Adaptive control method and system for air drying equipment

Abstract

The present invention discloses an adaptive control method and system for air-drying equipment, which relates to the field of drying processing technology. The method includes: performing surface detection on the food input into the air-drying equipment, obtaining the surface bacterial strain layer, and judging whether the bacterial strain activity needs to be maintained according to the food type and the preset bacterial strain activity requirements. If the bacterial strain activity needs to be maintained, the air-drying equipment is controlled to enter the activity retention air-drying mode to obtain a first control parameter combination; if the bacterial strain activity does not need to be maintained, the air-drying equipment is controlled to enter the stable drying air-drying mode to obtain a second control parameter combination. The present invention solves the technical problem in the prior art that the air-drying equipment cannot be adaptively controlled according to the specific bacterial strain activity requirements of the food, resulting in insufficient bacterial strain activity protection and unstable food quality. It achieves the technical effect of accurately controlling bacterial strain activity and improving food quality by intelligently judging and adjusting the working mode and parameter combination of the air-drying equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of drying processing, and in particular to an adaptive control method and system for air drying equipment. Background Art

[0002] In traditional air-dried food production, air-drying equipment typically operates using fixed temperature, humidity, and wind speed parameters, lacking precise adjustment for different foods and bacterial layers. This standardized air-drying method fails to effectively account for food type and bacterial activity requirements, resulting in the loss of bacterial activity in some air-dried foods (such as air-dried sausages and fermented cheeses) during the drying process, affecting their flavor, shelf life, and nutritional value. Furthermore, existing equipment often relies on empirically defined drying conditions and is unable to dynamically adjust operating parameters based on real-time food needs, resulting in wasted resources and impacting production efficiency and product quality. Summary of the Invention

[0003] The present application provides an adaptive control method and system for air-drying equipment, which is used to solve the technical problem in the prior art that air-drying equipment cannot be adaptively controlled according to the activity requirements of specific food strains, resulting in insufficient protection of strain activity and unstable food quality.

[0004] The first aspect of the present application provides an adaptive control method for air-drying equipment, the method comprising: performing surface detection on food input into the air-drying equipment to obtain a surface bacterial layer; identifying a preset bacterial activity requirement for the surface bacterial layer of the food type, and judging whether the food needs to maintain bacterial activity based on the preset bacterial activity requirement; if the food needs to maintain bacterial activity, controlling the air-drying equipment to enter an activity-retention air-drying mode, and obtaining a first control parameter combination for controlling the air-drying equipment based on the activity-retention air-drying mode with the preset bacterial activity requirement as an adaptive target; if the food does not need to maintain bacterial activity, controlling the air-drying equipment to enter a stable drying air-drying mode, and obtaining a second control parameter combination for controlling the air-drying equipment based on the stable drying air-drying mode.

[0005] The second aspect of the present application provides an adaptive control system for air-drying equipment, the system comprising: a food surface detection module, the food surface detection module being used to perform surface detection on food input into the air-drying equipment to obtain a surface bacterial layer; an activity requirement judgment module, the activity requirement judgment module being used to identify the preset bacterial activity requirements of the food type regarding the surface bacterial layer, and judging whether the food needs to maintain bacterial activity based on the preset bacterial activity requirements; an activity retention air-drying module, the activity retention air-drying module being used to control the air-drying equipment to enter an activity retention air-drying mode if the food needs to maintain bacterial activity, and obtaining a first control parameter combination for controlling the air-drying equipment based on the activity retention air-drying mode and the preset bacterial activity requirements as an adaptive target; and a stable drying air-drying module, the stable drying air-drying module being used to control the air-drying equipment to enter a stable drying air-drying mode if the food does not need to maintain bacterial activity, and obtaining a second control parameter combination for controlling the air-drying equipment based on the stable drying air-drying mode.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] The adaptive control method and system for air-drying equipment provided in this application relates to the field of drying processing technology. By detecting the bacterial layer on the surface of food and combining it with the preset bacterial activity requirements, it is intelligently judged whether the food needs to maintain bacterial activity. According to the judgment result, the working mode of the air-drying equipment is adjusted to enter the activity retention or stable drying air-drying mode, and the corresponding control parameter combination is adjusted to ensure the precise control of the air-drying process. It solves the technical problem in the prior art that the air-drying equipment cannot be adaptively controlled according to the activity requirements of specific bacterial strains in food, resulting in insufficient bacterial activity protection and unstable food quality. It achieves the technical effect of accurately controlling bacterial activity and improving food quality by intelligently judging and adjusting the working mode and parameter combination of the air-drying equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0009] Figure 1 A schematic flow chart of the adaptive control method for air-drying equipment provided in an embodiment of the present application;

[0010] Figure 2 Schematic diagram of the structure of the adaptive control system for air-drying equipment provided in an embodiment of the present application.

[0011] Description of the accompanying drawings: food surface detection module 11, activity demand judgment module 12, activity retention air-drying module 13, stable drying air-drying module 14. DETAILED DESCRIPTION

[0012] The present application provides an adaptive control method and system for air-drying equipment, which is used to solve the technical problem in the prior art that air-drying equipment cannot be adaptively controlled according to the activity requirements of specific food strains, resulting in insufficient protection of strain activity and unstable food quality.

[0013] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0014] It should be noted that the terms "first", "second", etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices.

[0015] Example 1, as Figure 1 As shown, the present application provides an adaptive control method for an air drying device, the method comprising:

[0016] P10: Perform surface inspection on food entering the air-drying device to obtain the surface bacterial layer. The air-drying device includes an edge computing chip embedded with a trained bacterial layer label recognition model.

[0017] Furthermore, step P10 in the embodiment of the present application further includes:

[0018] P11: A surface detection module is integrated in the air-drying equipment, and the surface detection module includes a visible light camera and a multispectral imaging component; P12: The surface image of the food is obtained according to the visible light camera; P13: The surface material reflection spectrum image of the food is obtained according to the multispectral imaging component; P14: Feature extraction is performed on the surface image and the surface material reflection spectrum image to obtain surface image convolution features and surface material convolution features; P15: The surface image convolution features and surface material convolution features are input as a combined vector into a strain layer label recognition model for analysis to obtain the surface strain layer of the food.

[0019] It should be understood that the surface of food entering the air-drying device is tested to obtain data on the surface bacterial layer, providing a basis for regulating bacterial activity during the subsequent air-drying process. To achieve this goal, the air-drying device integrates an edge computing chip with a trained bacterial layer label recognition model. Using this model, the device can quickly and accurately analyze the food surface, identifying and extracting relevant information about the surface bacterial layer.

[0020] Specifically, first, the air-drying equipment has an integrated surface detection module, which includes a visible light camera and a multispectral imaging component. The main function of the visible light camera is to obtain conventional images of the food surface and capture basic visual information such as the color, texture, and structure of the food surface. This information helps to understand the overall state and surface characteristics of the food. In contrast, the multispectral imaging component uses light of different wavelengths to image the food surface and obtain a reflective spectrum image. The reflective spectrum image can reveal the reflective characteristics of the food surface in multiple bands, especially highlighting the subtle changes in the bacterial layer. Different materials and bacterial layers on the food surface reflect light differently. Therefore, through multispectral imaging, it is possible to deeply analyze the surface details and obtain characteristic information related to the bacterial layer.

[0021] When the surface inspection module begins operation, the visible light camera in the air-drying equipment first captures images of the food surface. These images are standard two-dimensional color images, showing the general outline and characteristics of the food surface. Next, the multispectral imaging component acquires reflectance spectral images of the surface material using different spectral bands. Variations in the reflected light intensity in each spectral band reflect the physical properties of the surface material, particularly the microstructure of the bacterial layer. Analysis of these reflectance spectral images provides detailed information about the bacterial layer on the food surface.

[0022] Next, feature extraction is performed on the acquired surface image and surface material reflectance spectrum image. The surface image is processed using image processing techniques such as convolutional neural networks (CNNs) to extract surface image convolution features. This convolution operation automatically identifies key information in the image, such as edges, texture, and shape. These features can reveal surface details and textural variations in the food. Similarly, the reflectance spectrum image is processed using a convolutional neural network, but with a greater focus on extracting features related to the chemical composition and microstructure of the food surface from the spectral information. By extracting these two features, the system can comprehensively understand the characteristics of the food surface from both macroscopic and microscopic levels.

[0023] Finally, the extracted surface image convolution features and surface material convolution features are combined into a vector and input into the bacterial layer label recognition model for analysis. The bacterial layer label recognition model is a trained deep learning model that can identify the type of bacterial layer on the food surface based on the input feature vector. During training, the model has learned the characteristics of a large number of different bacterial layers. Therefore, it can accurately determine the presence of a specific bacterial layer on the food surface and identify its type and state, providing precise data support for subsequent control of the air-drying process.

[0024] Through the implementation of the above steps, the system can efficiently and accurately obtain the bacterial layer information on the food surface, providing important data support for the subsequent air-drying process control.

[0025] Furthermore, the strain layer label recognition model is trained. Step P10 of the embodiment of the present application further includes:

[0026] P15-1a: Collect a set of surface image samples, which includes surface image samples based on different types of food with known species layer labels and surface material reflectance spectrum image samples collected simultaneously; P15-2a: Extract texture feature vectors of the surface image samples in the surface image sample set, and output them as surface image convolution feature samples; P15-3a: Extract reflectance spectrum vectors of the surface material reflectance spectrum image samples in the surface image sample set, and output them as surface material convolution feature samples; P15-4a: Perform lightweight MLP classifier training based on the surface image convolution feature samples, the surface material convolution feature samples and the known species layer labels, until the accuracy of the lightweight MLP classifier is greater than a preset threshold, thereby obtaining a species layer label recognition model.

[0027] Optionally, the process of training the bacterial layer label recognition model can be further refined by collecting and processing surface image samples and reflectance spectrum image samples from multiple angles to ensure that the model can efficiently and accurately identify the characteristics of the bacterial layer on the food surface.

[0028] First, a surface image sample set is collected. This sample set includes surface image samples based on different food types and simultaneously collected surface material reflectance spectral image samples. These samples are labeled with known bacterial species for subsequent model training. Surface image samples can reflect macroscopic characteristics of the food surface, such as color and texture, while surface material reflectance spectral image samples provide information on the microscopic chemical composition. The combination of the two provides a rich data foundation for model training.

[0029] Next, texture features are extracted from each surface image sample in the collected surface image sample set, outputting a surface image convolution feature sample. Texture features are a key feature in image analysis, reflecting the regularity and pattern of pixel arrangement within an image. Using convolutional neural networks (CNNs) or other advanced image processing techniques, representative texture feature vectors can be extracted from surface images, capturing the texture details of the food surface and providing key information for subsequent model training.

[0030] A similar process is also applied to surface material reflectance spectrum image samples. By extracting a reflectance spectrum vector from each reflectance spectrum image sample, the light reflectance characteristics of the surface material at different wavelengths are determined. Each reflectance spectrum vector represents the degree of reflectance of the food surface material at a specific wavelength. These reflectance spectrum vectors can reveal the light reflectance characteristics of different types of bacterial layers, helping to identify the presence and status of the bacterial layer.

[0031] Subsequently, a lightweight multi-layer perceptron (MLP) classifier is trained based on the extracted surface image and material convolution feature samples, combined with known bacterial layer labels. The lightweight MLP classifier is an efficient neural network classification model that enables fast training and accurate classification even with limited computing resources. During training, the model continuously adjusts its parameters to minimize the error between its predictions and the known bacterial layer labels. Training continues until the accuracy of the lightweight MLP classifier reaches or exceeds a preset threshold, which is set based on actual application requirements to ensure the model's reliability and accuracy in real-world use. Ultimately, the rigorously trained lightweight MLP classifier is selected as the bacterial layer label recognition model for subsequent bacterial layer recognition tasks on food surfaces in air-drying equipment. This model can be embedded in the air-drying equipment's edge computing chip to analyze the food surface in real time, ensuring proper control of bacterial activity during the drying process and providing the air-drying equipment with real-time, efficient bacterial layer recognition and control capabilities.

[0032] Furthermore, step P15-4a of the embodiment of the present application further includes:

[0033] P15-41a: Train the twin network, feedback-connect the lightweight MLP classifier to the twin network, and feedback-connect the output end of the twin network to the input end of the lightweight MLP classifier; P15-42a: Construct positive sample pairs and negative sample pairs, the samples of the positive sample pair input into the first input end and the second input end of the twin network are samples of the same category of species layer, and the samples of the negative sample pair input into the first input end and the second input end of the twin network are samples of different categories of species layer; P15-43a: Use the comparison loss function to train the positive sample pairs and the negative sample pairs to obtain similarity score samples; P15-44a: Feedback optimize the model of the lightweight MLP classifier based on the similarity score samples, and output the optimized species layer label recognition model.

[0034] In a possible embodiment of the present application, the accuracy of the species layer label recognition model can be improved by further introducing a twin network for more refined training and optimization, especially in the recognition of subtle differences between species layers.

[0035] First, a twin network is trained and fed back into a lightweight MLP classifier. The twin network consists of two neural networks with shared parameters, used to compare the similarity of two sets of input data. Specifically, the output of the twin network is fed back into the input of the lightweight MLP classifier. This connection allows the twin network to assist in optimizing the lightweight MLP classifier's decision-making process, thereby improving the classifier's recognition accuracy at the species level.

[0036] Next, construct positive and negative sample pairs. Positive pairs are composed of samples from the same bacterial species layer that are fed into the first and second inputs of the twin network. These samples are highly similar in features. Negative pairs, on the other hand, are composed of samples from different bacterial species layers, with significant differences between them. By designing these positive and negative pairs, we can clearly distinguish between similar and different samples, thereby enhancing the model's sensitivity to subtle differences between bacterial species layers during classification.

[0037] Subsequently, the constructed positive and negative sample pairs are trained using a comparison loss function. This comparison loss function is primarily used to train the Siamese network. By calculating the similarity scores between positive and negative sample pairs, it aims to achieve higher similarity scores for samples of the same type and lower similarity scores for samples of different types. Through training with the comparison loss function, the system learns how to efficiently distinguish subtle differences between samples at different species levels, further optimizing its ability to identify species-level labels.

[0038] Finally, based on the obtained similarity score samples, the lightweight MLP classifier model undergoes feedback optimization. Through further training with positive and negative samples, the optimization process effectively increases the discriminability between labeled classification results, especially for subtle differences at the species level. The goal of feedback optimization is to reduce the false positive rate and improve the model's ability to distinguish species-level features, especially for subtle differences such as activity variations or microbial community characteristics across different species layers on the same food surface. The optimized model enables more precise segmentation and classification, adapting to more complex real-world application scenarios.

[0039] Ultimately, feedback optimization resulted in a more accurate strain-layer label recognition model with enhanced classification performance, especially when dealing with subtle strain-layer differences, reducing misidentification and improving accuracy. This enables air-drying equipment to fine-tune strain-layer activity in practical applications, resulting in more efficient food drying and strain-protection processes.

[0040] P20: Identify the preset bacterial activity requirement of the food type with respect to the surface bacterial layer, and determine whether the food needs to maintain bacterial activity based on the preset bacterial activity requirement.

[0041] Furthermore, step P20 in this embodiment of the present application further includes:

[0042] P21: Construct a bacterial strain activity requirement rule table to identify the preset bacterial strain activity requirements of the food type regarding the surface bacterial strain layer; wherein the storage structure of the bacterial strain activity requirement rule table includes a ternary storage field, and the storage field includes the food type, the bacterial strain layer type, and the activity requirement index.

[0043] Specifically, based on the identified food type and the type of bacterial layer on its surface, it is determined whether the bacterial strains need to be maintained active during the air-drying process. This process relies on constructing a bacterial strain activity requirement rule table, which provides a corresponding activity requirement index for each food type and bacterial layer type combination. This rule table is stored in a ternary storage field structure, which contains three fields: food type, bacterial layer type, and activity requirement index. The food type field lists all possible food types, the bacterial layer type field describes the types of bacterial strains that may be attached to the surface of these foods, and the activity requirement index defines the environmental conditions required for these foods during the air-drying process to maintain or optimize bacterial strain activity.

[0044] The process of building a rule table first requires identifying and listing the various food types and, for each type, determining the corresponding culture layer type. For example, air-dried sausages may be covered with white mold (white_mold), while fermented cheese may be covered with red mold (red_mold). In this way, the rule table can specify specific activity requirements for each food type and culture layer type combination. Activity requirements may include parameters such as temperature, humidity, and wind speed, with specific values ​​set based on the characteristics of the food and the culture layer. For foods that require the survival of the culture, the activity requirements will specify low temperatures, appropriate humidity, and moderate wind speeds during the drying process to avoid inactivation. For foods that do not require the survival of the culture, higher temperatures and wind speeds may be set to facilitate the drying process.

[0045] Once the bacterial strain activity requirement rule table is constructed, the system will use the table for real-time query during the air-drying process. When the food enters the air-drying equipment, the system will automatically identify the type of food and the type of surface bacterial layer, and obtain the activity requirement index of the combination according to the rule table. If the rule table shows that the combination of food type and bacterial layer type needs to maintain bacterial strain activity, it will enter a special activity-preserving air-drying mode. In this mode, the air-drying equipment will adjust control parameters such as temperature, humidity, and wind speed according to the preset activity requirement indicators to ensure that the activity of the bacterial strain is not destroyed. Conversely, if the combination does not need to maintain bacterial strain activity, the system will enter a stable drying mode. In this mode, the equipment focuses on rapid drying and no longer pays special attention to retaining bacterial strain activity. Parameters such as temperature and wind speed will be moderately increased to accelerate the drying process.

[0046] This method intelligently adjusts the air-drying process's environmental conditions based on the specific needs of each food product and bacterial layer, optimizing the drying process and ensuring effective control of both food flavor and bacterial activity. The use of this rule table not only enhances system automation and flexibility but also reduces manual intervention during production, improving efficiency and product quality.

[0047] Furthermore, whether the food needs to maintain bacterial activity is determined based on the preset bacterial activity requirement. Step P20 of the embodiment of the present application further includes:

[0048] P22: Based on the bacterial strain activity requirement rule table, a standard air-drying data sample group is collected for each food type and bacterial strain layer type, wherein each group of standard air-drying data samples includes surface activity detection samples under multiple air-drying stages, and the multiple air-drying stages include at least before air-drying and after air-drying; P23: For each group of standard air-drying data samples, the activity detection index difference between the surface activity detection index before air-drying and the surface activity detection index after air-drying is recorded; P24: If the activity detection index difference is less than or equal to the preset activity detection index, the food needs to maintain bacterial strain activity; if the activity detection index difference is greater than the preset activity detection index, the food needs to maintain bacterial strain activity.

[0049] It should be understood that in order to further accurately determine whether food needs to maintain bacterial activity based on preset bacterial activity requirements, a standard air-dried data sample group can be used to make a more accurate judgment on whether food needs to maintain bacterial activity.

[0050] First, based on the previously constructed bacterial strain activity requirement rule table, a standard air-drying data sample group is collected for each food type and bacterial strain layer type. Each set of standard air-drying data samples includes surface activity test samples at multiple different air-drying stages to ensure coverage of different time points in the air-drying process. These air-drying stages include at least two key points: before air-drying and after air-drying. Before air-drying, the activity of the bacterial strains on the surface of the food is in its initial state; after air-drying, the activity of the bacterial strains may be affected by factors such as temperature, humidity, and wind speed, and may change. By collecting activity test samples at these two stages, the impact of the air-drying process on bacterial strain activity can be effectively evaluated.

[0051] Next, for each set of standard air-drying data samples, the difference in activity detection indicators between the surface activity detection indicators before air-drying and the surface activity detection indicators after air-drying is recorded and calculated. Activity detection indicators may include a series of indicators such as the survival rate of the strain, the concentration of active substances, and the metabolic rate. By comparing the data difference before and after air-drying, the impact of the air-drying process on the activity of the strain can be quantified. For example, if the difference is small, it means that the air-drying process has no significant effect on the activity of the strain, and the activity of the strain is maintained; if the difference is large, it means that the air-drying process may have damaged the activity of the strain, and the system needs to make corresponding adjustments.

[0052] Next, the calculated activity detection index difference is compared with the preset activity detection index. If the difference is less than or equal to the preset activity detection index threshold, it indicates that the bacterial activity of the food has not been significantly affected and the food does not need to be specifically maintained, so the food can enter the normal drying mode. If the difference is greater than the preset activity detection index threshold, it indicates that the bacterial activity has undergone significant changes during the air-drying process and the food needs to maintain bacterial activity. At this time, the system will switch to the air-drying mode that maintains bacterial activity.

[0053] Through this process, we can not only determine whether the food needs to maintain bacterial strain activity based on the bacterial strain activity requirement rule table, but also verify and adjust this judgment through actual air-drying data to avoid the loss of bacterial strain activity due to inappropriate air-drying conditions, thereby effectively controlling the flavor, shelf life and fermentation effect of the food.

[0054] P30: If the food needs to maintain the activity of the bacteria, the air-drying equipment is controlled to enter the activity-retaining air-drying mode, and the first control parameter combination for controlling the air-drying equipment is obtained according to the activity-retaining air-drying mode with the preset bacteria activity requirement as the adaptive target.

[0055] Furthermore, step P30 in the embodiment of the present application further includes:

[0056] P31: Obtain an initial air-drying control parameter combination for air-drying the food, the initial air-drying control parameter combination including air-drying temperature, air-drying humidity, air-drying speed, and wind direction change frequency; P32: Based on the preset bacterial strain activity requirement, obtain the activity tolerance parameter threshold of the surface bacterial strain layer based on each control parameter in the initial air-drying control parameter combination; P33: Constrain the initial air-drying control parameter combination according to the activity tolerance parameter threshold to obtain a first control parameter combination for controlling the air-drying equipment.

[0057] Optionally, if the food needs to maintain bacterial activity, the air-drying equipment will enter the activity-preserving air-drying mode. This mode is adaptive to the preset bacterial activity requirement to ensure that the activity of the bacterial layer on the food surface is effectively maintained during the air-drying process.

[0058] First, the initial air-drying control parameter combination for the food needs to be determined. This combination includes the basic control parameters required for the air-drying process, such as air-drying temperature, air-drying humidity, air-drying speed, and the frequency of air direction changes. These parameters are essential factors in the air-drying process, directly determining the stability of the air-drying conditions and the activity of the bacterial layer on the food surface. The initial control parameter combination is usually preset based on standard air-drying requirements or the basic characteristics of the food, but further optimization is required based on the desired bacterial activity.

[0059] Next, based on the preset strain activity requirements, activity tolerance parameter thresholds are calculated. These thresholds establish a tolerance range for each air-drying control parameter (such as temperature, humidity, and wind speed). Activity tolerance refers to the range within which a control parameter can vary while maintaining strain activity. For example, a strain may remain active within a certain temperature range, but temperatures exceeding or falling below this range will cause the strain to become inactive. Based on the preset strain activity requirements, reasonable tolerance thresholds can be set for each control parameter to ensure that control conditions during the air-drying process remain within the range suitable for strain activity.

[0060] Next, the initial air-drying control parameter combination is constrained based on the activity tolerance parameter threshold, automatically adjusting the initial air-drying control parameters to ensure they are within the specified tolerance range. For example, if the air-drying temperature slightly exceeds the preset activity tolerance, the system automatically adjusts the temperature back to an acceptable range to avoid damaging the activity of the bacterial strain. After the constraint processing, the resulting first control parameter combination will serve as the final parameter for controlling the air-drying equipment, ensuring that the air-drying equipment operates according to optimized conditions, thereby effectively preserving the activity of the bacterial strains on the food surface.

[0061] Through the above steps, the air-drying equipment can automatically adjust its operating parameters in activity-preserving air-drying mode based on the specific bacterial activity requirements of the food. This adaptive control not only preserves the flavor and texture of the food, but also ensures its safety and nutritional value, thereby improving the overall quality of the air-dried food. Furthermore, this method is flexible and versatile, and can be widely applied to various food drying processes that require preserving bacterial activity.

[0062] P40: If the food does not need to maintain bacterial activity, the air-drying device is controlled to enter a stable drying mode, and a second control parameter combination for controlling the air-drying device is obtained according to the stable drying mode.

[0063] Furthermore, step P40 in the embodiment of the present application further includes:

[0064] P41: Performing phased stable adaptive processing on the initial air-drying control parameter combination and outputting a multi-stage air-drying control parameter combination; P42: Outputting the multi-stage air-drying control parameter combination as the second regulating parameter combination of the air-drying device.

[0065] It should be understood that when the food does not need to maintain bacterial activity, the air-drying equipment is controlled to enter the stable drying mode and a second control parameter combination for controlling the air-drying equipment is obtained. This step mainly focuses on how to optimize the air-drying process when maintaining bacterial activity is not required to improve air-drying efficiency and product quality.

[0066] First, the initial air-drying control parameter combination is subjected to stage-by-stage stable adaptive processing. The initial air-drying control parameter combination includes basic parameters such as temperature, humidity, and wind speed. During the air-drying process, food may go through multiple air-drying stages, and the requirements of each stage may be different. For example, a higher temperature may be required in the initial stage to accelerate water evaporation, while a lower temperature and higher humidity may be required in the later stage to prevent over-drying. In this step, the system will dynamically adjust the control parameters according to the needs of different stages to adapt them to the changing requirements of each stage. Through staged processing, the system can refine the adjustment of the air-drying control parameters according to the actual air-drying process, and output multiple air-drying control parameter combinations, each combination corresponding to a specific stage.

[0067] The resulting multi-stage air-drying control parameter combinations are then output as the final second control parameter combinations. These multi-stage parameter combinations cover all stages of the air-drying process, adjusting factors such as temperature, humidity, and air speed according to the actual needs of each stage. The final second control parameter combinations serve as the control parameters for the air-drying equipment, enabling the equipment to operate in a stable drying mode and continuously optimize control conditions at different stages.

[0068] This process intelligently adjusts the operating parameters of the drying equipment based on the varying needs of food types and the varying characteristics of the drying process, thereby improving drying efficiency, reducing resource waste, and ensuring food quality. Without the need to maintain bacterial activity, this effectively improves drying efficiency and food quality, achieving the desired drying effect.

[0069] In summary, the embodiments of the present application have at least the following technical effects:

[0070] This application intelligently identifies food types and bacterial layer activity requirements, dynamically adjusts the working mode and control parameters of the air-drying equipment to ensure that bacterial activity is effectively maintained or reasonably controlled; adjusts parameters such as temperature, humidity, and wind speed according to the specific needs of food and bacterial strains, thereby optimizing the flavor, shelf life, and fermentation effect of the food; reduces resource waste in traditional air-drying methods by precisely adjusting the operation of the air-drying equipment, and improves energy utilization efficiency and production benefits; automatically adjusts air-drying conditions based on real-time data, reduces manual operations, and improves the level of automation in the production process.

[0071] The technical effect of accurately controlling the activity of bacteria and improving food quality is achieved by intelligently judging and adjusting the working mode and parameter combination of the air-drying equipment.

[0072] Example 2, based on the same inventive concept as the adaptive control method for air drying equipment in the above embodiment, Figure 2As shown, the present application provides an adaptive control system for air drying equipment. The system and method embodiments in the present application are based on the same inventive concept. The system includes:

[0073] The food surface detection module 11 is used to perform surface detection on the food input into the air-drying device to obtain the surface bacterial layer. The air-drying device includes an edge computing chip embedded with a trained bacterial layer label recognition model.

[0074] The activity requirement judgment module 12 is used to identify the preset bacterial activity requirement of the food type with respect to the surface bacterial layer, and judge whether the food needs to maintain bacterial activity according to the preset bacterial activity requirement.

[0075] The activity-retaining air-drying module 13 is used to control the air-drying equipment to enter the activity-retaining air-drying mode if the food needs to maintain the activity of the bacteria, and obtain the first control parameter combination for controlling the air-drying equipment according to the activity-retaining air-drying mode with the preset bacteria activity requirement as the adaptive target.

[0076] The stable drying and air-drying module 14 is used to control the air-drying equipment to enter the stable drying and air-drying mode if the food does not need to maintain the activity of the bacteria, and then obtain the second control parameter combination for controlling the air-drying equipment according to the stable drying and air-drying mode.

[0077] Furthermore, the food surface detection module 11 is further configured to perform the following steps:

[0078] The food input into the air-drying device is subjected to surface inspection to obtain a surface bacterial species layer, wherein the air-drying device is integrated with a surface inspection module, and the surface inspection module includes a visible light camera and a multispectral imaging component; a surface image of the food is acquired according to the visible light camera; a surface material reflectance spectrum image of the food is acquired according to the multispectral imaging component; feature extraction is performed on the surface image and the surface material reflectance spectrum image to obtain surface image convolution features and surface material convolution features; the surface image convolution features and surface material convolution features are input as a combined vector into a bacterial species layer label recognition model for analysis to obtain the surface bacterial species layer of the food.

[0079] Furthermore, the food surface detection module 11 is further configured to perform the following steps:

[0080] A surface image sample set is collected, wherein the surface image sample set includes surface image samples of different food types with known strain layer labels and synchronously collected surface material reflectance spectrum image samples; texture feature vectors of the surface image samples in the surface image sample set are extracted and output as surface image convolution feature samples; reflectance spectrum vectors of the surface material reflectance spectrum image samples in the surface image sample set are extracted and output as surface material convolution feature samples; a lightweight MLP classifier is trained based on the surface image convolution feature samples, the surface material convolution feature samples and the known strain layer labels until the accuracy of the lightweight MLP classifier is greater than a preset threshold, thereby obtaining a strain layer label recognition model.

[0081] Furthermore, the food surface detection module 11 is further configured to perform the following steps:

[0082] Train the twin network, feedback-connect the lightweight MLP classifier to the twin network, and feedback-connect the output end of the twin network to the input end of the lightweight MLP classifier; construct positive sample pairs and negative sample pairs, the samples of the positive sample pair input to the first input end and the second input end of the twin network are samples of the same category of species layer, and the samples of the negative sample pair input to the first input end and the second input end of the twin network are samples of different categories of species layer; use the comparison loss function to train the positive sample pairs and the negative sample pairs to obtain similarity score samples; feedback optimize the model of the lightweight MLP classifier based on the similarity score samples, and output the optimized species layer label recognition model.

[0083] Furthermore, the activity demand determination module 12 is further configured to perform the following steps:

[0084] A bacterial strain activity requirement rule table is constructed to identify the preset bacterial strain activity requirements of the food type regarding the surface bacterial strain layer; wherein the storage structure of the bacterial strain activity requirement rule table includes a ternary storage field, and the storage field includes the food type, the bacterial strain layer type, and the activity requirement index.

[0085] Furthermore, the activity demand determination module 12 is further configured to perform the following steps:

[0086] Based on the bacterial strain activity requirement rule table, a standard air-drying data sample group is collected for each food type and bacterial strain layer type, wherein each group of standard air-drying data samples includes surface activity detection samples under multiple air-drying stages, and the multiple air-drying stages include at least before air-drying and after air-drying; for each group of standard air-drying data samples, the activity detection index difference between the surface activity detection index before air-drying and the surface activity detection index after air-drying is recorded; if the activity detection index difference is less than or equal to the preset activity detection index, the food needs to maintain bacterial strain activity; if the activity detection index difference is greater than the preset activity detection index, the food needs to maintain bacterial strain activity.

[0087] Furthermore, the active retention air-drying module 13 is further configured to perform the following steps:

[0088] An initial air-drying control parameter combination for air-drying the food is obtained, wherein the initial air-drying control parameter combination includes air-drying temperature, air-drying humidity, and air-drying speed and wind direction change frequency; based on the preset bacterial strain activity requirement, an activity tolerance parameter threshold of the surface bacterial strain layer based on each control parameter in the initial air-drying control parameter combination is obtained; the initial air-drying control parameter combination is constrained according to the activity tolerance parameter threshold to obtain a first control parameter combination for controlling the air-drying equipment.

[0089] Furthermore, the stable drying and air-drying module 14 is further configured to perform the following steps:

[0090] The initial air-drying control parameter combination is subjected to staged stable adaptive processing to output a multi-stage air-drying control parameter combination; and the multi-stage air-drying control parameter combination is output as a second regulating parameter combination of the air-drying device.

[0091] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0092] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0093] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. An adaptive control method for air drying equipment, characterized in that: The method comprises: Conduct surface testing on food entering the air drying equipment to obtain the surface bacterial layer; Identifying the preset bacterial activity requirement of the surface bacterial layer for the food type, and determining whether the food needs to maintain bacterial activity based on the preset bacterial activity requirement; If the food needs to maintain bacterial activity, the air-drying device is controlled to enter an activity-retaining air-drying mode, and a first control parameter combination for controlling the air-drying device is obtained according to the activity-retaining air-drying mode with the preset bacterial activity requirement as an adaptive target; If the food does not need to maintain bacterial activity, the air-drying device is controlled to enter a stable drying mode, and a second control parameter combination for controlling the air-drying device is obtained according to the stable drying mode.

2. The method according to claim 1, wherein Performing surface inspection on food input into an air-drying device to obtain a surface bacterial layer, wherein the air-drying device is integrated with a surface inspection module, the surface inspection module including a visible light camera and a multispectral imaging component; acquiring a surface image of the food using the visible light camera; Acquiring a surface material reflectance spectrum image of the food according to the multispectral imaging component; Performing feature extraction on the surface image and the surface material reflectance spectrum image to obtain surface image convolution features and surface material convolution features; The surface image convolution feature and the surface material convolution feature are input as a combined vector into a bacterial species layer label recognition model for analysis to obtain the surface bacterial species layer of the food.

3. The method according to claim 2, wherein The surface image convolution feature and the surface material convolution feature are input as a combined vector into a strain layer label recognition model for analysis. The method for training the strain layer label recognition model includes: Collecting a surface image sample set, the surface image sample set including surface image samples based on different food types with known bacterial species layer labels and synchronously collected surface material reflectance spectrum image samples; Extracting texture feature vectors of surface image samples in the surface image sample set, and outputting them as surface image convolution feature samples; Extracting the reflection spectrum vector of the surface material reflection spectrum image sample from the surface image sample set, and outputting it as a surface material convolution feature sample; A lightweight MLP classifier is trained based on the surface image convolution feature samples, the surface material convolution feature samples and the known bacterial species layer labels until the accuracy of the lightweight MLP classifier is greater than a preset threshold, thereby obtaining a bacterial species layer label recognition model.

4. The method according to claim 3, wherein Training the twin network, feedback-connecting the lightweight MLP classifier to the twin network, and feedback-connecting the output of the twin network to the input of the lightweight MLP classifier; Constructing positive sample pairs and negative sample pairs, wherein the samples input into the first input terminal and the second input terminal of the positive sample pair are samples of the same category of bacterial species layer, and the samples input into the first input terminal and the second input terminal of the negative sample pair are samples of different categories of bacterial species layer; Using a comparison loss function to train the positive sample pair and the negative sample pair to obtain a similarity score sample; Feedback optimization is performed on the model of the lightweight MLP classifier based on the similarity score samples, and an optimized strain layer label recognition model is output.

5. The method according to claim 1, wherein Constructing a bacterial strain activity requirement rule table to identify preset bacterial strain activity requirements of food types with respect to the surface bacterial strain layer; The storage structure of the bacterial strain activity requirement rule table includes a ternary storage field, and the storage field includes a food type, a bacterial strain layer type, and an activity requirement index.

6. The method according to claim 5, wherein The method of determining whether the food needs to maintain bacterial activity according to the preset bacterial activity requirement includes: Based on the bacterial strain activity requirement rule table, a standard air-drying data sample group is collected for each food type and bacterial strain layer type, wherein each group of standard air-drying data samples includes surface activity detection samples at multiple air-drying stages, and the multiple air-drying stages include at least before air-drying and after air-drying; For each set of standard air-dried data samples, the difference in activity detection index between the surface activity detection index before air-drying and the surface activity detection index after air-drying is recorded; If the difference in the activity detection index is less than or equal to the preset activity detection index, the food does not need to maintain the activity of the bacterial strain; if the difference in the activity detection index is greater than the preset activity detection index, the food needs to maintain the activity of the bacterial strain.

7. The method according to claim 1, wherein According to the activity-retaining air-drying mode, taking the preset bacterial strain activity requirement as an adaptive target, a first control parameter combination for controlling the air-drying device is obtained, the method comprising: Obtaining an initial air-drying control parameter combination for air-drying the food, wherein the initial air-drying control parameter combination includes air-drying temperature, air-drying humidity, and air-drying speed and wind direction change frequency; Based on the preset bacterial strain activity requirement, obtaining the activity tolerance parameter threshold of the surface bacterial strain layer based on each control parameter in the initial air-drying control parameter combination; The initial air-drying control parameter combination is constrained according to the activity tolerance parameter threshold to obtain a first control parameter combination for controlling the air-drying device.

8. The method according to claim 7, wherein According to the stable drying and air-drying mode, taking the preset bacterial strain activity requirement as an adaptive target, a second control parameter combination for controlling the air-drying device is obtained, the method comprising: Performing phased stable adaptive processing on the initial air-drying control parameter combination to output multiple air-drying control parameter combinations; The multi-stage air-drying control parameter combination is output as the second regulating parameter combination of the air-drying equipment.

9. The method according to claim 4, wherein The air-drying device includes an edge computing chip, which is embedded with a trained bacterial layer label recognition model.

10. An adaptive control system for air drying equipment, characterized in that: The system comprises: A food surface detection module, which is used to perform surface detection on the food input into the air-drying device to obtain a surface bacterial layer; an activity requirement determination module, the activity requirement determination module being configured to identify a preset bacterial activity requirement of a food type with respect to the surface bacterial layer, and determine whether the food needs to maintain bacterial activity based on the preset bacterial activity requirement; an activity-retaining air-drying module, the activity-retaining air-drying module being used to control the air-drying device to enter an activity-retaining air-drying mode if the food needs to maintain bacterial activity, and to obtain a first control parameter combination for controlling the air-drying device according to the activity-retaining air-drying mode with the preset bacterial activity requirement as an adaptive target; The stable drying and air-drying module is used to control the air-drying equipment to enter the stable drying and air-drying mode if the food does not need to maintain the activity of the bacteria, and then obtain the second control parameter combination for controlling the air-drying equipment according to the stable drying and air-drying mode.

Images