of the winter jujube fresh-keeping time.

[0046] Figure 4 FIG. 1 is a flow chart of step S140 in the winter jujube preservation control method based on IoT and AI according to an embodiment of the present application. Figure 4 As shown, the step S140 includes: S141, performing principal component analysis on the near-infrared spectrum characterization feature vector of the winter jujube quality and the multimodal time series correlation feature vector of the in-warehouse preservation parameters to obtain a set of principal component feature components of the near-infrared spectrum of the winter jujube quality and a set of principal component feature components of the in-warehouse preservation parameters; S142, performing optimal pairing multi-scale interactive coupling on the set of principal component feature components of the near-infrared spectrum of the winter jujube quality and the set of principal component feature components of the in-warehouse preservation parameters to obtain the principal component granularity significant interactive coupling feature vector of the winter jujube quality-preservation parameter.

[0047] Specifically, the step S141 includes: calculating the covariance matrix of the characteristic vector characterizing the near-infrared spectrum of the winter jujube quality to obtain the characteristic covariance matrix of the near-infrared spectrum of the winter jujube quality, and performing eigenvalue decomposition on the characteristic covariance matrix of the near-infrared spectrum of the winter jujube quality to obtain a set of principal component eigenvalues ​​of the near-infrared spectrum of the winter jujube quality and a corresponding set of principal component characteristic components of the near-infrared spectrum of the winter jujube quality.

[0048] That is, in the step S141, first, the principal component analysis (PCA) technology is used to process the near-infrared spectrum characterization feature vector of the winter jujube quality and the multimodal time series correlation feature vector of the in-warehouse preservation parameters to reduce the complexity of the data and extract the main features therein, so that the subsequent interactive response analysis process focuses on the main feature parts, thereby obtaining a set of principal component feature components of the near-infrared spectrum of the winter jujube quality and a set of principal component feature components of the in-warehouse preservation parameters.

[0049] Specifically, the step S142 includes: taking each principal component characteristic component of the near-infrared spectrum of the winter jujube quality in the set of principal component characteristic components of the near-infrared spectrum of the winter jujube quality as a query vector, and taking the set of principal component characteristic components of the in-warehouse preservation parameters as a query library, matching the principal component characteristic components of the in-warehouse preservation parameters that best match each of the query vectors from the query library to obtain a set of best matching pairs of {principal component characteristic components of the near-infrared spectrum of the winter jujube quality; principal component characteristic components of the in-warehouse preservation parameters}; performing multi-scale interactive fusion on the set of best matching pairs of {principal component characteristic components of the near-infrared spectrum of the winter jujube quality; principal component characteristic components of the in-warehouse preservation parameters} to obtain the principal component granularity significant interactive coupling characteristic vector of the winter jujube quality-preservation parameter.

[0050] In a specific example of the present application, matching the in-store fresh-keeping parameter principal component characteristic components that best match each of the query vectors from the query library includes: calculating the Mahalanobis distance between the query vector and each in-store fresh-keeping parameter principal component characteristic component in the query library to obtain a set of winter jujube quality-fresh-keeping parameter principal component matching difference coefficients; selecting the in-store fresh-keeping parameter principal component characteristic component corresponding to the minimum value in the set of winter jujube quality-fresh-keeping parameter principal component matching difference coefficients as the optimal matching result of the query vector to obtain the {winter jujube quality near-infrared spectrum principal component characteristic component; in-store fresh-keeping parameter principal component characteristic component} best matching pair.

[0051] In a specific example of the present application, the set of the best matching pairs of {principal component characteristic components of near-infrared spectrum of winter jujube quality; principal component characteristic components of in-warehouse preservation parameters} is subjected to multi-scale interactive fusion, including: inputting each best matching pair of {principal component characteristic components of near-infrared spectrum of winter jujube quality; principal component characteristic components of in-warehouse preservation parameters} in the set of the best matching pairs of {principal component characteristic components of near-infrared spectrum of winter jujube quality; principal component characteristic components of in-warehouse preservation parameters} into a multi-scale interactive response coupling module to obtain a set of multi-scale interactive coupling representation vectors of the best matching pairs of winter jujube quality and preservation parameters; cascading the set of multi-scale interactive coupling representation vectors of the best matching pairs of winter jujube quality and preservation parameters to obtain the principal component granularity significant interactive coupling feature vector of the winter jujube quality and preservation parameters. More specifically, the best matching pair of {principal component characteristic components of near-infrared spectrum of winter jujube quality; principal component characteristic components of in-warehouse preservation parameters} is input into a multi-scale interactive response coupling module, including: respectively calculating the position point addition, position point subtraction and position point multiplication between the principal component characteristic components of near-infrared spectrum of winter jujube quality and the principal component characteristic components of in-warehouse preservation parameters in the best matching pair of {principal component characteristic components of near-infrared spectrum of winter jujube quality; principal component characteristic components of in-warehouse preservation parameters} to obtain a first best matching interactive representation vector, a second best matching interactive representation vector and a third best matching interactive representation vector; the first best matching interactive representation vector, the second best matching interactive representation vector and the third best matching interactive representation vector are respectively calculated. The second best matching interaction representation vector and the third best matching interaction representation vector are cascaded and fused, and then input into a one-dimensional convolutional layer including a maximum pooling layer to obtain a multi-scale interaction representation vector of the best matching of the principal components of the winter jujube quality and the fresh-keeping parameters; the maximum value among the second norm of the principal component characteristic component of the near-infrared spectrum of the winter jujube quality, the second norm of the principal component characteristic component of the fresh-keeping parameters in the warehouse and the preset scale adjustment parameter is selected as a scale scaling factor, and each eigenvalue in the multi-scale interaction representation vector of the best matching of the principal component of the winter jujube quality and the fresh-keeping parameters is divided by the scale scaling factor to obtain the multi-scale interaction coupling representation vector of the best matching pair of the winter jujube quality and the fresh-keeping parameters.

[0052] That is, in step S142, the optimal matching algorithm is used to perform feature query matching on the set of two principal component feature components, and the Mahalanobis distance is used as a metric to query the most relevant feature pair between the two, and a one-to-one correspondence is established between the principal component feature components of the near-infrared spectrum of the winter jujube quality and the principal component feature components of the storage parameters in the warehouse, thereby achieving an accurate match of the interactive response pattern with strong correlation between the winter jujube quality and the storage environment. Next, each best matching pair is input into the multi-scale interactive response coupling module for interactive response analysis of the best pairing features. The multi-scale interactive response coupling module performs multi-level feature interaction and deep learning on the principal component feature components of the near-infrared spectrum of the winter jujube quality and the principal component feature components of the storage parameters in the best matching pair, which can effectively capture the interactive pattern between the winter jujube quality and the storage environment at different abstract levels, and automatically learn and generate the interactive response feature representation of each best matching pair. Finally, by performing a cascade operation on the interactive response feature representations of all the best matching pairs to integrate the global information, the significant interactive coupling feature vector of the principal component granularity of the winter jujube quality-preservation parameters was obtained, which comprehensively reflects the interactive response pattern between the winter jujube quality and the preservation environment, thereby providing a more comprehensive and in-depth analysis basis for the prediction of the winter jujube preservation time.

[0053] In summary, the step S140 includes: processing the winter jujube quality near infrared spectrum characterization feature vector and the in-store fresh-keeping parameter multimodal time series correlation feature vector using the following interactive response analysis formula to obtain the winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling feature vector, wherein the interactive response analysis formula is:

[0054]

[0055]

[0056]

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063] in, and Respectively represent the covariance matrix of the near infrared spectrum characterization feature vector of the winter jujube quality and the multimodal time series correlation feature vector of the storage parameter, is a diagonal matrix consisting of the set of principal component eigenvalues ​​of the near-infrared spectrum of winter jujube quality obtained by eigenvalue decomposition of the near-infrared spectrum characteristic covariance matrix of winter jujube quality, and Respectively represent the first and the second in the set of principal component eigenvalues ​​of the near infrared spectrum of the winter jujube quality The principal component eigenvalues ​​of the near infrared spectrum of winter jujube quality are as follows: is the number of principal component eigenvalues ​​of the near infrared spectrum of the winter jujube quality, is a matrix composed of the principal component characteristic components of the near infrared spectrum of winter jujube quality. , , and They represent the first, second, and third components of the near-infrared spectrum of the winter jujube quality. and The principal component characteristic components of the near infrared spectrum of winter jujube quality represents the transpose of a matrix, is a diagonal matrix consisting of the set of principal component eigenvalues ​​of the in-store fresh-keeping parameters obtained by eigenvalue decomposition of the in-store fresh-keeping parameter characteristic covariance matrix, and Respectively represent the first and the second in the set of principal component eigenvalues ​​of the storage parameter The principal component eigenvalues ​​of the fresh-keeping parameters in the library, is a matrix composed of the set of principal component characteristic components of the storage parameters. , , , and respectively represent the first, second, and third , and The principal component characteristic components of the fresh-keeping parameters in the library, Indicates The principal component characteristic components of near infrared spectrum of winter jujube quality The inverse matrix of the covariance matrix between the principal component characteristic components of the fresh-keeping parameters in the library, Indicates the index corresponding to the minimum value. Indicates that The index of the principal component characteristic component of the near infrared spectrum of the winter jujube quality with the smallest semantic difference in the storage parameter principal component characteristic component, represents dot product, Indicates point addition, Indicates point reduction, represents a one-dimensional convolution operation, represents the maximum pooling operation, represents the bi-norm of the eigenvector, To preset the scale adjustment parameters, is the maximum value function, , , and Respectively represent the first, second, and and The optimal matching pair of winter jujube quality and preservation parameters for multi-scale interactive coupling is represented by a vector. Indicates cascade operation, It represents the significant interactive coupling eigenvector of the principal component granularity of the winter jujube quality-preservation parameters.

[0064] In the above-mentioned winter jujube preservation control method based on IoT and AI, the step S150 determines whether to generate a fresh-keeping quality warning prompt based on the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector. In a specific example of the present application, the step S150 includes: inputting the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector into a decoder-based fresh-keeping time prediction module to obtain a fresh-keeping time prediction decoding value; based on the comparison between the fresh-keeping time prediction decoding value and a preset threshold, determining whether to generate the fresh-keeping quality warning prompt.

[0065] That is, a decoder is used to construct a fresh-keeping time prediction module, which performs feature analysis on the winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling eigenvector through a multi-layer neural network structure, and can effectively learn the interactive response relationship between the winter jujube quality and the fresh-keeping environment contained therein, so as to make an intelligent prediction of the fresh-keeping time of the winter jujube under the current fresh-keeping environment. It should be understood that the fresh-keeping time refers to the length of time when the quality of the winter jujube decreases to the point where it is no longer suitable for sale or consumption, and is a time indicator to measure whether the winter jujube can maintain its freshness and quality under specific storage conditions. When the predicted fresh-keeping time is lower than the preset threshold, a fresh-keeping quality warning prompt is automatically generated to help managers take timely measures, such as adjusting storage conditions or speeding up sales to reduce losses.

[0066] Preferably, the method of inputting the winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling feature vector into a decoder-based fresh-keeping time prediction module to obtain a fresh-keeping time prediction decoded value includes: respectively calculating the first characteristic sum value of all eigenvalues ​​of the winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling feature vector, the second characteristic sum value of all eigenvalues ​​of the winter jujube quality near-infrared spectrum characterization feature vector, and the third characteristic sum value of all eigenvalues ​​of the multimodal time series association feature vector of the in-warehouse fresh-keeping parameter; dividing the first characteristic sum value by the second characteristic sum value and the third characteristic sum value, respectively, and calculating the square of the sum of the squares of the two quotients. Root to obtain the significant interactive coupling distribution probability value of the principal component granularity of winter jujube quality-fresh-keeping parameters; calculate the product of the first winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling eigenvalue of the winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling eigenvalue and the winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling distribution probability value to obtain the first winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling characteristic distribution probability product; calculate the second winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling eigenvalue of the winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling eigenvalue and the product .... The product of the absolute values ​​of the difference in the probability values ​​of the significant interactive coupling distribution of the principal component particle size of the freshness parameter is obtained to obtain the second winter jujube quality-freshness parameter principal component particle size significant interactive coupling characteristic inverse distribution probability product; the quotient of the first winter jujube quality-freshness parameter principal component particle size significant interactive coupling characteristic value divided by one minus the absolute value of the difference in the probability values ​​of the significant interactive coupling distribution of the principal component particle size of the winter jujube quality-freshness parameter is calculated to obtain the first winter jujube quality-freshness parameter principal component particle size significant interactive coupling characteristic inverse distribution probability quotient; the quotient of the second winter jujube quality-freshness parameter principal component particle size significant interactive coupling characteristic value and the winter jujube quality-freshness parameter principal component particle size significant interactive coupling characteristic inverse distribution probability quotient is calculated. The quotient of the probability value of the interaction coupling distribution is added to obtain the probability quotient of the significant interaction coupling characteristic distribution of the second winter jujube quality-fresh-keeping parameter principal component particle size; after adding the probability product of the significant interaction coupling characteristic distribution of the first winter jujube quality-fresh-keeping parameter principal component particle size and the probability product of the significant interaction coupling characteristic inverse distribution of the second winter jujube quality-fresh-keeping parameter principal component particle size, the difference between the probability quotient of the significant interaction coupling characteristic inverse distribution of the first winter jujube quality-fresh-keeping parameter principal component particle size and the probability quotient of the significant interaction coupling characteristic distribution of the second winter jujube quality-fresh-keeping parameter principal component particle size is subtracted to obtain the significant interaction coupling correction characteristic value of the winter jujube quality-fresh-keeping parameter principal component particle size;For each first winter jujube quality-fresh-keeping parameter main component granularity significant interactive coupling eigenvalue and second winter jujube quality-fresh-keeping parameter main component granularity significant interactive coupling eigenvalue of the winter jujube quality-fresh-keeping parameter main component granularity significant interactive coupling eigenvalue, the winter jujube quality-fresh-keeping parameter main component granularity significant interactive coupling correction feature matrix composed of the winter jujube quality-fresh-keeping parameter main component granularity significant interactive coupling correction eigenvalue is matrix-multiplied with the winter jujube quality-fresh-keeping parameter main component granularity significant interactive coupling eigenvector to obtain the corrected winter jujube quality-fresh-keeping parameter main component granularity significant interactive coupling eigenvector; the corrected winter jujube quality-fresh-keeping parameter main component granularity significant interactive coupling eigenvector is input into the decoder-based fresh-keeping time prediction module to obtain the fresh-keeping time prediction decoding value. ;

[0067] The correction process of the winter jujube quality-preservation parameter principal component particle size significant interactive coupling eigenvector can be expressed by the following correction formula:

[0068]

[0069]

[0070]

[0071]

[0072] in, represents the characteristic vector of the near infrared spectrum characterization of the winter jujube quality, represents each eigenvalue of the near infrared spectrum characterization eigenvector of the winter jujube quality, represents the multi-modal time series correlation feature vector of the storage parameters, represents each eigenvalue of the multimodal time series correlation eigenvector of the storage parameter, It represents the significant interactive coupling eigenvector of the principal component granularity of the winter jujube quality-preservation parameter, represents each eigenvalue of the principal component granularity significant interactive coupling eigenvector of the winter jujube quality-preservation parameter, is the probability value of the significant interactive coupling distribution of the principal component particle size of the winter jujube quality-preservation parameter, Indicates the significant interaction coupling correction eigenvalue of the principal component particle size of each winter jujube quality-preservation parameter, The principal component granularity significant interactive coupling correction characteristic matrix of winter jujube quality-preservation parameters composed of the principal component granularity significant interactive coupling correction eigenvalues ​​of each winter jujube quality-preservation parameter is represented. represents the matrix multiplication operation, It represents the significant interactive coupling eigenvector of the principal component granularity of the corrected winter jujube quality-preservation parameters.

[0073] That is, when the near-infrared spectral characterization feature vector of the winter jujube quality and the multimodal temporal association feature vector of the in-store fresh-keeping parameters respectively represent the near-infrared spectral image semantic features of the winter jujube in the store and the temporal multimodal semantic features of the in-store fresh-keeping parameters, when performing the optimal matching interactive response analysis based on the feature principal component granularity, the winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling feature vector will also cause interactive response diversity based on the optimal matching difference of feature principal components and granularity due to the feature modality difference between the near-infrared spectral characterization feature vector of the winter jujube quality and the multimodal temporal association feature vector of the in-store fresh-keeping parameters, thereby affecting the iterative consistency of the decoding mapping and reducing the accuracy of the decoding result.

[0074] Therefore, the present application aims at the regional-boundary integral relationship of the high-dimensional characteristic manifold of the winter jujube quality-preservation parameter main component granularity significant interactive coupling feature vector, and defines the feature set of the winter jujube quality-preservation parameter main component granularity significant interactive coupling feature vector as a whole relative to the winter jujube quality near-infrared spectral characterization feature vector and the in-warehouse multimodal time series correlation feature vector as a manifold constraint boundary to approximate the high-dimensional manifold of the feature set of the winter jujube quality-preservation parameter main component granularity significant interactive coupling feature vector composed of eigenvalue pairs, so as to avoid the ambiguity in the mapping of the diversified feature representation of the winter jujube quality-preservation parameter main component granularity significant interactive coupling feature vector to the local manifold representation in the high-dimensional feature decoding regression convergence space, improve the execution iteration consistency of each local feature distribution in the mapping task, improve the feature decoding regression convergence effect, and improve the accuracy of the fresh-keeping time prediction decoding value obtained by the decoder-based fresh-keeping time prediction module.

[0075] In summary, the winter jujube preservation control method based on IoT and AI according to the embodiment of the present application is explained, which uses artificial intelligence technology to perform data analysis on the near-infrared spectral image and multiple preservation parameters of the winter jujube, respectively capturing the quality characteristics of the winter jujube and the time domain correlation pattern of the multimodal preservation parameters, and through fine-grained interactive response analysis of the two, to dig out the intrinsic relationship between the quality of the winter jujube and the preservation parameters, so as to intelligently predict the preservation time of the winter jujube, and issue a fresh-keeping quality warning prompt when the predicted preservation time is lower than expected. In this way, it is possible to achieve precise control of the winter jujube preservation process, effectively improve the preservation efficiency, reduce unnecessary waste, and reduce labor costs.

[0076] Figure 5 FIG. 1 is a block diagram of a winter jujube preservation control system based on IoT and AI according to an embodiment of the present application. Figure 5As shown, the IoT and AI-based winter jujube preservation control system 100 according to the embodiment of the present application includes: a data acquisition module 110, which is used to collect the time queue of the preservation parameters in the warehouse through the sensor matrix, and simultaneously obtain the near-infrared spectrum image of the winter jujube in the warehouse collected by the near-infrared spectrometer, wherein the preservation parameters in the warehouse include temperature value, humidity value, oxygen concentration value and carbon dioxide concentration value; a multimodal parameter time series association coding module 120, which is used to perform multimodal parameter time series association coding on the time queue of the preservation parameters in the warehouse to obtain the multimodal time series association feature vector of the preservation parameters in the warehouse; winter jujube quality The feature extraction module 130 is used to extract the winter jujube quality features from the near-infrared spectrum image of the winter jujube in the warehouse to obtain the near-infrared spectrum characterization feature vector of the winter jujube quality; the interactive response analysis module 140 is used to perform interactive response analysis on the near-infrared spectrum characterization feature vector of the winter jujube quality and the multimodal time series correlation feature vector of the storage parameter to obtain the winter jujube quality-preservation parameter main component granularity significant interactive coupling feature vector; the preservation quality early warning module 150 is used to determine whether to generate a preservation quality early warning prompt based on the winter jujube quality-preservation parameter main component granularity significant interactive coupling feature vector.

[0077] Here, those skilled in the art can understand that the specific operations of each module in the above-mentioned winter jujube preservation control system based on IoT and AI have been referred to above. Figures 1 to 4 The description of the winter jujube preservation control method based on IoT and AI is introduced in detail, and therefore, its repeated description will be omitted.

[0078] Example 1

[0079] Taking "Linyi winter jujube" as the test material, we selected fruits with regular shape, uniform size, no pests and diseases, and no mechanical damage. After picking, they were quickly sent back to the cold storage and pre-cooled at 0±0.5℃ for 12 hours. After pre-cooling, the fruits were packed into special microporous plastic wrap bags, with 4 kg in each bag. The experimental winter jujube is divided into two maturity levels: white jujube and 1 / 3 red jujube. The experiment was divided into two treatment groups: the experimental group used the AI ​​fresh-keeping warehouse constructed by the above-mentioned winter jujube preservation control method based on IoT and AI, while the control group used ordinary cold storage. The various indicators were measured regularly, and each measurement was repeated at least three times.

[0080] 1. Breathing intensity

[0081] Airflow method determination. Take 1kg of winter jujube, put it into the respiration chamber, seal it, start the vacuum pump to evacuate the CO2 in the respiration chamber and the pipeline for half an hour, and after half an hour, use a pipette to absorb 10mL of 0.4 mol / L NaOH solution and put it into the absorption tube, and connect the absorption tube to the airflow pipeline. Absorb the CO2 produced by the fruit respiration. After 1 hour, take out the absorption tube and transfer the absorption liquid into a 250mL triangular flask and wash it with distilled water 5 times until it is neutral, add 5mL of saturated BaCl2 solution to eliminate the interference of Na2CO3, 2 drops of phenolphthalein indicator, titrate with 0.4 mol / L oxalic acid standard solution, and use the same method to make a blank.

[0082] Calculation formula:

[0083]

[0084] Where: X is the breathing intensity , is the milliliters (mL) of the standard solution of oxalic acid consumed during blank titration, is the number of milliliters (mL) of the standard solution of oxalic acid consumed during sample titration, N is the equivalent concentration of the standard solution of oxalic acid (mol / L), W is the sample weight (kg), and t is the time (h).

[0085] The results of the measurement show that the respiration rate of winter jujube at different maturity stages has roughly the same changing trend under various treatment conditions. In the early stage of storage, the respiration rate will experience a decline and then gradually rise. In contrast, the peak respiration rate of winter jujubes of two maturity levels was lower than that of the control group under the storage conditions of the AI ​​fresh-keeping warehouse. During the entire storage period, winter jujubes with higher maturity showed a higher respiration rate. Through variance analysis, it was found that there was a significant difference in the respiration intensity between the experimental group and the control group. At 21 days of storage, the white jujube and 1 / 3 red jujube of the control group and the experimental group reached a respiration peak, but during this period, the respiration intensity of the white jujube and 1 / 3 red jujube of the experimental group was significantly lower than that of the control group, and the subsequent decline rate was more stable and faster than that of the control group. Further variance analysis showed that there were significant differences in the respiration intensity between the white jujube of the control group and the white jujube of the experimental group, and between the 1 / 3 red jujube of the control group and the 1 / 3 red jujube of the experimental group (P<0.05). Therefore, it can be concluded that regardless of the maturity, the respiration intensity of the control group is higher than that of the experimental group, which shows that the AI ​​fresh-keeping technology described in this application has a significant effect in inhibiting the respiration of winter jujube.

[0086] 2. Fruit hardness

[0087] Use a scraper to cut off 0.5 cm² of the peel at the symmetrical part of the fruit body and then use a fruit hardness tester to measure. The probe diameter is 3.5 mm. Repeat 10 jujubes each time. Take 2 points symmetrically on each jujube fruit and remove the peel at the measuring part. Use a hardness tester to test the hardness and take the average value.

[0088] The results showed that the hardness of winter jujube generally showed a downward trend during storage. At the beginning of storage, the hardness of winter jujube in different treatment groups and different maturity showed different degrees of increase, and then began to gradually decrease. Through variance analysis, it was found that there was a very significant negative correlation between the hardness and storage time between the white jujube in the experimental group and the white jujube in the control group, and between the 1 / 3 red jujube in the experimental group and the 1 / 3 red jujube in the control group (P<0.01). The correlation coefficients were r1=-0.8677, r2=-0.8500 for white winter jujube; r1=-0.8769, r2=-0.6867 for 1 / 3 red winter jujube.

[0089] Throughout the storage period, the pulp hardness of the experimental group was always higher than that of the control group, and the difference between the two groups was statistically significant (P<0.05). From the 15th to the 21st day of storage, the hardness of the winter jujube decreased relatively slowly; however, from the 22nd day to the end of storage, the rate of decline of the hardness of the control group accelerated. Among all maturity and treatment groups, the hardness of 1 / 3 of the red dates in the control group decreased most significantly, with a hardness of 9.26Kg / cm² in the early stage of storage and 5.02Kg / cm² at the end. Throughout the storage process, the hardness of the white dates in the experimental group remained at the highest level, with a hardness of 10.64Kg / cm² in the early stage of storage and 10.02Kg / cm² at the end. This shows that the AI ​​preservation technology described in this application has a significant effect in maintaining the hardness of winter jujube.

[0090] 3. Soluble solids content

[0091] The test was carried out using a handheld sugar meter. Ten jujube fruits were tested each time, and two points were taken symmetrically from each jujube fruit. The peel of the tested part was removed. The pulp was taken to measure the soluble solid content and the average value was taken.

[0092] The results showed that the total soluble solids (TSS) content of winter jujube showed a gradually decreasing trend during storage. Specifically, during the entire storage period, the TSS content of 1 / 3 red jujube in the experimental group was always higher than that of 1 / 3 red jujube in the control group; similarly, the TSS content of white jujube in the experimental group was also higher than that of white jujube in the control group during the entire storage period. Through variance analysis, it was found that there was a significant negative correlation between the white jujube in the experimental group and the control group, as well as between the storage time (P<0.05), with correlation coefficients of r1=-0.8179 and r2=-0.7779, respectively, and the difference between the two groups was statistically significant (P<0.05). For the 1 / 3 red jujube in the experimental group and the 1 / 3 red jujube in the control group, their negative correlation with the storage time was more significant (P<0.01), with correlation coefficients of r1=-0.9444 and r2=-0.8804, respectively, and the difference between the two groups also reached statistical significance (P<0.05). During the storage period, the TSS content of the white dates in the control group decreased by 3.635% from the beginning to the end, while that of the white dates in the experimental group decreased by 2.11%; the TSS content of 1 / 3 of the red dates in the control group decreased by 5.86%, while that of 1 / 3 of the red dates in the experimental group decreased by 2.46%. Throughout the storage period, the TSS content of winter dates with a higher degree of maturity was always higher than that of winter dates with a lower degree of maturity. Based on these data, it can be concluded that the AI ​​preservation technology described in this application has a significant effect in maintaining the TSS content of winter dates.

[0093] 4. Determination of ascorbic acid (VC) content

[0094] Weigh 15g of winter jujube pulp, dilute to 100mL with 1% oxalic acid, take 10mL of the diluted solution and put it in a 100mL conical flask, add 20 drops of 1% starch, then add 20mL of 1% oxalic acid, and titrate with standard iodine solution (0.01mol / L) until it turns blue and does not fade within 15 seconds.

[0095] Calculation formula:

[0096]

[0097] Where: T is the mg of VC in every 100g of fresh sample, H is the titer of iodine solution, V is the milliliters of iodine solution consumed, is the volume of the filtrate used in the measurement, and m is the mass of the fresh sample.

[0098] Ascorbic acid content is an important indicator of the quality of winter jujube. The measurement results show that the ascorbic acid content of winter jujube has a slow increase process in the early storage period, then reaches a peak value, and then gradually decreases. This indicates that during the storage of winter jujube, ascorbic acid has a process of accumulation in the body. The ascorbic acid content of the white jujube in the control group at the early storage period, peak period and late storage period were 213.2mg / 100g·Fw, 282.86mg / 100g·Fw, and 176.88mg / 100g·Fw, respectively; the ascorbic acid content of the white jujube in the experimental group at the early storage period, peak period and late storage period were 213.2mg / 100g·Fw, 299.33mg / 100g·Fw, and 182.5mg / 100g·Fw, respectively; The ascorbic acid content of the 1 / 3 red dates in the group at the initial storage, peak storage and end storage were 223.5mg / 100g·Fw, 275.89mg / 100g·Fw, 150mg / 100g·Fw; the ascorbic acid content of the 1 / 3 red dates in the experimental group at the initial storage, peak storage and end storage were 223.5mg / 100g·Fw, 264.64mg / 100g·Fw, 210.55mg / 100g·Fw. The variance analysis showed that the white dates in the control group and the experimental group were extremely significantly negatively correlated with the storage time (P<0.01), with correlation coefficients r1=-0.8044 and r2=-0.8593, respectively. During the entire storage period, the difference between the two reached a significant level (P<0.05). The 1 / 3 red dates in the control group and the 1 / 3 red dates in the experimental group were significantly negatively correlated with the storage time (P<0.01), with correlation coefficients of r1=-0.8440 and r2=-0.8777, respectively. During the entire storage period, the difference between the two reached a significant level (P<0.05). It can be seen that at the end of the storage period, the ascorbic acid content of winter dates at each maturity level was higher than that of the control, indicating that the AI ​​preservation technology described in this application has a significant effect on maintaining VC.

[0099] 5. Ethylene release

[0100] Determination by gas chromatography. Take gas from the top of the vacuum dryer to collect ethylene, collect it in a small bottle by drainage method, and seal it with a cap. Store in a constant temperature room (18±0.5)℃. The gas sample is measured by Vav-ian CP-3800 gas chromatograph, hydrogen flame ionization detector (FID), carrier gas N2 flow rate 29mL / min, fuel gas H2 flow rate 1mL / min. Injection temperature 150℃, column temperature 85℃, detection temperature 200℃. The gas chromatograph is equipped with a chromatographic workstation, and chromatographic software is used for data processing. All prepared samples are measured individually, and each treatment is repeated 3 times.

[0101] Ethylene is a vital plant hormone that promotes fruit ripening and senescence. One of the main functions of ethylene is to trigger the respiratory climacteric phenomenon of respiratory climacteric fruits, which accelerates the softening process of the fruit and makes the fruit suitable for consumption. The results showed that the ethylene release of the jujube fruits treated with two different methods showed a significant downward trend at the beginning of storage. Subsequently, during the storage process, the two treated jujube fruits showed a smaller ethylene release peak, after which the ethylene release decreased again until it reached a stable state. Specifically, the white jujube and 1 / 3 red jujube in the control group reached a significant ethylene release peak on the 18th day and the 14th day, respectively, while the white jujube and 1 / 3 red jujube in the experimental group only showed ethylene release peaks on the 20th day and the 18th day, respectively. After reaching the peak, the ethylene release of both the experimental and control groups decreased. In this study, the ethylene release of the jujube fruits in the experimental group was always lower than that of the control group.

[0102] AI fresh-keeping warehouse effectively slows down the respiration and ethylene production of winter jujube, thus extending the shelf life of winter jujube. Figure 6 , 7 As shown in the figure, the redness of the surface of winter jujube in AI fresh-keeping warehouse is significantly lower than that in ordinary cold storage, and the browning degree of the cut surface of the jujube kernel is also lighter. This shows that AI fresh-keeping warehouse has a significant positive effect on the preservation effect of winter jujube.

[0103] 6. Ethanol Determination

[0104] Weigh 20 g of the sample, wash it with 150 mL of water into a 500 mL flask, connect a condenser to the bottom of the flask for heating and distillation, collect the distillate in a 100 mL volumetric flask until it reaches the scale, cover the bottle cap, and mix well.

[0105] Put 20 mL of 0.01 mol / L K2Cr2O7 in a 200 mL Erlenmeyer flask, take 5 mL of concentrated sulfuric acid with a measuring cylinder, slowly drip 10 mL of distilled water, and shake continuously. Connect a condenser and place it on an asbestos net to heat and reflux, and boil the solution in the flask slightly for 10 minutes.

[0106] After the refluxed solution has cooled, rinse the condenser with water so that all the solution can be contained in the 200mL Erlenmeyer flask intact. Then carefully transfer the solution into a 500mL Erlenmeyer flask, rinse the 200mL Erlenmeyer flask with about 200mL of water, add 1g of potassium iodide, cover the flask, and place it in a dark place for 5min.

[0107] Add 0.01 mol / L sodium thiosulfate solution from the burette. When the color of the solution changes from orange to light yellow, add 5 mL of starch solution and continue titrating until the solution changes from blue to light green. Record the volume of sodium thiosulfate consumed.

[0108] Calculation formula:

[0109]

[0110] Where: m is the weight of the sample, is the number of milliliters of potassium dichromate solution added, is the concentration of potassium dichromate solution, is the milliliters of sodium thiosulfate consumed during titration, is the concentration of sodium thiosulfate solution.

[0111] The results showed that the ethanol content of winter jujube showed a trend of gradual increase during the entire storage period. Specifically, the ethanol content of the white jujube in the control group at the beginning and end of storage was 0.04599g / 100g·Fw and 0.0846g / 100g·Fw, respectively, an increase of 0.03841g / 100g·Fw; while the ethanol content of the white jujube in the experimental group at the same time point increased from 0.04599g / 100g·Fw to 0.0725g / 100g·Fw, an increase of 0.02651g / 100g·Fw. For 1 / 3 of the red dates in the control group, the ethanol content at the beginning and end of storage was 0.0529g / 100g·Fw and 0.1092g / 100g·Fw, respectively, an increase of 0.0563g / 100g·Fw; the ethanol content of 1 / 3 of the red dates in the experimental group increased from 0.0529g / 100g·Fw to 0.06557g / 100g·Fw, an increase of 0.01653g / 100g·Fw.

[0112] The results of variance analysis showed that there was a significant positive correlation between the ethanol content of the white dates in the control group and the experimental group and the storage time (P<0.05), with correlation coefficients of r1=0.8782 and r2=0.8612, respectively, and there was a significant difference between the two (P<0.05). Similarly, there was a significant positive correlation between the ethanol content of 1 / 3 red dates in the control group and 1 / 3 red dates in the experimental group and the storage time (P<0.05), with correlation coefficients of r1=0.9299 and r2=0.9254, respectively, and there was a significant difference between the two (P<0.05). In summary, during the entire storage process, the ethanol content of the white dates and red dates of each maturity in the different treatment groups was higher in the control group than in the experimental group, and the growth rate of the control group was greater than that of the experimental group. This shows that the AI ​​preservation technology described in this application has played an effective role in inhibiting the increase in ethanol content.

[0113] 7. Acetaldehyde determination

[0114] The extraction process of acetaldehyde is the same as that of ethanol.

[0115] Dissolve 1 mL of redistilled triacetaldehyde solution in 1000 mL of distilled water, i.e., 1000 μL of the stock solution. Take 2 mL of the above triacetaldehyde stock solution and quantify it to 100 mL, with a concentration of 20 μL / L, and then dilute it to a standard solution of 2~15 μL / L according to the table below to make a standard curve.

[0116]

[0117] Take 1mL of the extract and put it into a test tube, then add 0.05mL of 5% copper sulfate, then put the test tube in ice water, slowly add 8mL of concentrated sulfuric acid, shake well. Then add 0.2mL of 1% p-hydroxybiphenyl, shake well, and place at 30℃ for 30min, gently shake the test tube several times in the middle, and then put the test tube into boiling water for 1.5min. After cooling, it will turn purple, and the color is compared at 560nm of the spectrophotometer.

[0118] Calculation formula:

[0119]

[0120] Where: p is the acetaldehyde concentration obtained from the standard curve, V is the volume of winter jujube juice used for colorimetry, 10 is the total volume of the absorption liquid, 100 is used to convert the juice volume to 100 mL, and 1000 is used to convert the acetaldehyde content (μg) to mg.

[0121] The results showed that the acetaldehyde content in winter jujube fruits showed a gradual increase during the entire storage period. At each maturity stage, the acetaldehyde content of the untreated control group was always higher than that of the treated experimental group. In particular, the acetaldehyde content of the white jujube in the experimental group increased the least. Specifically, the acetaldehyde content of the white jujube in the control group was 2.078μg / mL and 4.815μg / mL at the beginning and end of storage, respectively, an increase of 2.737μg / mL; while the acetaldehyde content of the white jujube in the experimental group was 2.078μg / mL and 2.105μg / mL at the same period, an increase of only 0.027μg / mL. For 1 / 3 mature red jujube, the acetaldehyde content of the control group at the beginning and end of storage was 2.486μg / mL and 4.302μg / mL, an increase of 1.816μg / mL; the acetaldehyde content of the experimental group was 2.486μg / mL and 3.302μg / mL, an increase of 0.816μg / mL.

[0122] Variance analysis showed that there was a significant positive correlation between the acetaldehyde content of the white dates in the control group and the experimental group and the storage time (P<0.05), and the correlation coefficients were r1=0.9952 and r2=0.758, respectively, indicating that there was a significant difference between the two during the entire storage process (P<0.05). Similarly, there was a significant positive correlation between the acetaldehyde content of 1 / 3 red dates in the control group and 1 / 3 red dates in the experimental group and the storage time (P<0.05), and the correlation coefficients were r1=0.9175 and r2=0.9391, respectively, indicating that there was also a significant difference between the two during the entire storage process (P<0.05). In summary, the acetaldehyde content of the white dates in the control group increased the most, while the acetaldehyde content of the white dates in the experimental group increased the least, which fully demonstrated that the AI ​​preservation technology described in this application has a significant effect in inhibiting the increase in acetaldehyde content.

[0123] 8. PPO Activity Assay

[0124] Weigh 1g of winter jujube pulp, add an appropriate amount of 0.05mol / L, pH 6.8 phosphate buffer, add a small amount of quartz sand, grind in an ice bath, fix the volume in a centrifuge tube, and centrifuge at 13000g and 4℃ for 10min. The supernatant is the enzyme solution.

[0125] The total volume of the reaction system is 3 mL, including 1 mL buffer, 1.5 mL 0.02 mol / L catechol, and 0.5 mL enzyme solution. The control tube is added with enzyme solution boiled for 3 minutes. The reaction time is 30 minutes, in a 37°C water bath. When the time is up, add 1 mL 20% trichloroacetic acid to terminate the reaction. Measure the absorbance at 420 nm every 1 second for a total of one minute, with a change of 0.01 absorbance as one unit of activity.

[0126] Calculation formula:

[0127]

[0128] Where: is the absorbance change value, w is the pulp weight (g), and t is the reaction time (min).

[0129] The results showed that the content of polyphenol oxidase (PPO) in winter jujube fruits showed a gradual upward trend during the entire storage period. At each maturity stage, the PPO content of the control group was always higher than that of the experimental group. Specifically, the PPO content of the white jujube in the control group increased by 1.224 (unit: 0.01△A / gFw·min) at the beginning and end of storage; while the PPO content of the white jujube in the experimental group increased by 0.564 (unit: 0.01△A / gFw·min) at the same stage. The PPO content of 1 / 3 red jujube in the control group increased by 0.878 (unit: 0.01△A / gFw·min) at the beginning and end of storage; while the PPO content of 1 / 3 red jujube in the experimental group increased by 0.417 (unit: 0.01△A / gFw·min). Variance analysis showed that there were significant differences in storage effects between the control group and the experimental group during the entire storage period. In summary, the PPO content of the white dates in the control group increased the most during the entire storage process, while the PPO content of the red dates in the experimental group 1 / 3 increased the least, which shows that the AI ​​preservation technology described in the present application has a significant effect in inhibiting the increase of PPO content in winter dates.

[0130] In summary, the AI ​​preservation technology described in this application has outstanding performance in the storage of winter jujube. It not only successfully extends the shelf life of winter jujube, effectively reduces the phenomenon of rotting, weight loss and mildew, but also ensures the hardness and solid content of the fruit.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A winter jujube preservation control method based on IoT and AI, characterized in that: include: The time queue of the preservation parameters in the warehouse is collected by the sensor matrix, and the near-infrared spectrum image of the winter jujube in the warehouse is obtained by the near-infrared spectrometer, wherein the preservation parameters in the warehouse include temperature value, humidity value, oxygen concentration value and carbon dioxide concentration value; Performing multi-modal parameter time series association encoding on the time queue of the in-store fresh-keeping parameter to obtain a multi-modal time series association feature vector of the in-store fresh-keeping parameter; Extracting winter jujube quality features from the near infrared spectrum image of the winter jujube in the warehouse to obtain a near infrared spectrum characterization feature vector for the winter jujube quality; Performing interactive response analysis on the near infrared spectrum characterization feature vector of the winter jujube quality and the multimodal time series correlation feature vector of the storage fresh-keeping parameters to obtain the winter jujube quality-fresh-keeping parameter principal component granularity significant interactive coupling feature vector; Determine whether to generate a fresh-keeping quality warning prompt based on the significant interactive coupling eigenvector of the winter jujube quality-fresh-keeping parameter principal component granularity; Among them, the interactive response analysis of the near-infrared spectrum characterization feature vector of the winter jujube quality and the multimodal time series association feature vector of the in-warehouse preservation parameters is performed, including: performing principal component analysis on the near-infrared spectrum characterization feature vector of the winter jujube quality and the multimodal time series association feature vector of the in-warehouse preservation parameters respectively to obtain a set of principal component feature components of the near-infrared spectrum of the winter jujube quality and a set of principal component feature components of the in-warehouse preservation parameters; performing optimal pairing multi-scale interactive coupling on the set of principal component feature components of the near-infrared spectrum of the winter jujube quality and the set of principal component feature components of the in-warehouse preservation parameters to obtain the principal component granularity significant interactive coupling feature vector of the winter jujube quality-preservation parameters.

2. The method for controlling winter jujube preservation based on IoT and AI according to claim 1, characterized in that: The time queue of the storage fresh-keeping parameter is subjected to multi-modal parameter time series association encoding to obtain a multi-modal time series association feature vector of the storage fresh-keeping parameter, including: Data structuring is performed on the time queue of the in-store fresh-keeping parameters to obtain a multi-modal time-domain joint matrix of the in-store fresh-keeping parameters; The multimodal time-domain joint matrix of the in-warehouse preservation parameters is input into a preservation parameter time-series correlation feature extractor based on a hole convolutional neural network model to obtain a multimodal time-series correlation feature vector of the in-warehouse preservation parameters.

3. The method for controlling winter jujube preservation based on IoT and AI according to claim 2, characterized in that: Extracting winter jujube quality features from the near infrared spectrum image of the winter jujube in the warehouse to obtain a near infrared spectrum characterization feature vector of the winter jujube quality includes: The near infrared spectrum image of the winter jujube in the warehouse is input into a winter jujube quality feature extractor based on a fully connected convolutional neural network model to obtain a near infrared spectrum characterization feature vector of the winter jujube quality.

4. The method for controlling winter jujube preservation based on IoT and AI according to claim 3, characterized in that: The principal component analysis is performed on the near infrared spectrum characterization feature vector of the winter jujube quality to obtain a set of principal component feature components of the near infrared spectrum of the winter jujube quality, including: The covariance matrix of the characteristic vectors of the near-infrared spectrum characterization of the winter jujube quality is calculated to obtain the characteristic covariance matrix of the near-infrared spectrum of the winter jujube quality, and the eigenvalue decomposition of the characteristic covariance matrix of the near-infrared spectrum of the winter jujube quality is performed to obtain a set of principal component eigenvalues ​​of the near-infrared spectrum of the winter jujube quality and a corresponding set of principal component characteristic components of the near-infrared spectrum of the winter jujube quality.

5. The method for controlling winter jujube preservation based on IoT and AI according to claim 4, characterized in that: The optimal pairing multi-scale interactive coupling is performed on the set of principal component characteristic components of the near infrared spectrum of the winter jujube quality and the set of principal component characteristic components of the storage parameter to obtain the principal component granularity significant interactive coupling characteristic vector of the winter jujube quality-preservation parameter, including: Taking each principal component characteristic component of the near infrared spectrum of winter jujube quality in the set of principal component characteristic components of the near infrared spectrum of winter jujube quality as a query vector, and taking the set of principal component characteristic components of the in-store fresh-keeping parameters as a query library, matching the principal component characteristic components of the in-store fresh-keeping parameters that best match each of the query vectors from the query library to obtain a set of best matching pairs of {principal component characteristic components of the near infrared spectrum of winter jujube quality; principal component characteristic components of the in-store fresh-keeping parameters}; The set of the best matching pairs of {near-infrared spectrum principal component characteristic components of winter jujube quality; principal component characteristic components of storage parameters} is subjected to multi-scale interactive fusion to obtain the principal component granularity significant interactive coupling characteristic vector of winter jujube quality-preservation parameters.

6. The method for controlling winter jujube preservation based on IoT and AI according to claim 5, characterized in that: Match the main component characteristic components of the fresh-keeping parameters in the library that best match each of the query vectors from the query library to obtain {main component characteristic components of the near infrared spectrum of winter jujube quality; The set of the best matching pairs of the main component characteristic components of the storage parameters in the library includes: Calculating the Mahalanobis distance between the query vector and the characteristic components of the main components of the fresh-keeping parameters in each library in the query library to obtain a set of winter jujube quality-fresh-keeping parameter main component matching difference coefficients; Select the in-store fresh-keeping parameter principal component characteristic component corresponding to the minimum value in the set of the winter jujube quality-fresh-keeping parameter principal component matching difference coefficients as the optimal matching result of the query vector to obtain the {winter jujube quality near-infrared spectrum principal component characteristic component; The best matching pair of the principal component characteristic component of the storage parameter.

7. The method for controlling winter jujube preservation based on IoT and AI according to claim 6, characterized in that: The multi-scale interactive fusion is performed on the set of the best matching pairs of {near infrared spectrum principal component characteristic components of winter jujube quality; principal component characteristic components of storage parameters} to obtain the principal component granularity significant interactive coupling characteristic vector of winter jujube quality-preservation parameters, including: Input each best matching pair of {principal component characteristic component of near infrared spectrum of winter jujube quality; principal component characteristic component of in-store fresh-keeping parameter} in the set of best matching pairs of {principal component characteristic component of near infrared spectrum of winter jujube quality; principal component characteristic component of in-store fresh-keeping parameter} into the multi-scale interactive response coupling module to obtain a set of multi-scale interactive coupling representation vectors of best matching pairs of winter jujube quality and fresh-keeping parameter; The set of multi-scale interactive coupling representation vectors of the best matching pair of winter jujube quality-fresh-keeping parameters is cascaded to obtain the principal component granularity significant interactive coupling feature vector of the winter jujube quality-fresh-keeping parameters.

8. The method for controlling winter jujube preservation based on IoT and AI according to claim 7, characterized in that: The best matching pairs of {principal component characteristic components of near infrared spectrum of winter jujube quality; principal component characteristic components of in-warehouse preservation parameters} in the set of the best matching pairs of {principal component characteristic components of near infrared spectrum of winter jujube quality; principal component characteristic components of in-warehouse preservation parameters} are respectively input into the multi-scale interactive response coupling module to obtain a set of multi-scale interactive coupling representation vectors of the best matching pairs of winter jujube quality and preservation parameters, including: Respectively calculate the position point addition, position point subtraction and position point multiplication between the principal component characteristic components of the near infrared spectrum of winter jujube quality and the principal component characteristic components of the in-store fresh-keeping parameters in the best matching pair of {principal component characteristic components of winter jujube quality near infrared spectrum; principal component characteristic components of in-store fresh-keeping parameters} to obtain a first best matching interaction representation vector, a second best matching interaction representation vector and a third best matching interaction representation vector; The first best matching interaction representation vector, the second best matching interaction representation vector and the third best matching interaction representation vector are cascaded and fused, and then input into a one-dimensional convolutional layer including a maximum pooling layer to obtain a multi-scale interaction representation vector of the best matching of the principal components of the winter jujube quality-keeping parameter; The second norm of the principal component characteristic component of the near-infrared spectrum of the winter jujube quality, the second norm of the principal component characteristic component of the in-warehouse preservation parameter and the maximum value of the preset scale adjustment parameter are selected as the scale scaling factor, and each eigenvalue in the multi-scale interaction representation vector of the best match of the winter jujube quality-preservation parameter principal component is divided by the scale scaling factor to obtain the multi-scale interaction coupling representation vector of the best match of the winter jujube quality-preservation parameter.

9. The method for controlling winter jujube preservation based on IoT and AI according to claim 8, characterized in that: Based on the winter jujube quality-preservation parameter principal component granularity significant interactive coupling eigenvector, determining whether to generate a fresh-keeping quality warning prompt includes: Inputting the winter jujube quality-preservation parameter principal component granularity significant interactive coupling feature vector into a decoder-based preservation time prediction module to obtain a freshness-preservation time prediction decoding value; Based on the comparison between the predicted decoded value of the fresh-keeping time and a preset threshold, it is determined whether to generate the fresh-keeping quality warning prompt.

10. A winter jujube preservation control system based on IoT and AI, characterized in that: include: A data acquisition module is used to collect the time queue of the preservation parameters in the warehouse through the sensor matrix, and simultaneously obtain the near-infrared spectrum image of the winter jujube in the warehouse collected by the near-infrared spectrometer, wherein the preservation parameters in the warehouse include temperature value, humidity value, oxygen concentration value and carbon dioxide concentration value; A multimodal parameter time series association coding module is used to perform multimodal parameter time series association coding on the time queue of the storage storage parameter to obtain a multimodal time series association feature vector of the storage storage parameter; A winter jujube quality feature extraction module is used to extract the winter jujube quality features from the near infrared spectrum image of the winter jujube in the warehouse to obtain a winter jujube quality near infrared spectrum characterization feature vector; An interactive response analysis module is used to perform interactive response analysis on the near-infrared spectrum characterization feature vector of the winter jujube quality and the multimodal time series correlation feature vector of the storage fresh-keeping parameters to obtain a significant interactive coupling feature vector of the winter jujube quality-fresh-keeping parameter principal component granularity; The fresh-keeping quality warning module is used to determine whether to generate a fresh-keeping quality warning prompt based on the significant interactive coupling eigenvector of the winter jujube quality-fresh-keeping parameter principal component granularity.

Images