Fresh-keeping state detection system and method for transporting fresh fruits and vegetables

Abstract

The invention discloses a fresh-keeping state detection system and method for transportation of fresh fruits and vegetables, and relates to the technical field of transportation of fresh fruits and vegetables, the system comprises a multi-parameter sensor array, a transportation area image acquisition module, a real-time fresh-keeping data processing module and a fresh-keeping state prediction module; the real-time fresh-keeping data processing module is used for receiving data acquired in real time by a plurality of sensors included in the parameter sensor array and data acquired in real time by the transportation area image acquisition module; preprocessing data acquired by a plurality of sensors included in the parameter sensor array in real time and data acquired by a transportation area image acquisition module in real time to obtain sensor data and transportation area image data; and the fresh-keeping state prediction module is used for determining the fresh-keeping state of the fresh fruits and vegetables by utilizing the trained and optimized fresh-keeping state prediction model according to the sensor data and the transportation area image data.

Description

Technical Field

[0001] The present application relates to the technical field of fresh fruit and vegetable transportation, and in particular to a system and method for detecting the freshness of fresh fruits and vegetables during transportation. Background Art

[0002] With the development of global trade, long-distance transportation of fresh fruits and vegetables has become increasingly frequent. The freshness of these perishable products during transportation is directly related to their final nutritional value and market value. In order to ensure that fresh fruits and vegetables remain fresh during transportation, traditional transportation methods often rely on low-temperature refrigeration technology, but this cannot completely solve the preservation problem. Because fresh fruits and vegetables may be affected by temperature fluctuations, humidity changes, physical damage and other factors during transportation, these factors may lead to a decline in the quality of fruits and vegetables.

[0003] However, existing detection systems mostly rely on manual inspection, which is not only inefficient but also difficult to ensure the accuracy of detection.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present application provide a system and method for detecting the freshness of transported fresh fruits and vegetables to solve the above-mentioned technical problems.

[0006] The present application provides a fresh-keeping state detection system for transporting fresh fruits and vegetables, including a multi-parameter sensor array, a transport area image acquisition module, a real-time fresh-keeping data processing module, and a fresh-keeping state prediction module; wherein:

[0007] The multi-parameter sensor array includes a temperature sensor, a humidity sensor, a gas concentration sensor and a light intensity sensor, and the multiple sensors included in the multi-parameter sensor array are arranged in a transportation device for fresh fruits and vegetables;

[0008] The real-time preservation data processing module is used to receive the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module, and pre-process the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module to obtain sensor data and transportation area image data;

[0009] The fresh-keeping state prediction module is used to determine the fresh-keeping state of the fresh fruits and vegetables according to the sensor data and the transportation area image data using the fresh-keeping state prediction model that has been trained and optimized.

[0010] Furthermore, the real-time preservation data processing module preprocesses the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module to obtain sensor data and transportation area image data, and is configured as follows:

[0011] Performing denoising and normalization processing on the data collected in real time by the multiple sensors included in the parameter sensor array to obtain the sensor data;

[0012] The data collected in real time by the transport area image collection module is compressed and enhanced to obtain the transport area image data.

[0013] Furthermore, the fresh-keeping state prediction model is trained and optimized based on the following steps:

[0014] Obtain historical sensor data and historical transport area image data;

[0015] Extracting statistical features and frequency domain features of the historical sensor data; wherein the frequency domain features include Fourier transform coefficients and harmonics;

[0016] Extracting regional image features of the historical transportation regional image data; wherein the regional image features include texture features, color distribution features, shape features and size features;

[0017] Using the mutual information and maximum information coefficient algorithm, a first weighted vector, a first feature importance vector, and a first feature relevance vector are assigned to the statistical feature; a second weighted vector, a second feature importance vector, and a second feature relevance vector are assigned to the frequency domain feature; and a third weighted vector, a third feature importance vector, and a third feature relevance vector are assigned to the regional image feature;

[0018] Determine a comprehensive weighted vector, a comprehensive feature importance vector, and a comprehensive feature relevance vector according to the first weighted vector, the first feature importance vector, the first feature relevance vector, the second weighted vector, the second feature importance vector, the second feature relevance vector, the third weighted vector, the third feature importance vector, and the third feature relevance vector;

[0019] Using a causal forest algorithm, analyzing the causal relationship between the statistical characteristics of the historical sensor data and the frequency domain characteristics and the fresh-keeping state, and analyzing the causal relationship between the regional image characteristics of the historical transportation area image data and the fresh-keeping state;

[0020] Generative adversarial networks are used to simulate the appearance changes of fresh fruits and vegetables under different preservation conditions, so as to strengthen the causal relationship between the statistical characteristics of the enhanced historical sensor data and the frequency domain characteristics and the preservation state, and the causal relationship between the regional image characteristics and the preservation state, and obtain the first enhanced causal relationship and the second enhanced causal relationship; wherein different preservation conditions correspond to different temperatures, humidity, gas concentrations and light intensities;

[0021] Based on the first reinforced causal relationship and the second reinforced causal relationship, combined with the historical sensor data and the historical transportation area image data, the freshness state prediction model is trained and optimized through a gradient boosting machine.

[0022] Furthermore, the fresh-keeping status detection system for transporting fresh fruits and vegetables also includes:

[0023] The user personalized setting module is used to initialize the monitoring parameter threshold according to the type and preservation requirements of fresh fruits and vegetables; and receive the modification instruction of the monitoring parameter threshold from the transportation manager of the fresh fruits and vegetables, and adjust the monitoring parameter threshold.

[0024] Furthermore, the fresh-keeping status detection system for transporting fresh fruits and vegetables also includes:

[0025] The abnormal state alarm module is used to send an alarm message to the transportation manager when the sensor data or the transportation area image data exceeds the monitoring parameter threshold, and generate a transportation plan optimization suggestion based on the sensor data or the transportation area image data.

[0026] Furthermore, the fresh-keeping status detection system for transporting fresh fruits and vegetables also includes:

[0027] The wireless data transmission module is used to transmit the sensor data and the transport area image data to the cloud database in real time.

[0028] The present application provides a method for detecting the freshness of fresh fruits and vegetables during transportation, receiving data collected in real time by multiple sensors included in a parameter sensor array and data collected in real time by a transportation area image acquisition module, and preprocessing the data collected in real time by multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module to obtain sensor data and transportation area image data;

[0029] Wherein, the multi-parameter sensor array includes a temperature sensor, a humidity sensor, a gas concentration sensor and a light intensity sensor, and the multiple sensors included in the multi-parameter sensor array are arranged in the transportation device for fresh fruits and vegetables;

[0030] The freshness status of the fresh fruits and vegetables is determined according to the sensor data and the transport area image data using the freshness status prediction model that has been trained and optimized.

[0031] Furthermore, the fresh-keeping state prediction model is trained and optimized based on the following steps:

[0032] Obtain historical sensor data and historical transport area image data;

[0033] Extracting statistical features and frequency domain features of the historical sensor data; wherein the frequency domain features include Fourier transform coefficients and harmonics;

[0034] Extracting regional image features of the historical transportation regional image data; wherein the regional image features include texture features, color distribution features, shape features and size features;

[0035] Using the mutual information and maximum information coefficient algorithm, a first weighted vector, a first feature importance vector, and a first feature relevance vector are assigned to the statistical feature; a second weighted vector, a second feature importance vector, and a second feature relevance vector are assigned to the frequency domain feature; and a third weighted vector, a third feature importance vector, and a third feature relevance vector are assigned to the regional image feature;

[0036] Determine a comprehensive weighted vector, a comprehensive feature importance vector, and a comprehensive feature relevance vector according to the first weighted vector, the first feature importance vector, the first feature relevance vector, the second weighted vector, the second feature importance vector, the second feature relevance vector, the third weighted vector, the third feature importance vector, and the third feature relevance vector;

[0037] Using a causal forest algorithm, analyzing the causal relationship between the statistical characteristics of the historical sensor data and the frequency domain characteristics and the fresh-keeping state, and analyzing the causal relationship between the regional image characteristics of the historical transportation area image data and the fresh-keeping state;

[0038] Generative adversarial networks are used to simulate the appearance changes of fresh fruits and vegetables under different preservation conditions, so as to strengthen the causal relationship between the statistical characteristics of the enhanced historical sensor data and the frequency domain characteristics and the preservation state, and the causal relationship between the regional image characteristics and the preservation state, and obtain the first enhanced causal relationship and the second enhanced causal relationship; wherein different preservation conditions correspond to different temperatures, humidity, gas concentrations and light intensities;

[0039] Based on the first reinforced causal relationship and the second reinforced causal relationship, combined with the historical sensor data and the historical transportation area image data, the freshness state prediction model is trained and optimized through a gradient boosting machine.

[0040] Based on the embodiment provided by the present application, the multi-parameter sensor array includes a temperature sensor, a humidity sensor, a gas concentration sensor and a light intensity sensor, and the multiple sensors included in the multi-parameter sensor array are arranged in the transportation device of fresh fruits and vegetables; the real-time preservation data processing module is used to receive the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module, and pre-process the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module to obtain sensor data and transportation area image data; the preservation state prediction module is used to determine the preservation state of fresh fruits and vegetables according to the sensor data and the transportation area image data, using the preservation state prediction model completed by training optimization. Thereby, the real-time monitoring of the preservation state of fruits and vegetables is realized, and the detection efficiency and accuracy of the preservation state detection system for transporting fresh fruits and vegetables is improved; and the preservation quality of fruits and vegetables during transportation can be improved, and decision support can be provided for transportation managers, thereby optimizing the transportation process, reducing losses, and improving economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The drawings described herein are used to provide a further understanding of the embodiments of the present invention and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0042] Figure 1 It is a structural diagram of an optional fresh-keeping status detection system for transporting fresh fruits and vegetables according to an embodiment of the present application;

[0043] Figure 2 The present invention is a flowchart of an optional method for detecting the freshness status of fresh fruits and vegetables in transportation according to an embodiment of the present application.

[0044] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0045] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0046] Alternatively, if Figure 1 As shown, the present application provides a fresh-keeping state detection system for transporting fresh fruits and vegetables, including a multi-parameter sensor array 101, a transport area image acquisition module 102, a real-time fresh-keeping data processing module 103 and a fresh-keeping state prediction module 104; wherein,

[0047] The multi-parameter sensor array 101 includes a temperature sensor, a humidity sensor, a gas concentration sensor, and a light intensity sensor. The multiple sensors included in the multi-parameter sensor array 101 are arranged in the transportation device for fresh fruits and vegetables;

[0048] The real-time preservation data processing module 103 is used to receive the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module 102, and pre-process the data collected in real time by the multiple sensors included in the parameter sensor array 101 and the data collected in real time by the transportation area image acquisition module 102 to obtain sensor data and transportation area image data;

[0049] The freshness state prediction module 104 is used to determine the freshness state of fresh fruits and vegetables based on the sensor data and the transportation area image data using the freshness state prediction model that has been trained and optimized.

[0050] Based on the embodiment provided by the present application, the multi-parameter sensor array includes a temperature sensor, a humidity sensor, a gas concentration sensor and a light intensity sensor, and the multiple sensors included in the multi-parameter sensor array are arranged in the transportation device of fresh fruits and vegetables; the real-time preservation data processing module is used to receive the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module, and pre-process the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module to obtain sensor data and transportation area image data; the preservation state prediction module is used to determine the preservation state of fresh fruits and vegetables according to the sensor data and the transportation area image data, using the preservation state prediction model completed by training optimization. Thereby, the real-time monitoring of the preservation state of fruits and vegetables is realized, and the detection efficiency and accuracy of the preservation state detection system for transporting fresh fruits and vegetables is improved; and the preservation quality of fruits and vegetables during transportation can be improved, and decision support can be provided for transportation managers, thereby optimizing the transportation process, reducing losses, and improving economic benefits.

[0051] Furthermore, the real-time preservation data processing module pre-processes the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module to obtain sensor data and transportation area image data, and is configured as follows:

[0052] Performing denoising and normalization processing on the data collected in real time by multiple sensors included in the parameter sensor array to obtain sensor data;

[0053] The data collected in real time by the transport area image acquisition module is compressed and enhanced to obtain transport area image data.

[0054] Furthermore, the freshness state prediction model is trained and optimized based on the following steps:

[0055] Obtain historical sensor data and historical transport area image data;

[0056] Extract statistical features and frequency domain features of historical sensor data; wherein the frequency domain features include Fourier transform coefficients and harmonics;

[0057] Extracting regional image features of historical transportation regional image data; wherein the regional image features include texture features, color distribution features, shape features and size features;

[0058] Using the mutual information and maximum information coefficient algorithm, a first weighted vector, a first feature importance vector, and a first feature relevance vector are assigned to the statistical feature; a second weighted vector, a second feature importance vector, and a second feature relevance vector are assigned to the frequency domain feature; and a third weighted vector, a third feature importance vector, and a third feature relevance vector are assigned to the regional image feature;

[0059] Determine a comprehensive weighted vector, a comprehensive feature importance vector, and a comprehensive feature relevance vector according to the first weighted vector, the first feature importance vector, the first feature relevance vector, the second weighted vector, the second feature importance vector, the second feature relevance vector, the third weighted vector, the third feature importance vector, and the third feature relevance vector;

[0060] Using the causal forest algorithm, the statistical characteristics of historical sensor data and the causal relationship between frequency domain characteristics and freshness status are analyzed, as well as the causal relationship between regional image characteristics and freshness status of historical transportation area image data are analyzed;

[0061] Generative adversarial networks are used to simulate the appearance changes of fresh fruits and vegetables under different preservation conditions, so as to strengthen the statistical characteristics of historical sensor data, the causal relationship between frequency domain characteristics and preservation status, and the causal relationship between regional image characteristics and preservation status, and obtain the first strengthened causal relationship and the second strengthened causal relationship; where different preservation conditions correspond to different temperatures, humidity, gas concentrations, and light intensities;

[0062] Based on the first reinforcement causal relationship and the second reinforcement causal relationship, combined with historical sensor data and historical transportation area image data, the freshness status prediction model is trained and optimized through the gradient boosting machine.

[0063] Further, according to the first weighted vector, the first feature importance vector, the first feature relevance vector, the second weighted vector, the second feature importance vector, the second feature relevance vector, the third weighted vector, the third feature importance vector and the third feature relevance vector, determining a comprehensive weighted vector, a comprehensive feature importance vector and a comprehensive feature relevance vector, including:

[0064] W releva =α×W1+β×W2+(1-α-β)×W3

[0065] H releva =α×H1+β×H2+(1-α-β)×H3

[0066] C releva =α×C1+β×C2+(1-α-β)×C3

[0067] Among them, W releva is the comprehensive weighted vector; W1 is the first weighted vector; W2 is the second weighted vector; W3 is the third weighted vector; H releva is the comprehensive feature importance vector; H1 is the first feature importance vector; H2 is the second feature importance vector; H3 is the third feature importance vector; C releva is the comprehensive feature correlation vector; C1 is the first feature correlation vector; C2 is the second feature correlation vector; C3 is the third feature correlation vector; α and β are both adjustment factors; α is used to adjust the contribution of feature importance; β is used to adjust the contribution of feature correlation.

[0068] Furthermore, in the process of using the generative adversarial network to simulate the appearance changes of fresh fruits and vegetables under different preservation conditions, the following loss function is used;

[0069] L GAN =γ×L LSGAN +(1-γ)×L WGAN

[0070] Among them, L GAN is the total loss function of the generated adversarial network, which is used to balance the least squares GAN loss and the Wasserstein GAN loss; L LSGAN is the loss function of the least squares GAN; L WGAN is the loss function of Wasserstein GAN; γ is the importance adjustment parameter.

[0071] Furthermore, the freshness detection system for transporting fresh fruits and vegetables also includes:

[0072] The user personalized setting module is used to initialize the monitoring parameter thresholds according to the types and preservation requirements of fresh fruits and vegetables; and to receive modification instructions for the monitoring parameter thresholds from the transportation manager of fresh fruits and vegetables, and adjust the monitoring parameter thresholds.

[0073] Furthermore, the freshness detection system for transporting fresh fruits and vegetables also includes:

[0074] The abnormal status alarm module is used to send an alarm message to the transportation manager when the sensor data or the transportation area image data exceeds the monitoring parameter threshold, and generate transportation plan optimization suggestions based on the sensor data or the transportation area image data.

[0075] Furthermore, the freshness detection system for transporting fresh fruits and vegetables also includes:

[0076] Wireless data transmission module, used to transmit sensor data and transportation area image data to the cloud database in real time.

[0077] Alternatively, if Figure 2 As shown, the present application provides a method for detecting the freshness of transported fresh fruits and vegetables, comprising:

[0078] S201, receiving data collected in real time by multiple sensors included in the parameter sensor array and data collected in real time by the transport area image acquisition module, and preprocessing the data collected in real time by multiple sensors included in the parameter sensor array and the data collected in real time by the transport area image acquisition module to obtain sensor data and transport area image data;

[0079] The multi-parameter sensor array includes a temperature sensor, a humidity sensor, a gas concentration sensor, and a light intensity sensor. The multi-parameter sensor array includes a plurality of sensors arranged in a transport device for fresh fruits and vegetables.

[0080] S202, determining the freshness status of fresh fruits and vegetables based on the sensor data and the transport area image data using the trained and optimized freshness status prediction model.

[0081] Furthermore, the freshness state prediction model is trained and optimized based on the following steps:

[0082] Obtain historical sensor data and historical transport area image data;

[0083] Extract statistical features and frequency domain features of historical sensor data; wherein the frequency domain features include Fourier transform coefficients and harmonics;

[0084] Extracting regional image features of historical transportation regional image data; wherein the regional image features include texture features, color distribution features, shape features and size features;

[0085] Using the mutual information and maximum information coefficient algorithm, a first weighted vector, a first feature importance vector, and a first feature relevance vector are assigned to the statistical feature; a second weighted vector, a second feature importance vector, and a second feature relevance vector are assigned to the frequency domain feature; and a third weighted vector, a third feature importance vector, and a third feature relevance vector are assigned to the regional image feature;

[0086] Determine a comprehensive weighted vector, a comprehensive feature importance vector, and a comprehensive feature relevance vector according to the first weighted vector, the first feature importance vector, the first feature relevance vector, the second weighted vector, the second feature importance vector, the second feature relevance vector, the third weighted vector, the third feature importance vector, and the third feature relevance vector;

[0087] Using the causal forest algorithm, the statistical characteristics of historical sensor data and the causal relationship between frequency domain characteristics and freshness status are analyzed, as well as the causal relationship between regional image characteristics and freshness status of historical transportation area image data are analyzed;

[0088] Generative adversarial networks are used to simulate the appearance changes of fresh fruits and vegetables under different preservation conditions, so as to strengthen the statistical characteristics of historical sensor data, the causal relationship between frequency domain characteristics and preservation status, and the causal relationship between regional image characteristics and preservation status, and obtain the first strengthened causal relationship and the second strengthened causal relationship; where different preservation conditions correspond to different temperatures, humidity, gas concentrations, and light intensities;

[0089] Based on the first reinforcement causal relationship and the second reinforcement causal relationship, combined with historical sensor data and historical transportation area image data, the freshness status prediction model is trained and optimized through the gradient boosting machine.

[0090] It should be noted that in the present application, the embodiments implemented by the freshness status detection system for transporting fresh fruits and vegetables can be referenced to each other with the embodiments implemented by the freshness status detection method for transporting fresh fruits and vegetables, and the present application will not repeat them one by one.

[0091] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A fresh-keeping status detection system for transporting fresh fruits and vegetables, characterized in that: It includes a multi-parameter sensor array, a transport area image acquisition module, a real-time preservation data processing module, and a preservation status prediction module; The multi-parameter sensor array includes a temperature sensor, a humidity sensor, a gas concentration sensor and a light intensity sensor, and the multiple sensors included in the multi-parameter sensor array are arranged in a transportation device for fresh fruits and vegetables; The real-time preservation data processing module is used to receive the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module, and pre-process the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module to obtain sensor data and transportation area image data; The fresh-keeping state prediction module is used to determine the fresh-keeping state of the fresh fruits and vegetables according to the sensor data and the transportation area image data using the fresh-keeping state prediction model that has been trained and optimized.

2. The fresh-keeping status detection system for transporting fresh fruits and vegetables according to claim 1 is characterized in that: The real-time preservation data processing module pre-processes the data collected in real time by the multiple sensors included in the parameter sensor array and the data collected in real time by the transportation area image acquisition module to obtain sensor data and transportation area image data, and is configured as follows: Performing denoising and normalization processing on the data collected in real time by the multiple sensors included in the parameter sensor array to obtain the sensor data; The data collected in real time by the transport area image collection module is compressed and enhanced to obtain the transport area image data.

3. The fresh-keeping status detection system for transporting fresh fruits and vegetables according to claim 2 is characterized in that: The fresh-keeping state prediction model is trained and optimized based on the following steps: Obtain historical sensor data and historical transport area image data; Extracting statistical features and frequency domain features of the historical sensor data; wherein the frequency domain features include Fourier transform coefficients and harmonics; Extracting regional image features of the historical transportation regional image data; wherein the regional image features include texture features, color distribution features, shape features and size features.

4. The fresh-keeping status detection system for transporting fresh fruits and vegetables according to claim 3 is characterized in that: The fresh-keeping state prediction model is trained and optimized based on the following steps: Using mutual information and maximum information coefficient algorithm, assign a first weight vector, a first feature importance vector, and a first feature relevance vector to the statistical feature; and assigning a second weighting vector, a second feature importance vector, and a second feature relevance vector to the frequency domain feature; and assigning a third weighting vector, a third feature importance vector, and a third feature relevance vector to the regional image feature; Determine a comprehensive weighted vector, a comprehensive feature importance vector, and a comprehensive feature relevance vector according to the first weighted vector, the first feature importance vector, the first feature relevance vector, the second weighted vector, the second feature importance vector, the second feature relevance vector, the third weighted vector, the third feature importance vector, and the third feature relevance vector; The causal forest algorithm is used to analyze the causal relationship between the statistical characteristics of the historical sensor data and the frequency domain characteristics and the fresh-keeping state, and to analyze the causal relationship between the regional image characteristics of the historical transportation area image data and the fresh-keeping state.

5. The fresh-keeping status detection system for transporting fresh fruits and vegetables according to claim 4 is characterized in that: The fresh-keeping state prediction model is trained and optimized based on the following steps: Generative adversarial networks are used to simulate the appearance changes of fresh fruits and vegetables under different preservation conditions, so as to strengthen the causal relationship between the statistical characteristics of the enhanced historical sensor data and the frequency domain characteristics and the preservation state, and the causal relationship between the regional image characteristics and the preservation state, and obtain the first enhanced causal relationship and the second enhanced causal relationship; wherein different preservation conditions correspond to different temperatures, humidity, gas concentrations and light intensities; Based on the first reinforced causal relationship and the second reinforced causal relationship, combined with the historical sensor data and the historical transportation area image data, the freshness state prediction model is trained and optimized through a gradient boosting machine.

6. The fresh-keeping status detection system for transporting fresh fruits and vegetables according to claim 1 is characterized in that: The fresh-keeping status detection system for transporting fresh fruits and vegetables also includes: The user personalized setting module is used to initialize the monitoring parameter threshold according to the type and preservation requirements of fresh fruits and vegetables; and receive the modification instruction of the monitoring parameter threshold from the transportation manager of the fresh fruits and vegetables, and adjust the monitoring parameter threshold.

7. The fresh-keeping status detection system for transporting fresh fruits and vegetables according to claim 6, characterized in that: The fresh-keeping status detection system for transporting fresh fruits and vegetables also includes: The abnormal state alarm module is used to send an alarm message to the transportation manager when the sensor data or the transportation area image data exceeds the monitoring parameter threshold, and generate a transportation plan optimization suggestion based on the sensor data or the transportation area image data.

8. The fresh-keeping status detection system for transporting fresh fruits and vegetables according to claim 1, characterized in that: The fresh-keeping status detection system for transporting fresh fruits and vegetables also includes: The wireless data transmission module is used to transmit the sensor data and the transport area image data to the cloud database in real time.

9. A method for detecting the freshness of fresh fruits and vegetables in transportation, comprising: Receiving data collected in real time by multiple sensors included in the parameter sensor array and data collected in real time by the transport area image acquisition module, and preprocessing the data collected in real time by multiple sensors included in the parameter sensor array and the data collected in real time by the transport area image acquisition module to obtain sensor data and transport area image data; Wherein, the multi-parameter sensor array includes a temperature sensor, a humidity sensor, a gas concentration sensor and a light intensity sensor, and the multiple sensors included in the multi-parameter sensor array are arranged in the transportation device for fresh fruits and vegetables; The freshness status of the fresh fruits and vegetables is determined according to the sensor data and the transport area image data using the freshness status prediction model that has been trained and optimized.

10. The method for detecting the freshness of fresh fruits and vegetables in transportation according to claim 9, characterized in that: The fresh-keeping state prediction model is trained and optimized based on the following steps: Obtain historical sensor data and historical transport area image data; Extracting statistical features and frequency domain features of the historical sensor data; wherein the frequency domain features include Fourier transform coefficients and harmonics; Extracting regional image features of the historical transportation area image data; wherein the regional image features include texture features, color distribution features, shape features and size features; Using the mutual information and maximum information coefficient algorithm, a first weighted vector, a first feature importance vector, and a first feature relevance vector are assigned to the statistical feature; a second weighted vector, a second feature importance vector, and a second feature relevance vector are assigned to the frequency domain feature; and a third weighted vector, a third feature importance vector, and a third feature relevance vector are assigned to the regional image feature; Determine a comprehensive weighted vector, a comprehensive feature importance vector, and a comprehensive feature relevance vector according to the first weighted vector, the first feature importance vector, the first feature relevance vector, the second weighted vector, the second feature importance vector, the second feature relevance vector, the third weighted vector, the third feature importance vector, and the third feature relevance vector; Using a causal forest algorithm, analyzing the causal relationship between the statistical characteristics of the historical sensor data and the frequency domain characteristics and the fresh-keeping state, and analyzing the causal relationship between the regional image characteristics of the historical transportation area image data and the fresh-keeping state; Generative adversarial networks are used to simulate the appearance changes of fresh fruits and vegetables under different preservation conditions, so as to strengthen the causal relationship between the statistical characteristics of the enhanced historical sensor data and the frequency domain characteristics and the preservation state, and the causal relationship between the regional image characteristics and the preservation state, and obtain the first enhanced causal relationship and the second enhanced causal relationship; wherein different preservation conditions correspond to different temperatures, humidity, gas concentrations and light intensities; Based on the first reinforced causal relationship and the second reinforced causal relationship, combined with the historical sensor data and the historical transportation area image data, the freshness state prediction model is trained and optimized through a gradient boosting machine.

Images