Monitoring and tracing system for agricultural product supply chain transportation

Abstract

The invention discloses a monitoring and traceability system for agricultural product supply chain transportation, and particularly relates to the field of supply chain transportation monitoring and traceability, which comprises the following steps: carrying out damage resistance quantitative evaluation of sliding, collision and extrusion on each agricultural product category in a supply chain, training a sound recognition model based on marked audio signal time-frequency domain features, collecting cargo information and audio data of each link segment in the transportation line, and uploading the cargo information and the audio data to a monitoring center; classifying and recognizing the audio data through a voice recognition model, and calculating a damage assessment score of each link segment in combination with a damage resistance quantitative index; and carrying out visual labeling on a damage evaluation result, sending an abnormal signal to a corresponding link segment, adjusting the transportation density to reduce the damage risk, realizing accurate traceability of the transportation link with abnormal loss, and ensuring that the damage degree of goods in transportation is controllable.

Description

Technical Field

[0001] The present invention relates to the technical field of supply chain transportation monitoring and tracing technology, and more specifically, to a monitoring and tracing system for agricultural product supply chain transportation. Background Art

[0002] During the transportation of agricultural products in the supply chain, goods are often subject to sliding, collision, extrusion and other forces, which can cause varying degrees of damage to the quality and appearance of agricultural products. Traditional damage monitoring methods rely on manual inspection or visual inspection, which is not only labor-intensive, but also difficult to achieve full-process monitoring, prone to missed inspections and misjudgments, and unable to accurately trace the transportation links that frequently experience sliding, collision, and extrusion damage, resulting in no basis for targeted prevention and optimization of such damage during transportation.

[0003] With the rapid development of the Internet of Things and intelligent sensing technology, sound signals can become an analytical basis that can reflect the external forces acting on goods during transportation. Sound analysis has the characteristics of high flexibility and strong detail recognition ability. Therefore, how to accurately identify different types of damage and transportation sections that are prone to damage in the transportation link through the analysis of sound signals, and realize a monitoring and traceability system for agricultural product supply chain transportation, meet the precise traceability of transportation links with abnormal losses, and ensure that the degree of damage to goods during transportation is controllable, is an urgent problem to be solved.

[0004] In order to solve the above problems, a technical solution is now provided. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a monitoring and traceability system for agricultural product supply chain transportation to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions: A monitoring and traceability system for agricultural product supply chain transportation, including a damage resistance quantification module, a sound recognition model training module, a data acquisition and upload module, a sound recognition module, a link segment damage assessment module, and a damage traceability module; The damage resistance quantification module conducts quantitative evaluation of the damage resistance of sliding, impact and extrusion for each agricultural product category in the supply chain; The sound recognition model training module labels the sound type of the audio signal generated by the contact between the vehicle and the cargo, and trains the sound recognition model based on the time-frequency domain characteristics of the sound signal after type labeling; The data collection and upload module decomposes the supply chain transportation route into link segments, uses the batch number of the goods as the monitoring dimension to record the agricultural product category, quantity, transportation vehicle specifications and collected audio data of the goods in each link segment, and uploads the data to the transportation link monitoring center; The sound recognition module classifies and recognizes the audio data of the transport link monitoring center through the sound recognition model; The link segment damage assessment module calculates the damage assessment score by calculating the frequency of sliding, impact, and squeezing sounds in the audio segments of each link segment and combining the damage resistance quantitative index value corresponding to the agricultural product category to which the goods belong. The damage tracing module visually marks the damage of link segments on the transportation route and sends abnormal transportation damage signals to specific link segments to adjust the transportation density of goods.

[0007] Furthermore, the damage resistance of each agricultural product category in the supply chain is quantitatively evaluated for sliding, impact, and extrusion, including: Obtain all agricultural product categories in the agricultural product supply chain, select several test samples from each product category to form test sample groups A, B, and C, and conduct damage tests based on sliding, impact, and extrusion; Select test sample group A, use a sliding friction tester to simulate the sliding friction during transportation, gradually increase the friction force on the test sample until the test sample surface is cracked and damaged, and record the instrument friction force reading; Select test sample group B to conduct a free fall experiment. Let the test samples fall freely from different heights onto the surface of the mechanical sensor until the surface of the test samples is broken and damaged, and record the sensor readings; Select test sample group C, use a pressure loading device to gradually apply pressure to simulate the squeezing during transportation, until rupture damage appears on the surface of the test sample, and record the loading pressure reading; Calculate the mean of the reading records of different test samples of each agricultural product category in the sliding, impact and extrusion damage tests, normalize the mean to remove the unit and map it to the interval expression of [0,1]. The result is used as the quantitative index value of the sliding, impact and extrusion damage resistance of this agricultural product category.

[0008] Furthermore, the audio signal generated by the contact between the vehicle and the cargo is labeled with the sound type, and the sound recognition model is trained based on the time-frequency domain features of the sound signal after the type labeling, specifically including: Acoustic sensors are placed on transport vehicles to collect audio data generated by the contact between the vehicle and the cargo. The audio data is processed using signal processing algorithms to remove the operating noise of the vehicle. Extract the sound clips containing sliding, impact and squeezing sounds from the denoised audio, and label the sound types; Short-time Fourier transform is used to analyze the signal in the time and frequency domain to extract the time and frequency domain features of the sound signals of different types of sound clips. The time and frequency domain features include amplitude, waveform shape, duration, change rate, frequency distribution and frequency peak value. A sound recognition model for the transportation process is established. The labeled time-frequency domain features of the sound signal are used as the training data set. The sound recognition model is trained and verified through a machine learning algorithm. The verified sound recognition model is deployed to the transportation link monitoring center.

[0009] Furthermore, the supply chain transport route is decomposed into link segments, and the batch number of the goods is used as the monitoring dimension to record the agricultural product category, cargo quantity, transport vehicle specifications and collected audio data of the goods in each link segment, and upload the data to the transport link monitoring center. Specifically, it includes: Mark the transport transfer stations in all transport routes in the supply chain as link nodes; Taking the batch number of goods as the monitoring dimension, the actual transportation route of each batch of goods is obtained, and the transportation route of the same batch of goods is split into link segments based on adjacent link nodes; When the same batch of goods passes through a specific link segment, the timestamps of entering and leaving the link segment, as well as the agricultural product category, quantity and transport vehicle specifications of the goods transported in the link segment are recorded; The acoustic sensors on the transport vehicle collect the audio data generated by the contact between the vehicle and the goods in real time. After all the data are integrated, a corresponding relationship with the batch number of the goods is established and uploaded to the transport link monitoring center.

[0010] Furthermore, the audio data of the transport link monitoring center is classified and identified by the sound recognition model, specifically including: Select the batch number of goods that need to be monitored for transportation, and obtain the agricultural product category of the goods corresponding to the batch number, the entry and exit timestamps of each link segment, and the complete audio data collected from the transportation link monitoring center; Dividing the audio data into audio segments aligned with the link segment transit time periods based on the link segment ingress and egress timestamps; Extract the time-frequency domain features of the audio segments as input recognition features of the sound recognition model, and select the main classification features corresponding to the sliding, impacting, and squeezing sounds from the input recognition features; Obtain the sliding, impact, and extrusion damage resistance quantitative index values ​​corresponding to the agricultural product category to which the transport monitored goods belong; Based on the damage resistance quantitative index values ​​of various types of damage, the classification decision sensitivity of the sound recognition model is regulated by controlling the input recognition feature expression; The final classification and recognition output results of the sound recognition model after the classification decision sensitivity adjustment are obtained, and the number of occurrences of the three sounds of sliding, impact, and squeezing in the audio segments of each link segment is recorded.

[0011] Furthermore, based on the damage resistance quantitative index values ​​of various types of damage, by controlling the input recognition feature expression, the classification decision sensitivity of the sound recognition model is regulated, specifically including: Set the benchmark reference values ​​of the quantitative indicators of sliding, impact and extrusion damage resistance. If the quantitative indicator of damage resistance of a certain agricultural product category to which the goods belong is greater than its corresponding benchmark reference value, the goods are marked as having high damage resistance to the damage type. Otherwise, the goods are marked as having low damage resistance to the damage type. Based on the deviation ratio of each damage resistance quantitative indicator of the agricultural product category to which the goods belong and the corresponding benchmark value, the classification main features of the sound category corresponding to the damage type with low damage resistance in the input recognition features are subjected to feature enhancement with equal deviation ratio. At the same time, the classification main features corresponding to the sound category corresponding to the damage type with high damage resistance in the input recognition features are subjected to feature dilution with equal deviation ratio.

[0012] Furthermore, by calculating the frequency of sliding, impact, and squeezing sounds in the audio segments of each link segment, combined with the damage resistance quantitative index value corresponding to the agricultural product category to which the goods belong, the damage assessment score is calculated specifically including: The sound frequency is calculated based on the number of times the sliding, impact, and squeezing sounds appear in the audio segments of each link segment and the travel time of the link segment. The damage assessment score is calculated using the damage assessment formula, combined with the damage resistance quantitative index value corresponding to the agricultural product category to which the goods belong. The damage assessment formula is as follows: ;

[0013] In the formula, is the damage assessment score of the link segment, is the quantitative index value of damage resistance of the i-th damage, is the frequency of occurrence of the sound of the ith type of damage, is the damage type, and its value range is [1,3].

[0014] Furthermore, visual marking of link segment damage on the transport route and sending transport damage abnormality signals to specific link segments to adjust the transport density of goods specifically include: Based on the damage assessment score corresponding to each link segment, the link segment damage of the transportation line is visually marked; Set a damage assessment score abnormal threshold, filter link segments with damage assessment scores greater than or equal to the damage assessment score abnormal threshold, send a transport damage abnormal signal to the link node corresponding to the segment, and adjust the cargo transport density at the node for subsequent transportation until the damage assessment score is less than the damage assessment score abnormal threshold.

[0015] The technical effects and advantages of the monitoring and tracing system for agricultural product supply chain transportation of the present invention are as follows: The transport links are managed in segments, which enables fine-grained damage monitoring and data tracking, provides real-time feedback for the transport process of each link segment, and effectively reduces missed damage detections due to insufficient overall assessment. The sound recognition model is trained with time-frequency domain features, and can efficiently distinguish different types of damage sounds, and can flexibly adapt to various noise and interference conditions in the actual transportation environment, improving the classification accuracy and system stability. The data acquisition module monitors based on the cargo batch number, and can record the transportation process of different batches of cargo in detail, ensuring that the transportation status of each cargo can be traced back to the source. Through the link segment damage assessment and damage traceability modules, the system can accurately locate the specific link where the damage occurs, send abnormal signals in real time, and guide the adjustment of transportation density to avoid excessive damage, thereby ensuring the quality and safety of agricultural products and reducing transportation losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The present invention is a schematic diagram of the structure of a monitoring and traceability system for agricultural product supply chain transportation. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0018] Example Figure 1 The present invention provides a monitoring and tracing system for agricultural product supply chain transportation, including a damage resistance quantification module, a sound recognition model training module, a data acquisition and uploading module, a sound recognition module, a link segmentation damage assessment module, and a damage tracing module; The damage resistance quantification module conducts quantitative evaluation of the damage resistance of sliding, impact and extrusion for each agricultural product category in the supply chain; The sound recognition model training module labels the sound type of the audio signal generated by the contact between the vehicle and the cargo, and trains the sound recognition model based on the time-frequency domain characteristics of the sound signal after type labeling; The data collection and upload module decomposes the supply chain transportation route into link segments, uses the batch number of the goods as the monitoring dimension to record the agricultural product category, quantity, transportation vehicle specifications and collected audio data of the goods in each link segment, and uploads the data to the transportation link monitoring center; The sound recognition module classifies and recognizes the audio data of the transport link monitoring center through the sound recognition model; The link segment damage assessment module calculates the damage assessment score by calculating the frequency of sliding, impact, and squeezing sounds in the audio segments of each link segment and combining the damage resistance quantitative index value corresponding to the agricultural product category to which the goods belong. The damage tracing module visually marks the damage of link segments on the transportation route and sends abnormal transportation damage signals to specific link segments to adjust the transportation density of goods.

[0019] The damage resistance of each agricultural product category in the supply chain to sliding, impact and extrusion is quantitatively evaluated.

[0020] Obtain all agricultural product categories in the agricultural product supply chain, select several test samples to form test sample groups A, B, and C, and conduct damage tests based on sliding, impact, and extrusion. According to the characteristics of different agricultural products, select at least three different agricultural product categories to ensure that the sample group covers enough physical properties and appearance characteristics. Select at least three test samples for each agricultural product category to form test sample groups A (sliding), B (impact), and C (extrusion). Each test sample is a representative actual transported cargo to ensure that the test results have practical application significance.

[0021] Select test sample group A and use a sliding friction tester to simulate the sliding friction during transportation. During the simulation, the friction force is gradually increased by 0.5N increments. The surface changes of the sample are recorded each time the friction force increases until the surface of the test sample is cracked and damaged, and the instrument friction reading is recorded.

[0022] Select test sample group B for free fall experiment, and gradually increase the drop height from 0.5m to 3m, increasing by 0.5m each time, to ensure the simulation of different impact forces. The mechanical sensor is used to measure the impact force of each impact, record the change of impact force, and record the maximum reading of impact force when cracks or damage appear on the surface of the test sample.

[0023] Test sample group C was selected, and a pressure loading device was used to gradually apply pressure to simulate the extrusion during transportation. The pressure applied by the pressure loading device started from 0N and increased by 5N each time until rupture damage occurred on the surface of the test sample. The relationship between pressure and sample surface damage was recorded to ensure that the pressure increase process matched the extrusion force that might be encountered during actual transportation.

[0024] Calculate the mean of the reading records of different test samples of each agricultural product category in the sliding, impact and extrusion damage tests, normalize the mean to remove the unit and map it to the interval expression of [0,1]. The result is used as the quantitative index value of the sliding, impact and extrusion damage resistance of this agricultural product category.

[0025] The audio signals generated by the contact between the vehicle and the cargo are labeled with sound types, and the sound recognition model is trained based on the time-frequency domain characteristics of the labeled sound signals.

[0026] Multiple acoustic sensors are installed on the transport vehicle. These sensors should be arranged in areas close to the cargo to ensure that the audio signals generated when the cargo contacts the vehicle can be clearly captured. The installed sensors should have high sensitivity and a wide frequency response range to cover different types of sounds such as sliding, impact, and squeezing. The audio data collected by these sensors usually contains multiple noise sources, among which the operating noise will affect the accuracy of the sound signal. To this end, a signal processing algorithm is used to suppress the noise of the collected audio data to remove components unrelated to mechanical noise, engine noise, wind noise, etc. during the operation of the transport vehicle. Specifically, the unloaded sections in the transportation are first selected for static noise estimation, and then the spectral characteristics of the noise are extracted from these sections. Then, the noise spectrum is subtracted from the spectrum of the original audio by spectral subtraction. After the above processing, the influence of vehicle operation and environmental noise in the audio can be removed, while effectively retaining the audio information related to the contact with the cargo.

[0027] In the denoised audio data, by analyzing the time domain and frequency domain features of the audio signal, audio clips containing sliding, impact and squeezing sounds are intercepted. Each clip needs to be manually or automatically labeled as the corresponding sound type based on the characteristics of the audio. The labeling process can be completed through manual labeling or semi-automatic labeling. Manual labeling requires experts to judge and label the sound type based on the time and frequency characteristics of the audio. Automatic labeling can use the preliminary trained classification model to pre-classify the sound, which can then be verified and corrected by experts. The labeled data set will be used as the training data set for the sound classification model to provide high-quality labeled samples for subsequent model training.

[0028] Short-time Fourier transform is used to perform time-frequency domain analysis on the signal to extract the time-frequency domain features of sound signals of different types of sound clips. The time-frequency domain features include amplitude, waveform shape, duration, change rate, frequency distribution and frequency peak.

[0029] The time-frequency domain features of the annotated sound signal are used as training data sets, and machine learning algorithms (such as support vector machine (SVM), random forest, convolutional neural network (CNN), etc.) are used to train the sound classification model. Through model training, the rules of identifying different types of sounds from time-frequency domain features are learned and verified. The model is verified by comparing with the actual annotated data set to ensure the accuracy and robustness of the model in practical applications. In the verification stage, the cross-validation method can be used to avoid overfitting and ensure the generalization ability of the model. The verified sound recognition model is deployed to the transportation link monitoring center for real-time monitoring of the classification and recognition of audio data during transportation.

[0030] Decompose the supply chain transportation routes into link segments, and use the batch number of the goods as the monitoring dimension to record the agricultural product category, cargo quantity, transportation vehicle specifications and collected audio data of the goods in each link segment, and upload the data to the transportation link monitoring center.

[0031] In all transportation routes within the supply chain, first identify all transfer stations and transportation nodes, which include the transfer points of goods from one transportation link to another. Each transfer station will be marked as a link node, which are usually the places where goods are transshipped, changed vehicles, changed ships or exchanged.

[0032] According to the batch number of the goods, the actual transportation route of each batch of goods is tracked and divided into multiple link segments according to adjacent link nodes. Each link segment starts from one link node and ends at the next link node, representing a specific stage of the goods in the transportation process. Each batch of goods is identified by the batch number, and the path of each batch of goods in the transportation process is tracked in real time to ensure that the information of each link segment can be accurately recorded.

[0033] When the goods pass through each link segment, the system will record the timestamp of the goods entering and leaving the link segment in real time. The recorded timestamp information can accurately reflect the time the goods stay in each link segment. In addition to the time information, it is also necessary to record the detailed data of the agricultural product category, cargo volume and transportation vehicle specifications of the goods transported in the link segment.

[0034] The acoustic sensors on the transport vehicle collect the audio data generated by the contact between the vehicle and the goods in real time. After all the data are integrated, a corresponding relationship with the batch number of the goods is established and uploaded to the transport link monitoring center.

[0035] The audio data of the transportation link monitoring center is classified and identified through the sound recognition model.

[0036] Select the batch number of goods that need to be monitored for transportation, and obtain the agricultural product category (e.g., fruit, vegetable, grain, etc.) to which the batch of goods belongs from the transportation link monitoring center, the entry and exit timestamps of each link segment (i.e., the time information when the goods enter and exit different link nodes), and the complete audio data collected by the acoustic sensor. All information is matched one-to-one according to the batch number of the goods, ensuring that the acquired data and audio can accurately reflect the transportation process of the batch of goods.

[0037] By aligning the acquired audio data with the entry and exit timestamps of the link segments, the audio data can be divided into multiple time periods, each of which corresponds to the passage time of the goods through different link segments. Specifically, the collected complete audio data is divided into multiple audio segments according to the entry and exit time of the goods in each link segment.

[0038] The time-frequency domain features of the audio segments are extracted as the input recognition features of the sound recognition model, and the main classification features corresponding to the sliding, impacting and squeezing sounds are selected from the input recognition features.

[0039] Obtain the quantitative index value of damage resistance to sliding, impact, and extrusion corresponding to the agricultural product category to which the goods under transportation monitoring belong.

[0040] Based on the damage resistance quantitative index values ​​of various types of damage, the classification decision sensitivity of the sound recognition model is regulated by controlling the input recognition feature expression. Specifically, the damage resistance quantitative index values ​​corresponding to the agricultural product categories are used to enhance or weaken the input features, thereby indirectly achieving dynamic regulation of the classification decision sensitivity of the sound recognition model.

[0041] The final classification and recognition output results of the sound recognition model after the classification decision sensitivity adjustment are obtained, and the number of occurrences of the three sounds of sliding, impact, and squeezing in the audio segments of each link segment is recorded.

[0042] Based on the quantitative index values ​​of damage resistance of various types of damage, the classification decision sensitivity of the sound recognition model is regulated by controlling the input recognition feature expression.

[0043] Set the benchmark reference value of the sliding, impact, and extrusion damage resistance quantitative indicators (the default setting is 0.5). If a certain damage resistance quantitative indicator of the agricultural product category to which the goods belong is greater than its corresponding benchmark reference value, the goods are marked as having high damage resistance to this damage type. Otherwise, the goods are marked as having low damage resistance to this damage type. For example, watermelon has high damage resistance to sliding damage and low damage resistance to impact and extrusion damage.

[0044] For low-resistance agricultural products, a stricter classification rule is used for the sound classification of damage types such as impact, sliding, and extrusion, that is, the tolerance is lowered (sensitivity is increased to prevent missed identification). For high-resistance agricultural products, a looser classification rule is used for the sound classification of damage types such as impact, sliding, and extrusion, that is, the tolerance is increased (sensitivity is reduced to prevent misidentification).

[0045] The main characteristics of the classification of sliding damage sounds are frequency distribution, amplitude and waveform. Because the frequency distribution of sliding sounds is higher frequency components, the amplitude is relatively stable, and the waveform is usually more regular, showing a stable repetitive pattern.

[0046] The main characteristics of the classification of impact damage sound are duration, amplitude and frequency distribution. Because the impact sound is short in duration, the impact sound will have a large instantaneous amplitude and the frequency spectrum of the impact sound is usually wide.

[0047] The main characteristics of the classification of crush injury sound are amplitude, duration and frequency distribution. Because the amplitude of crush sound is large, the duration is long and it appears as a low-frequency sound.

[0048] Based on the deviation ratio of each damage resistance quantitative indicator of the agricultural product category to which the goods belong and the corresponding benchmark value, the classification main features of the sound category corresponding to the damage type with low damage resistance in the input recognition features are subjected to feature enhancement with equal deviation ratio. At the same time, the classification main features corresponding to the sound category corresponding to the damage type with high damage resistance in the input recognition features are subjected to feature dilution with equal deviation ratio.

[0049] The frequency of sliding, impact, and squeezing sounds in the audio segments of each link segment is calculated, and the damage assessment score is calculated based on the damage resistance quantitative index value corresponding to the agricultural product category to which the goods belong.

[0050] The sound frequency is calculated based on the number of times the sliding, impact, and squeezing sounds appear in the audio segments of each link segment and the travel time of the link segment. The damage assessment score is calculated using the damage assessment formula, combined with the damage resistance quantitative index value corresponding to the agricultural product category to which the goods belong. The damage assessment formula is as follows: ;

[0051] In the formula, is the damage assessment score of the link segment, is the quantitative index value of damage resistance of the i-th damage, is the frequency of occurrence of the sound of the ith type of damage, is the damage type, and its value range is [1,3].

[0052] Visually mark the link segment damage of the transportation route, and send transportation damage abnormal signals to specific link segments to adjust the transportation density of goods.

[0053] During transportation, the damage assessment score of each link segment is calculated based on the audio data and damage resistance quantitative indicators obtained in the previous steps. The damage assessment score of each link segment reflects the degree of damage suffered by the goods in the segment. By corresponding the damage assessment score to the link segment, the damage risk of each link segment in the transportation route can be intuitively displayed. In the visual annotation, each link segment is assigned a different color or mark according to its damage assessment score. The link segment with greater damage is marked in red, and the link segment with less damage is marked in green. This step uses visualization to enable managers to quickly identify which link segments have a higher risk of damage during transportation, which is convenient for subsequent monitoring and adjustment.

[0054] After completing the visual annotation of the damage assessment score, set an abnormal threshold for the damage assessment score. The abnormal threshold is obtained by multiple iterations of fitting based on regression analysis of the historical records of actual transportation damage. When the damage assessment score of a link segment is greater than or equal to the abnormal threshold, it means that the segment has a high risk of damage and immediate measures need to be taken. By screening the damage assessment scores of all link segments, those link segments with higher risks are found, and the corresponding abnormal signals are sent to the link nodes corresponding to the segments. After receiving the abnormal signal, the link node will start the subsequent emergency response mechanism, including adjusting the cargo transportation density at the node (calculated by the cargo volume and transportation vehicle specifications recorded by the transportation link monitoring center) to reduce the potential damage to the cargo. When the impact of impact and sliding damage is greater, the cargo transportation density is increased, and when the impact of extrusion damage is greater, the cargo transportation density is reduced.

[0055] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0056] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or may be transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium may be a solid-state hard disk.

[0057] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0058] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0059] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0060] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0061] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0062] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0063] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0064] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A monitoring and traceability system for agricultural product supply chain transportation, including a damage resistance quantification module, a sound recognition model training module, a data acquisition and upload module, a sound recognition module, a link segment damage assessment module, and a damage traceability module; The damage resistance quantification module conducts quantitative evaluation of the damage resistance of sliding, impact and extrusion for each agricultural product category in the supply chain; The sound recognition model training module labels the sound type of the audio signal generated by the contact between the vehicle and the cargo, and trains the sound recognition model based on the time-frequency domain characteristics of the sound signal after type labeling; The data collection and upload module decomposes the supply chain transportation route into link segments, uses the batch number of the goods as the monitoring dimension to record the agricultural product category, quantity, transportation vehicle specifications and collected audio data of the goods in each link segment, and uploads the data to the transportation link monitoring center; The sound recognition module classifies and recognizes the audio data of the transport link monitoring center through the sound recognition model; The link segment damage assessment module calculates the damage assessment score by calculating the frequency of sliding, impact, and squeezing sounds in the audio segments of each link segment and combining the damage resistance quantitative index value corresponding to the agricultural product category to which the goods belong. The damage tracing module visually marks the damage of link segments on the transportation route and sends abnormal transportation damage signals to specific link segments to adjust the transportation density of goods.

2. A monitoring and traceability system for agricultural product supply chain transportation according to claim 1, characterized in that: The quantitative evaluation of the damage resistance of sliding, impact and extrusion for each type of agricultural product in the supply chain includes: Obtain all agricultural product categories in the agricultural product supply chain, select several test samples from each product category to form test sample groups A, B, and C, and conduct damage tests based on sliding, impact, and extrusion; Select test sample group A, use a sliding friction tester to simulate the sliding friction during transportation, gradually increase the friction force on the test sample until the test sample surface is cracked and damaged, and record the instrument friction force reading; Select test sample group B to conduct a free fall experiment. Let the test samples fall freely from different heights onto the surface of the mechanical sensor until the surface of the test samples is broken and damaged, and record the sensor readings; Select test sample group C, use a pressure loading device to gradually apply pressure to simulate the squeezing during transportation, until rupture damage appears on the surface of the test sample, and record the loading pressure reading; Calculate the mean of the reading records of different test samples of each agricultural product category in the sliding, impact and extrusion damage tests, normalize the mean to remove the unit and map it to the interval expression of [0,1]. The result is used as the quantitative index value of the sliding, impact and extrusion damage resistance of this agricultural product category.

3. A monitoring and traceability system for agricultural product supply chain transportation according to claim 2, characterized in that: The audio signal generated by the contact between the vehicle and the cargo is labeled with the sound type, and the sound recognition model is trained based on the time-frequency domain characteristics of the sound signal after type labeling. Specifically, the following steps are performed: Acoustic sensors are placed on transport vehicles to collect audio data generated by the contact between the vehicle and the cargo. The audio data is processed using signal processing algorithms to remove the operating noise of the vehicle. Extract the sound clips containing sliding, impact and squeezing sounds from the denoised audio, and label the sound types; Short-time Fourier transform is used to analyze the signal in the time and frequency domain to extract the time and frequency domain features of the sound signals of different types of sound clips. The time and frequency domain features include amplitude, waveform shape, duration, change rate, frequency distribution and frequency peak value. A sound recognition model for the transportation process is established. The labeled time-frequency domain features of the sound signal are used as the training data set. The sound recognition model is trained and verified through a machine learning algorithm. The verified sound recognition model is deployed to the transportation link monitoring center.

4. A monitoring and traceability system for agricultural product supply chain transportation according to claim 3, characterized in that: Decompose the supply chain transportation route into link segments, use the batch number of the goods as the monitoring dimension to record the agricultural product category, cargo quantity, transportation vehicle specifications and collected audio data of the goods in each link segment, and upload the data to the transportation link monitoring center. Specifically include: Mark the transport transfer stations in all transport routes in the supply chain as link nodes; Taking the batch number of goods as the monitoring dimension, the actual transportation route of each batch of goods is obtained, and the transportation route of the same batch of goods is split into link segments based on adjacent link nodes; When the same batch of goods passes through a specific link segment, the timestamps of entering and leaving the link segment, as well as the agricultural product category, quantity and transport vehicle specifications of the goods transported in the link segment are recorded; The acoustic sensors on the transport vehicle collect the audio data generated by the contact between the vehicle and the goods in real time. After all the data are integrated, a corresponding relationship with the batch number of the goods is established and uploaded to the transport link monitoring center.

5. A monitoring and traceability system for agricultural product supply chain transportation according to claim 4, characterized in that: The audio data of the transport link monitoring center is classified and identified through the sound recognition model, including: Select the batch number of goods that need to be monitored for transportation, and obtain the agricultural product category of the goods corresponding to the batch number, the entry and exit timestamps of each link segment, and the complete audio data collected from the transportation link monitoring center; Dividing the audio data into audio segments aligned with the link segment transit time periods based on the link segment ingress and egress timestamps; Extract the time-frequency domain features of the audio segments as input recognition features of the sound recognition model, and select the main classification features corresponding to the sliding, impacting, and squeezing sounds from the input recognition features; Obtain the sliding, impact, and extrusion damage resistance quantitative index values ​​corresponding to the agricultural product category to which the transport monitored goods belong; Based on the damage resistance quantitative index values ​​of various types of damage, the classification decision sensitivity of the sound recognition model is regulated by controlling the input recognition feature expression; The final classification and recognition output results of the sound recognition model after the classification decision sensitivity adjustment are obtained, and the number of occurrences of the three sounds of sliding, impact, and squeezing in the audio segments of each link segment is recorded.

6. A monitoring and traceability system for agricultural product supply chain transportation according to claim 5, characterized in that: Based on the damage resistance quantitative index values ​​of various types of damage, the classification decision sensitivity of the sound recognition model is regulated by controlling the input recognition feature expression. Specifically, the following are included: Set the benchmark reference values ​​of the quantitative indicators of sliding, impact and extrusion damage resistance. If the quantitative indicator of damage resistance of a certain agricultural product category to which the goods belong is greater than its corresponding benchmark reference value, the goods are marked as having high damage resistance to the damage type. Otherwise, the goods are marked as having low damage resistance to the damage type. Based on the deviation ratio of each damage resistance quantitative indicator of the agricultural product category to which the goods belong and the corresponding benchmark value, the classification main features of the sound category corresponding to the damage type with low damage resistance in the input recognition features are subjected to feature enhancement with equal deviation ratio. At the same time, the classification main features corresponding to the sound category corresponding to the damage type with high damage resistance in the input recognition features are subjected to feature dilution with equal deviation ratio.

7. A monitoring and traceability system for agricultural product supply chain transportation according to claim 6, characterized in that: By calculating the frequency of sliding, impact, and squeezing sounds in the audio segments of each link segment, combined with the damage resistance quantitative index value corresponding to the agricultural product category to which the goods belong, the damage assessment score is calculated specifically including: The sound frequency is calculated based on the number of times the sliding, impact, and squeezing sounds appear in the audio segments of each link segment and the travel time of the link segment. The damage assessment score is calculated using the damage assessment formula, combined with the damage resistance quantitative index value corresponding to the agricultural product category to which the goods belong. The damage assessment formula is as follows: ； In the formula, is the damage assessment score of the link segment, is the quantitative index value of damage resistance of the i-th damage, is the frequency of occurrence of the sound of the ith type of damage, is the damage type, and its value range is [1,3].

8. A monitoring and traceability system for agricultural product supply chain transportation according to claim 7, characterized in that: Visually mark the link segment damage of the transportation route and send the transportation damage abnormal signal to the specific link segment to adjust the transportation density of the goods. Specifically include: Based on the damage assessment score corresponding to each link segment, the link segment damage of the transportation line is visually marked; Set a damage assessment score abnormal threshold, filter link segments with damage assessment scores greater than or equal to the damage assessment score abnormal threshold, send a transport damage abnormal signal to the link node corresponding to the segment, and adjust the cargo transport density at the node for subsequent transportation until the damage assessment score is less than the damage assessment score abnormal threshold.

Images