Edible mushroom quality monitoring method and system

Abstract

The invention discloses an edible mushroom quality monitoring method and system, belongs to the technical field of agricultural Internet of Things and intelligent detection, and is technically characterized in that the system comprises an image acquisition module, a gas sensing module, a weight measurement module, an environmental parameter acquisition module, a data processing module and a display alarm module, image features, volatile gas concentration, weight loss rate and storage environment information of edible mushrooms can be obtained in real time. The system calculates a comprehensive quality index Q (t) through a fusion algorithm, judges the quality grade according to a preset threshold value, and achieves the dynamic quality evaluation and deterioration early warning of the edible mushrooms in the storage and transportation process. The system and the method are combined with multi-source sensing and an intelligent algorithm, have the characteristics of non-contact, automation, predictability and the like, can be widely applied to storage monitoring, logistics quality control, intelligent grading and other scenes of edible mushrooms such as shiitake mushrooms, needle mushrooms and oyster mushrooms, and improve the quality management efficiency and the intelligent level of a supply chain.

Description

Technical Field

[0001] The present invention relates to the field of agricultural Internet of Things and intelligent detection technology, and in particular to a method and system for monitoring the quality of edible fungi. Background Art

[0002] Edible fungi, due to their rich nutritional value and unique flavor, have become an indispensable component of the modern diet. With the development of agricultural technology and the increase in market demand, the large-scale cultivation and distribution of edible fungi such as shiitake mushrooms, enoki mushrooms, oyster mushrooms, and oyster mushrooms are becoming increasingly popular. However, edible fungi have physiological characteristics such as high water content, soft tissue structure, susceptibility to mechanical damage, and high respiratory intensity. They are prone to quality deterioration within a short period of time after harvest, with problems such as wilting, discoloration, decay, stickiness, or the production of odors. This type of quality decline not only affects the eating quality and consumer experience, but also causes significant losses in logistics, warehousing, and sales, seriously restricting the efficiency and economic benefits of the edible fungi industry chain.

[0003] Traditional methods for evaluating the quality of edible fungi rely primarily on visual inspection, empirical judgment, or offline testing methods, such as sampling and weighing, moisture determination, colorimetry, or odor analysis. These methods suffer from cumbersome operation, delayed response, poor real-time performance, isolated data, and significant human error. They are unable to meet the technical requirements for "full-process, full-parameter, fully closed-loop" quality monitoring in modern cold-chain logistics, intelligent packaging, and direct-from-source supply models. While some companies have attempted to deploy single methods such as temperature and humidity recorders, modified atmosphere packaging, or video surveillance to slow the deterioration of edible fungi or assist in their management, these methods are mostly static judgments or environmental side-tests, and are unable to achieve dynamic perception and comprehensive evaluation of the quality status of the fungi themselves.

[0004] The rapid development of artificial intelligence, the Internet of Things, and multimodal sensing technologies has provided a technological foundation for the digitalization, intelligence, and automation of edible fungus quality monitoring. However, a systematic solution that integrates multi-dimensional information such as image recognition, gas analysis, quality testing, and environmental perception, and provides algorithmic support and trend prediction capabilities, remains lacking. Developing a method and system for edible fungus quality monitoring based on multi-source sensing and fusion modeling has become a pressing technical challenge. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for monitoring the quality of edible fungi in order to address the deficiencies of the prior art and solve the problems raised in the background art.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An edible fungus quality monitoring system, comprising:

[0008] Image acquisition module, used to obtain edible fungus surface image data and extract color, grayscale, and structural image features;

[0009] Volatile organic compound gas sensor module, used to detect the concentration of gas components released by edible fungi i (t);

[0010] The weight measurement module is used to record the mass change per unit time and obtain the weight loss rate W(t);

[0011] Environmental parameter acquisition module, used to monitor the temperature T(t), humidity RH(t) and carbon dioxide concentration in the storage environment

[0012] A data processing module is used to fuse and normalize the sensor data of the above-mentioned environmental parameter acquisition module to obtain the environmental parameter combination E(t) and calculate the edible fungus quality index Q(t);

[0013] The display and alarm module is used to output quality levels and trigger quality deterioration warnings.

[0014] The calculation formula of the quality index Q(t) is:

[0015] Q(t)=α1·f1(I(t))+α2·f2(W(t))+α3·f3(V(t))+α4·f4(E(t))

[0016] Where: I(t): moment image feature vector; W(t): weight loss rate; V(t): VOC concentration index, E(t): environmental parameter combination, α1, α2, α3, α4: weight coefficients; f1~f4: normalization functions of each dimension.

[0017] Furthermore, the image feature vector I(t) includes the color mean C mean , image grayscale standard deviation σ G and edge definition parameter E s , which is in the form of:

[0018] I(t)=[C mean ,σ G ,E s ]

[0019] Where: C mean : the average value of bacterial surface color; σ G : Image grayscale fluctuation degree; E s : Clarity of the cap edge outline.

[0020] Furthermore, the data processing module has a built-in quality discrimination model based on a support vector machine, which outputs a quality level L(t) by comparing the calculated Q(t) with a preset threshold:

[0021]

[0022] Among them: θ1, θ2, θ3 are the boundary values ​​of each quality grade, and the unit is dimensionless index value.

[0023] Furthermore, the VOC concentration index V(t) is obtained by weighting the concentrations of multiple gas components, and its expression is:

[0024]

[0025] Where: c i (t): real-time concentration of the i-th odor component, unit: ppm; β i : The weight of the component's impact on quality, satisfying ∑β i =1.

[0026] Furthermore, the weight loss rate W(t) is obtained by regular measurement by the weight measurement module and is defined as:

[0027]

[0028] Where: M(t): momentary mass, unit: g; Δt: sampling time interval, unit: h.

[0029] Furthermore, the environmental parameter combination E(t) includes temperature T(t), relative humidity RH(t) and carbon dioxide concentration And normalization is used:

[0030]

[0031] Among them, T min ,T max : Upper and lower limits of normal temperature range, unit: °C; Standard reference concentration value, unit: ppm.

[0032] Furthermore, the display and alarm module is connected to a remote platform via a communication interface, and supports pushing quality index, grade and alarm information to a cloud server or mobile terminal.

[0033] The present invention also provides a method for monitoring the quality of edible fungi, which comprises the following steps:

[0034] S1: Initialize the device and record sample information;

[0035] S2: Periodically collects image, weight, odor, and environmental data;

[0036] S3: Extract image features I(t), calculate weight loss rate W(t), VOC concentration V(t) and environmental parameters E(t);

[0037] S4: Calculate the comprehensive quality index Q(t) according to the formula;

[0038] S5: Compare Q(t) with the quality level threshold and output the level L(t);

[0039] S6: When the quality level is in the "Warning" or "Corruption" state, an alarm is triggered or processing suggestions are prompted.

[0040] Furthermore, image feature extraction is performed using a convolutional neural network (CNN)-based model, and VOC features are dimensionally compressed through principal component analysis to improve discrimination accuracy and computational efficiency.

[0041] Furthermore, the quality index Q(t) of the past n consecutive moments is used to train a recurrent neural network (RNN) to predict the quality index of the kth hour in the future.

[0042]

[0043] Where: n: the length of the historical time window, in hours; k: the prediction step, in hours; Future forecast quality index.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1. By constructing a multi-module collaborative edible fungus quality monitoring system that includes image acquisition, gas sensing, weight measurement, and environmental monitoring, this invention enables comprehensive, multi-source, real-time sensing of key quality parameters during storage and transportation. This effectively overcomes the limitations of traditional methods, which suffer from single-source detection, lag time, and large errors. The system continuously tracks internal changes in fungus surface color, structural changes, odor release, and water loss, reflecting the freshness, spoilage trends, and storage risks of edible fungi, significantly improving the scientific nature and accuracy of quality monitoring.

[0046] 2. The integrated assessment model employed in this invention calculates a comprehensive quality index based on image features, weight loss rate, VOC gas concentration, and environmental factors. It also incorporates a machine learning algorithm for intelligent quality grade determination and supports RNN trend prediction, enabling early warning of future quality changes. Compared to traditional approaches that rely on threshold judgment, this system possesses intelligent decision-making, self-learning, and predictive capabilities, offering significant practical value in scenarios such as cold chain logistics, agricultural product grading, and freshness control.

[0047] 3. This invention boasts a high degree of structural integration and flexible deployment, making it suitable for a variety of storage containers or transport modules. It possesses excellent engineering feasibility and potential for industrialization and promotion. It can be widely applied to the quality testing of mainstream edible fungi, such as shiitake mushrooms, enoki mushrooms, and oyster mushrooms. It can also be expanded to other perishable agricultural products, providing digital and intelligent quality control technology support for the agricultural product supply chain, promoting the development of smart agriculture and intelligent cold chains. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a system block diagram of a method and system for monitoring edible fungi quality in an embodiment of the present invention; Figure 2 This is a flow chart of a method and system for monitoring the quality of edible fungi in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0050] The specific implementation of the present invention is described in detail below with reference to the accompanying drawings and examples.

[0051] Example: This invention proposes a system and method for dynamic monitoring and intelligent assessment of edible fungi quality, aiming to enable real-time tracking and quantitative assessment of the quality evolution of fungi during post-harvest transportation and storage. The system integrates image processing, gas sensing, weight monitoring, and environmental data collection modules, combined with a fusion algorithm, to output a comprehensive quality index for the edible fungi and provide quality grade warnings.

[0052] In specific implementation, the system structure of the present invention includes: image acquisition module, VOC gas sensor module, weight measurement module, environmental parameter acquisition module data processing module and display alarm module. The image acquisition module uses adjustable fill light LED and high-resolution camera to regularly capture the surface image of edible fungi and extract the color mean (C mean ), grayscale standard deviation (σ G ) and edge definition (E s ) and other image features. The image features constitute the vector I(t) = [C mean ,σ G ,E s], reflecting visual information such as color changes and structural changes on the bacterial surface. The gas sensor module deploys a variety of gas sensors, including ammonia (NH3), ethanol, hydrogen sulfide, etc., to monitor the concentration of odors produced by bacterial metabolism or corruption, which are recorded as c i (t), where i = 1, 2, ..., n, represents different types of odor molecules. The system uses a weighted method to fuse the odor concentration values ​​to form a comprehensive VOC index:

[0053]

[0054] Among them, β i is the weight of each component's influence on quality, satisfying ∑β i = 1. The weight measurement module uses a high-precision electronic scale to record the cell mass M(t) in real time and calculate the weight loss rate per unit time at a set interval Δt (e.g., 1 hour):

[0055]

[0056] The weight loss rate reflects the rate of water loss in the bacteria and indirectly measures the freshness. The environmental parameter acquisition module is used to synchronously record temperature T(t), relative humidity RH(t) and CO2 concentration. Taking into account the inconsistency of the physical quantities of different indicators, the system normalizes the environmental parameter vector:

[0057]

[0058] Among them, T min and T max They are the lower and upper limits of the allowable temperature range (e.g. 5-25°C), is the CO2 reference concentration (e.g. 1000ppm). All sensor data is uploaded to the data processing module. The system has an embedded quality assessment algorithm that inputs the four types of data vectors into the fusion calculation model to ultimately derive the comprehensive quality index Q(t), which is expressed as:

[0059] Q(t)=α1·f1(I(t))+α2·f2(W(t))+α3·f3(V(t))+α4·f4(E(t))

[0060] Among them, f1~f4 are normalized mapping functions, α1~α4 are weight coefficients of each factor, which are determined by fitting the training samples and usually satisfy ∑α i =1.

[0061] The quality index Q(t) is compared with the preset thresholds θ1, θ2, and θ3, and the corresponding quality level L(t) is output. The level rules are as follows:

[0062]

[0063] Quality levels and alarm information are displayed on the local touch screen and remote terminal platforms. The system supports threshold self-learning and integrates a convolutional neural network (CNN) model to automatically extract image features and improve recognition robustness.

[0064] To predict trends, the system further builds a recursive neural network (RNN) prediction model. Based on the quality index sequence Q(t), Q(t-1), ..., Q(t-n+1) at n historical time points, it predicts the quality change trend in the next k hours:

[0065]

[0066] in, is the predicted quality index, where n is the lookback window length (e.g., 6 hours) and k is the prediction step size (e.g., 2 hours). When the prediction results indicate that the quality is about to enter a "warning" or "corruption" state, the system will send a handling suggestion in advance.

[0067] In the embodiment of the present invention, the system is deployed in a standard storage box or cold chain transport container, operates stably, and has industrial-grade monitoring capabilities. In actual application, the system can cover a variety of mushrooms such as shiitake mushrooms, enoki mushrooms, oyster mushrooms, and shimeji mushrooms, and complete the whole process quality monitoring without disturbing the mushrooms.

[0068] To verify the practicality of the proposed system, a 48-hour quality monitoring experiment was conducted on freshly harvested Enoki mushroom samples (net weight 100g). The system was installed in a laboratory storage box equipped with constant temperature control, set at 18°C ​​and 85% relative humidity to simulate a normal temperature transportation environment. After the system was started, image, gas, quality, and environmental data were collected every hour, and the following information was recorded:

[0069] 1. Image acquisition and processing: An image directly above the bacteria is acquired every hour, and the following image features are extracted using the CNN model: Initial color mean: C mean =182 (unit pixel value, grayscale range 0-255); grayscale standard deviation: σ G =13.5; Edge clarity: E s =0.84. As time goes by, the image parameters change at 24 hours as follows: C mean =170, indicating that the color becomes darker; σ G =18.3, structural details intensified; E s =0.65, the edge is blurred, indicating that the cap begins to sink.

[0070] After normalizing the above parameters, we get: f1(I(t))=0.72;

[0071] 2. Weight change monitoring: Initial mass M(0) = 100g, mass at 24 hours is M(24) = 94.2g. Calculate the current weight loss rate per hour:

[0072] After normalization, weight loss index: f2(W(t))=0.74;

[0073] 3. Gas composition analysis: At the 24th hour, the sensor detected the following component concentrations:

[0074] Ammonia c1 = 4.2 ppm;

[0075] Ethanol c2 = 2.5 ppm;

[0076] Hydrogen sulfide c3 = 0.8 ppm.

[0077] The weight of each component β = [0.5, 0.3, 0.2], then the comprehensive value of VOC is: V (24) = 0.5·4.2 + 0.3·2.5 + 0.2·0.8 = 2.1 + 0.75 + 0.16 = 3.01;

[0078] After normalization, we get: f3(V(t))=0.78;

[0079] 4. Environmental parameter collection: temperature T = 18 ° C, T min =5℃,T max =25℃; humidity RH=85%, carbon dioxide concentration C CO2 =1200ppm, standard reference value but:

[0080]

[0081] After normalization, the average is obtained: f4(E(t))=0.90;

[0082] 5. Calculation of quality index: The weight coefficients are set as: α1 = 0.3, α2 = 0.2, α3 = 0.3, α4 = 0.2, then: Q(24) = 0.216 + 0.148 + 0.234 + 0.18 = 0.778;

[0083] 6. Quality level determination and alarm: The system sets the level determination threshold as follows:

[0084] θ1=0.85: excellent;

[0085] θ2=0.75: good;

[0086] θ3=0.60: Warning.

[0087] According to the calculation result Q(24)=0.778∈[0.75,0.85), the quality level at the 24th hour is “good”.

[0088] If the quality index drops by more than 0.1 within 2 hours during subsequent monitoring, the system will trigger an alarm, prompting "Please check the storage conditions of the fungus. Refrigeration is recommended." This example verifies the system's ability to identify the quality evolution of Enoki mushrooms under normal temperature storage and transportation scenarios, and can issue early warnings of quality decline, providing data support for fresh food circulation.

[0089] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to replace some of the technical features therein with equivalents. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An edible fungus quality monitoring system, characterized in that: include: Image acquisition module, used to obtain edible fungus surface image data and extract color, grayscale, and structural image features; Volatile organic compound gas sensor module, used to detect the concentration of gas components released by edible fungi i (t); The weight measurement module is used to record the mass change per unit time and obtain the weight loss rate W(t); Environmental parameter acquisition module, used to monitor the temperature T(t), humidity RH(t) and carbon dioxide concentration in the storage environment A data processing module is used to fuse and normalize the sensor data of the above-mentioned environmental parameter acquisition module to obtain the environmental parameter combination E(t) and calculate the edible fungus quality index Q(t); The display and alarm module is used to output quality levels and trigger quality deterioration warnings. The calculation formula of the quality index Q(t) is: Q(t)=α1·f1(I(t))+α2·f2(W(t))+α3·f3(V(t))+α4·f4(E(t)) Where: I(t): moment image feature vector; W(t): weight loss rate; V(t): VOC concentration index, E(t): environmental parameter combination, α1, α2, α3, α4: weight coefficients; f1~f4: normalization functions of each dimension.

2. The edible fungus quality monitoring system according to claim 1, characterized in that: The image feature vector I(t) includes the color mean C mean , image grayscale standard deviation σ G and edge definition parameter E s , which is in the form of: I(t)=[C mean ,σ G ,AND S ] Where: C mean : the average value of bacterial surface color; σ G : Image grayscale fluctuation degree; E s : Clarity of the cap edge outline.

3. The edible fungus quality monitoring system according to claim 1, characterized in that: The data processing module has a built-in quality discrimination model based on support vector machine, which outputs the quality level L(t) by comparing the calculated Q(t) with the preset threshold: Among them: θ1, θ2, θ3 are the boundary values ​​of each quality grade, and the unit is dimensionless index value.

4. The edible fungus quality monitoring system according to claim 1, characterized in that: The VOC concentration index V(t) is obtained by weighting the concentrations of multiple gas components, and its expression is: Where: c i (t): real-time concentration of the i-th odor component, unit: ppm; β i : The weight of the component's impact on quality, satisfying ∑β i =1.

5. The edible fungus quality monitoring system according to claim 1, characterized in that: The weight loss rate W(t) is obtained by regular measurement of the weight measurement module and is defined as: Where: M(t): momentary mass, unit: g; Δt: sampling time interval, unit: h.

6. The edible fungus quality monitoring system according to claim 1, characterized in that: The environmental parameter combination E(t) includes temperature T(t), relative humidity RH(t) and carbon dioxide concentration And normalization is used: Where: T min ,T max : Upper and lower limits of normal temperature range, unit: °C; Standard reference concentration value, unit: ppm.

7. The edible fungus quality monitoring system according to claim 1, characterized in that: The display and alarm module is connected to the remote platform through a communication interface, and supports pushing the quality index, grade and alarm information to the cloud server or mobile terminal.

8. A method for monitoring the quality of edible fungi based on the system according to any one of claims 1 to 7, characterized in that: include: S1: Initialize the device and record sample information; S2: Periodically collects image, weight, odor, and environmental data; S3: Extract image features I(t), calculate weight loss rate W(t), VOC concentration V(t) and environmental parameters E(t); S4: Calculate the comprehensive quality index Q(t); S5: Compare Q(t) with the quality level threshold and output the level L(t); S6: When the quality level is in the "Warning" or "Corruption" state, an alarm is triggered or processing suggestions are prompted.

9. The method according to claim 8, characterized in that Image feature extraction is processed using a model based on convolutional neural network (CNN). VOC features are dimensionally compressed through principal component analysis to improve discrimination accuracy and computational efficiency.

10. The method according to claim 8, characterized in that Use the quality index Q(t) of the past n consecutive moments to train a recurrent neural network (RNN) to predict the quality index of the kth hour in the future Where n is the length of the historical time window, in hours; k is the prediction step, in hours; Future forecast quality index.

Images