Product anti-counterfeiting traceability monitoring system and method based on Internet of Things

Abstract

The embodiment of the invention provides a product anti-counterfeiting traceability monitoring system and method based on the Internet of Things, and belongs to the technical field of data traceability. The system comprises an exception handling module which performs exception identification on first data by using a platform layer to obtain first exception data; performing exception identification on the second data to obtain second exceptional data, and determining a target quality type according to the first exceptional data and the second exceptional data; the false detection module performs false detection on the first information according to the first data by using the platform layer to obtain a target detection result; the data association module uses the platform layer to correct the first information according to the target quality type and the target detection result to obtain second information, and associates the first abnormal data with the second abnormal data according to the second information to obtain related traceability information; and the data monitoring module obtains a target query identifier corresponding to the target user by using the application layer, and obtains third information and target traceability information corresponding to the target query identifier in the database according to the target query identifier.

Description

Technical Field

[0001] The present invention relates to the technical field of data traceability, and in particular to a product anti-counterfeiting traceability monitoring system and method based on the Internet of Things. Background Art

[0002] In modern society, with the intensification of globalization and market competition, product quality and safety have become the focus of consumers and enterprises. However, most of the existing product anti-counterfeiting technologies rely on simple identification methods such as barcodes and QR codes. Although these technologies are easy to implement, they have the defects of being easy to copy and tamper with, and cannot effectively provide full traceability information of products.

[0003] With the rapid development of technologies such as the Internet of Things, big data, and cloud computing, intelligent anti-counterfeiting and traceability technologies based on the Internet of Things have gradually become an important means to solve this problem. Through technologies such as RFID, sensors, and GPS, relevant data on products in the production, transportation, and sales stages can be collected and monitored in real time to ensure the authenticity and traceability of products. However, the existing anti-counterfeiting and traceability systems based on the Internet of Things still have the following technical problems: First, it is impossible to ensure the accuracy and completeness of the data collected during large-scale production and transportation. Second, it is impossible to promptly identify and handle abnormal situations in the production and transportation process in order to take effective countermeasures. Third, it is impossible to prevent falsified information and false detection to ensure the authenticity of product information. Finally, it is impossible to effectively associate and store data from different sources so that subsequent queries and traceability can provide a complete and reliable basis.

[0004] Therefore, there is an urgent need for a product anti-counterfeiting traceability monitoring system based on the Internet of Things that can comprehensively improve the accuracy, real-time and security of traceability information to solve at least one of the above technical problems. Summary of the invention

[0005] The main purpose of the embodiments of the present invention is to provide a product anti-counterfeiting traceability monitoring system and method based on the Internet of Things, aiming to solve the problem in the related technology that the anti-counterfeiting traceability process cannot timely identify and handle abnormal situations in the production and transportation process, which leads to inaccurate traceability results.

[0006] In a first aspect, an embodiment of the present invention provides a product anti-counterfeiting traceability monitoring system based on the Internet of Things, comprising: A data collection module, used to determine first information according to a target identifier of a target object using a perception layer of the Internet of Things, and to collect first data of the target object in a production process and second data of the target object in a transportation process according to the perception layer; an exception handling module, configured to use the platform layer of the Internet of Things to perform exception identification on the first data to obtain first exception data; perform exception identification on the second data to obtain second exception data, and determine the target quality type corresponding to the target object according to the first exception data and the second exception data; a false detection module, configured to use the platform layer to perform false detection on the first information according to the first data to obtain a target detection result; a data association module, configured to use the platform layer to modify the first information according to the target quality type and the target detection result to obtain second information, and to associate the first abnormal data with the second abnormal data according to the second information to obtain relevant traceability information corresponding to the second information; A data storage module, used to associate the target identifier, the second information and the related traceability information with each other using the platform layer and store them in a database; The data monitoring module is used to obtain a target query identifier corresponding to a target user by using the application layer of the Internet of Things, and obtain third information and target traceability information corresponding to the target query identifier in the database according to the target query identifier.

[0007] In a second aspect, an embodiment of the present invention provides a product anti-counterfeiting traceability monitoring method based on the Internet of Things, comprising: Determine first information according to a target identifier of a target object using a perception layer of the Internet of Things, and collect first data of the target object in a production process and second data of the target object in a transportation process according to the perception layer; Using the platform layer of the Internet of Things to perform abnormality identification on the first data to obtain first abnormal data; performing abnormality identification on the second data to obtain second abnormal data, and determining the target quality type corresponding to the target object according to the first abnormal data and the second abnormal data; Using the platform layer to perform false detection on the first information according to the first data to obtain a target detection result; Using the platform layer to modify the first information according to the target quality type and the target detection result to obtain second information, and correlating the first abnormal data with the second abnormal data according to the second information to obtain relevant traceability information corresponding to the second information; Using the platform layer, the target identifier, the second information and the related traceability information are associated and stored in a database; The target query identifier corresponding to the target user is obtained by utilizing the application layer of the Internet of Things, and the third information and target tracing information corresponding to the target query identifier are obtained in the database according to the target query identifier.

[0008] The embodiment of the present invention provides a product anti-counterfeiting traceability monitoring system and method based on the Internet of Things, the system includes: a data acquisition module uses the perception layer of the Internet of Things to determine the first information according to the target identification of the target object, and can accurately focus on a specific target object. At the same time, the first data of the production process and the second data of the transportation process are collected respectively, so as to achieve data coverage of the entire process of the target object from production to transportation. This helps to fully understand the state of the target object at different stages, and provides a rich and accurate data basis for subsequent quality assessment, abnormality detection, etc., and then the abnormality processing module performs abnormal identification on the first data and the second data respectively, and can timely discover abnormal situations occurring during production and transportation. By determining the target quality type corresponding to the target object, the product quality can be effectively evaluated and classified, and then the false detection module performs false detection on the first information according to the first data, which can effectively prevent data falsification, so that by obtaining the target detection result, it can be ensured that the subsequent analysis and processing based on this information are reliable. Therefore, the data association module corrects the first information according to the target quality type and the target detection result to obtain the second information. Thus, the first abnormal data and the second abnormal data are associated with the corrected second information to form relevant traceability information, and effective data integration is achieved. This helps to establish a comprehensive and systematic information system, facilitating subsequent data analysis and traceability. The data storage module then associates the target identification, the second information and the relevant traceability information and stores them in the database, thus realizing the long-term preservation of the data. Finally, the data monitoring module enables the target user to use the application layer of the Internet of Things to obtain the corresponding third information and target traceability information through the target query identification, so that the target user can obtain accurate and authentic traceability information under the target query identification, and also solves the problem in the related technology that the anti-counterfeiting traceability process cannot timely identify and handle abnormal situations in the production and transportation process, which leads to inaccurate traceability results. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0010] Figure 1 A schematic diagram of the module structure of a product anti-counterfeiting traceability monitoring system based on the Internet of Things provided by an embodiment of the present invention; Figure 2 A flowchart of a product anti-counterfeiting traceability monitoring method based on the Internet of Things is provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0011] 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 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.

[0012] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0013] It should be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0014] The embodiment of the present invention provides a product anti-counterfeiting traceability monitoring system and method based on the Internet of Things. The product anti-counterfeiting traceability monitoring system based on the Internet of Things can be applied to a terminal device, which can be an electronic device such as a tablet computer, a laptop computer, a desktop computer, a personal digital assistant, and a wearable device. The terminal device can be a server or a server cluster.

[0015] Some embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0016] Please refer to Figure 1 , Figure 1 A schematic diagram of the module structure of a product anti-counterfeiting traceability monitoring system based on the Internet of Things provided in an embodiment of the present invention.

[0017] like Figure 1As shown, the product anti-counterfeiting traceability monitoring system 100 based on the Internet of Things includes a data acquisition module 101, an exception handling module 102, a false detection module 103, a data association module 104, a data storage module 105, and a data monitoring module 106, wherein the data acquisition module 101 is used to determine the first information according to the target identification of the target object by using the perception layer of the Internet of Things, and collect the first data of the target object in the production process and the second data in the transportation process according to the perception layer; the exception handling module 102 is used to use the platform layer of the Internet of Things to perform anomaly identification on the first data to obtain first abnormal data; perform anomaly identification on the second data to obtain second abnormal data, and determine the target quality type corresponding to the target object according to the first abnormal data and the second abnormal data; the false detection module 103 , used to use the platform layer to perform false detection on the first information according to the first data to obtain a target detection result; a data association module 104, used to use the platform layer to correct the first information according to the target quality type and the target detection result to obtain second information, and associate the first abnormal data with the second abnormal data according to the second information to obtain relevant traceability information corresponding to the second information; a data storage module 105, used to use the platform layer to associate the target identifier, the second information and the relevant traceability information and store them in a database; a data monitoring module 106, used to use the application layer of the Internet of Things to obtain a target query identifier corresponding to the target user, and obtain third information and target traceability information corresponding to the target query identifier in the database according to the target query identifier.

[0018] Exemplarily, the target identifier is a unique identifier such as an RFID tag or a QR code affixed to the target object when it is sold. Then, the data acquisition module 101 uses the perception layer of the Internet of Things to obtain the first information corresponding to the target object from the database according to the target identifier of the target object. The first information is the basic information corresponding to the target object, including but not limited to the production time, production address, target type, target quality grade, etc. corresponding to the target object, that is, the first information is the basic information describing the target object.

[0019] Exemplarily, at the same time, the data acquisition module 101 uses an image sensor to collect first data of the target object in the production process according to the perception layer, and the first data is the image data corresponding to the target object in the production process, and uses a temperature and humidity sensor to collect the environmental parameters of the target object in the transportation process in real time to obtain the corresponding second data, that is, the second data is the environmental parameter information corresponding to the target object in the transportation process.

[0020] Exemplarily, a data access interface is established at the platform layer of the Internet of Things to ensure that the first data and the second data can be accurately and timely transmitted to the exception handling module 102. The first data contains image data related to the operation, which contains rich operation information, and then the image in the first data is preprocessed, including operations such as image cropping, scaling, and normalization to improve the quality of the image data. Then, features such as color features, texture features, and shape features are extracted from the preprocessed image. Finally, the extracted features are input into the trained machine learning model for classification to determine whether the image has an operation abnormality, thereby obtaining the corresponding first abnormal data. The second data mainly includes environmental parameter data, such as temperature, humidity, and air pressure. In order to identify data in which environmental parameter abnormalities exist, it is necessary to make a judgment based on a pre-established abnormal identification rule, and then compare the environmental parameter data in the second data with the abnormal identification rule. When a certain environmental parameter data does not meet the rule, it is marked as abnormal data, thereby obtaining the corresponding second abnormal data.

[0021] Exemplarily, a large amount of historical data is collected, including first abnormal data and corresponding quality grade labels. Then, these data are preprocessed and feature extracted, and input into the model for training. During the training process, the model parameters are continuously adjusted so that the model can learn the relationship between the first abnormal data and the quality grade. Thus, the current first abnormal data is input into the trained quality grade prediction model for prediction, and the first quality type is determined according to the prediction result. For example, if the prediction result shows that the quality grade of the target object is high, the first quality type can be defined as "high quality"; if the prediction result shows that the quality grade is low, the first quality type can be defined as "defective", etc. On the basis of obtaining the first quality type, the second abnormal data is used to adjust its quality, so as to obtain the target quality type corresponding to the target object. The quality adjustment rule needs to comprehensively consider the influence of the first quality type and the second abnormal data. For example, if the first quality type is "high quality", but the second abnormal data shows that the environmental parameters are seriously abnormal, it may have a certain impact on the quality of the target object. At this time, the level of the target quality type can be appropriately reduced, so that according to the quality adjustment rule, the first quality type is adjusted in combination with the first quality type and the second abnormal data, so as to determine the target quality type corresponding to the target object.

[0022] Exemplarily, the false detection module 103 uses the platform layer to perform attribute recognition on the first data to obtain target attribute information corresponding to the target object, and then compares the target attribute information with the first information, thereby performing false detection on the first information according to the comparison result to obtain a target detection result.

[0023] Exemplarily, the data association module 104 compares the target quality type with the quality type in the first information, and when the level of the quality type in the first information is greater than the target quality type, the quality type in the first information is modified. At the same time, the information in the target detection result that is inconsistent with the first information is modified accordingly according to the target attribute information to obtain the second information, and the second information is more accurate basic information after the production data and transportation data are corrected. In addition, the first abnormal data and the second abnormal data in the process of determining the target quality type in the second information are associated to facilitate subsequent data traceability, and then the first abnormal information and the second abnormal information are determined as relevant traceability information corresponding to the second information.

[0024] Exemplarily, the data storage module 105 uses the platform layer to associate the target identifier, the second information and the related traceability information and store them in the database.

[0025] Exemplarily, the data monitoring module 106 utilizes the application layer of the Internet of Things to receive the target query identifier corresponding to the target user, thereby obtaining the third information and target traceability information corresponding to the target query identifier after comparing the target query identifier with the target identifier in the database.

[0026] In some embodiments, the abnormality handling module includes: a data acquisition module, which is used to collect normal image data corresponding to the target object during the production process using the platform layer, and obtain the abnormal description text corresponding to the target object based on expert experience; an image generation module, which is used to use the platform layer to perform abnormal adjustment on the normal image data according to the abnormal description text to obtain the abnormal image data corresponding to the target object; a model determination module, which is used to use the platform layer to determine the target abnormality recognition model corresponding to the target object according to the normal image data and the abnormal image data; an abnormality recognition module, which is used to use the platform layer to perform abnormality recognition on the first data according to the target abnormality recognition model to obtain the first abnormal data; a data analysis module, which is used to use the platform layer to obtain normal sensor data corresponding to the target object during transportation, and perform data association analysis based on the normal sensor data and the second data to obtain the second abnormal data; a damage determination module, which is used to use the platform layer to obtain frequency information and time information corresponding to the second abnormal data, and determine the target damage parameter corresponding to the target object during transportation according to the second abnormal data, the frequency information and the time information; a quality determination module, which is used to use the platform layer to determine the target quality type corresponding to the target object according to the first abnormal data and the target damage parameter.

[0027] For example, a stable connection channel with various image acquisition devices (such as industrial cameras, surveillance cameras, etc.) in the production process is established at the platform layer, and the normal image data of the target object in the production process is obtained regularly or in real time through the data interface. Experts use their rich experience and in-depth understanding of the production process of the target object to describe in detail possible abnormal situations, such as product surface defects, dimensional deviations, etc., and organize these descriptions into standardized abnormal description texts.

[0028] Exemplarily, the image generation module analyzes the image feature change rules corresponding to different abnormal situations based on the abnormal description text. For example, if the abnormal description is "there are scratches on the surface of the product", a strategy is formulated to simulate the scratch effect on the normal image, such as adding lines, changing local colors, etc., and then the image processing algorithm is used to process the normal image data according to the abnormal adjustment strategy to generate abnormal image data corresponding to the target object.

[0029] Exemplarily, the model determination module divides the normal image data and the abnormal image data into a training set and a test set according to a certain ratio (such as 7:3). The training set is used to train the abnormal recognition model, and the test set is used to evaluate the performance of the model, and then use machine learning or deep learning models, such as convolutional neural networks (CNN). The model is trained using the training set, and the parameters of the model are continuously adjusted so that it can accurately distinguish between normal images and abnormal images. The trained model is evaluated using the test set, and the model's accuracy, recall rate, F1 value and other indicators are calculated. The model is optimized according to the evaluation results, such as adjusting the model structure, increasing training data, etc., until the model achieves satisfactory performance and is determined as the target abnormal recognition model.

[0030] Exemplarily, the abnormality recognition module inputs the first data into the target abnormality recognition model, and the model classifies the image according to the learned features and patterns to determine whether there is an abnormality, and marks the image data identified as abnormal as the first abnormal data.

[0031] Exemplarily, the data analysis module establishes connections with various sensors (such as temperature sensors, humidity sensors, vibration sensors, etc.) during transportation at the platform layer, and collects normal sensor data of the target object during transportation in real time. Perform statistical analysis on the normal sensor data to determine the normal range and fluctuation pattern of each parameter, and then use statistical analysis methods and machine learning algorithms to compare and correlate the second data with the normal sensor data to detect whether there are data points in the second data that deviate greatly from the normal range. These data points that deviate from the normal range are marked as abnormal data, i.e., second abnormal data.

[0032] Exemplarily, the corresponding frequency information and time information are extracted from the second abnormal data. The frequency information reflects the frequency of the abnormal situation, and the time information records the specific time and duration of the abnormal situation. Thus, a damage assessment model is established based on the second abnormal data, the frequency information and the time information. For example, a weighted summation method can be used to quantify and comprehensively evaluate different factors to determine the target damage parameters corresponding to the target object during transportation.

[0033] Exemplarily, the quality determination module inputs the current first abnormal data into the trained quality grade prediction model for prediction, determines the first quality type according to the prediction result, and then adjusts the quality of the first quality type using the target damage parameter on the basis of the first quality type, thereby obtaining the target quality type corresponding to the target object. The quality adjustment rule needs to comprehensively consider the influence of the first quality type and the target damage parameter.

[0034] Specifically, the target quality type is determined by comprehensively considering the first abnormal data and the target damage parameter, making the quality assessment more comprehensive and accurate. Traditional quality assessment may only focus on one aspect of the production process or the transportation process, while this method takes factors of both processes into consideration and can more truly reflect the actual quality status of the target object.

[0035] In some embodiments, the image generation module includes: a text feature extraction module, which is used to use the text feature extraction network of the image generation model in the platform layer to extract features from the abnormal description text to obtain first text feature information; an image feature extraction module, which is used to use the image feature extraction network of the image generation model to extract features from the normal image data to obtain first image feature information; a feature fusion module, which is used to use the parameter sharing network of the image generation model to perform feature fusion on the first text feature information and the first image feature information to obtain second text feature information and second image feature information; and an association analysis module, which is used to use the image-text analysis network of the image generation model to perform an association degree analysis on the second text feature information and the second image feature information. Obtaining a picture-text association value, and determining a corresponding picture-text association matrix according to the picture-text association value; an anomaly detection module, used to use the anomaly recognition network of the image generation model to determine a segmentation threshold using an optimization algorithm according to the picture-text association matrix, and determine the image anomaly features corresponding to the normal image data under the anomaly description text and the text anomaly features corresponding to the anomaly description text under the normal image data according to the segmentation threshold and the picture-text association matrix; an initial generation module, used to use the image generation network of the image generation model to generate an image according to the text anomaly features and the image anomaly features to obtain initial image data; an image fusion module, used to fuse the initial image data with the normal image data to obtain the abnormal image data corresponding to the target object.

[0036] Exemplarily, the text feature extraction module selects a text feature extraction network in a suitable image generation model at the platform layer, such as a network based on the Transformer architecture, initializes and configures the network, including setting hyperparameters such as learning rate and batch size to ensure that it can run efficiently, and then inputs the abnormal description text into the configured text feature extraction network to obtain the first text feature information.

[0037] Exemplarily, an image feature extraction network in an image generation model, such as a convolutional neural network (CNN), is selected, and then normal image data is input into the image feature extraction network. The network extracts image features such as edges, textures, and colors through operations such as convolution and pooling, and finally outputs the first image feature information.

[0038] Exemplarily, the feature fusion module constructs a parameter sharing network of the image generation model, and the function of the network is to input the first text feature information and the first image feature information into the parameter sharing network. The network will interact and integrate the two features according to its internal parameters and structure, thereby obtaining the second text feature information and the second image feature information.

[0039] Exemplarily, the association analysis module uses a graph-text analysis network of an image generation model, which can be constructed based on an attention mechanism, etc., to better analyze the degree of association between text and image features, and then input the second text feature information and the second image feature information into the graph-text analysis network. The network calculates the similarity or correlation between the text and image features to obtain a graph-text association value. Based on these association values, a corresponding graph-text association matrix is ​​constructed, and the elements in the matrix represent the strength of association between different text and image features.

[0040] Exemplarily, the anomaly detection module uses an optimization algorithm, such as a genetic algorithm, a simulated annealing algorithm, etc., to determine the segmentation threshold, thereby finding the optimal segmentation threshold based on the image-text association matrix so that the abnormal features can be accurately identified, and then analyzing the normal image data and the abnormal description text according to the determined segmentation threshold and the image-text association matrix. The portion exceeding the segmentation threshold is determined as an abnormal image feature and an abnormal text feature, corresponding to the abnormal portion of the normal image data under the abnormal description text and the abnormal portion of the abnormal description text under the normal image data, respectively.

[0041] Exemplarily, the initial generation module trains the image generation network of the image generation model so that it can learn the mapping relationship between abnormal features and image generation, and then inputs the text abnormal features and image abnormal features into the trained image generation network. The network generates corresponding initial image data based on these abnormal features, so that the abnormal part of the normal image data under the abnormal description text is generated according to the abnormal part of the abnormal description text under the normal image data.

[0042] Exemplarily, the image fusion module performs weighted fusion on the initial image data and the abnormal part of the normal image data under the abnormal description text to obtain the abnormal image data corresponding to the target object.

[0043] In some embodiments, the model determination module includes: a data encoding module, used to encode the normal image data using the first encoder of the target abnormality recognition model to obtain a first feature representation, and simultaneously encode the abnormal image data using the second encoder of the target abnormality recognition model to obtain a second feature representation, wherein the network structure between the first encoder and the second encoder is the same and the network parameters are shared; an information reconstruction module, used to use the information reconstruction layer of the target abnormality recognition model to perform feature dimensionality reduction on the first feature representation to obtain a first low-dimensional representation and to perform feature dimensionality reduction on the second feature representation to obtain a second low-dimensional representation, then perform a high-dimensional restoration operation on the first low-dimensional representation to obtain a first high-dimensional representation and perform a high-dimensional restoration operation on the second low-dimensional representation to obtain a second high-dimensional representation, and then perform similarity calculation on the first feature representation and the first high-dimensional representation to obtain a first similarity value and perform similarity calculation on the first feature representation and the second high-dimensional representation to obtain a second similarity value, and train the information reconstruction layer according to the first similarity value and the second similarity value to obtain the information reconstruction layer corresponding to the information reconstruction layer information reconstruction parameters, and adjusting the first high-dimensional representation according to the information reconstruction parameters to obtain a third high-dimensional representation and adjusting the second high-dimensional representation to obtain a fourth high-dimensional representation; a feature processing module, used to use the feature processing layer of the target abnormality recognition model to perform feature compression on the third high-dimensional representation to obtain a third feature representation and to perform feature compression on the fourth high-dimensional representation to obtain a fourth feature representation; a data decoding module, used to use the first decoder of the target abnormality recognition model to decode the third feature representation to obtain a fifth feature representation and to decode the fourth feature representation to obtain a sixth feature representation; a loss calculation module, used to use the difference calculation layer of the target abnormality recognition model to perform difference calculation on the fifth feature representation and the sixth feature representation to obtain the corresponding target difference value between the abnormal image data and the normal image data, and determine the loss function between the abnormal image data and the normal image data according to the target difference value; a parameter adjustment module, used to adjust the parameters of the target abnormality recognition model according to the loss function until the corresponding target abnormality recognition model that meets the preset conditions is obtained.

[0044] Exemplarily, the data encoding module determines the first encoder and the second encoder in the target anomaly recognition model, and the two have the same network structure and share parameters, and then inputs the normal image data into the first encoder, and the encoder performs a series of transformation and feature extraction operations on the image data to convert it into a first feature representation. At the same time, the abnormal image data is input into the second encoder, and due to parameter sharing, the second encoder encodes the abnormal image data in the same way to obtain a second feature representation.

[0045] Exemplarily, the information reconstruction module uses the information reconstruction layer of the target anomaly recognition model to perform a feature dimensionality reduction operation on the first feature representation and the second feature representation to convert the high-dimensional feature representation into a low-dimensional representation, respectively obtaining a first low-dimensional representation and a second low-dimensional representation, and then performing a high-dimensional restoration operation on the first low-dimensional representation and the second low-dimensional representation. A decoder or a deconvolution network or other method can be used to restore the low-dimensional representation to a high-dimensional representation to obtain a first high-dimensional representation and a second high-dimensional representation, thereby calculating the similarity between the first feature representation and the first high-dimensional representation to obtain a first similarity value; and calculating the similarity between the first feature representation and the second high-dimensional representation to obtain a second similarity value. According to these two similarity values, an optimization algorithm (such as stochastic gradient descent) is used to train the information reconstruction layer and adjust its parameters until a satisfactory effect is achieved, thereby obtaining information reconstruction parameters corresponding to the information reconstruction layer. Thus, according to the obtained information reconstruction parameters, the first high-dimensional representation and the second high-dimensional representation are adjusted to obtain a third high-dimensional representation and a fourth high-dimensional representation, respectively.

[0046] Exemplarily, the feature processing module determines the feature processing layer of the target anomaly recognition model, which can adopt a fully connected layer, a pooling layer and other structures. The feature processing layer is configured to set its parameters and hyperparameters. The third high-dimensional representation and the fourth high-dimensional representation are respectively input into the feature processing layer to perform feature compression operations. By reducing the dimension and number of features, the third feature representation and the fourth feature representation are obtained, which can reduce data redundancy and improve the efficiency and generalization ability of the model.

[0047] Exemplarily, the data decoding module selects a first decoder of the target anomaly recognition model, and the structure and function of the decoder correspond to the encoder. The decoder is set, including initializing parameters and setting hyperparameters. The third feature representation is input into the first decoder to obtain the fifth feature representation, and the fourth feature representation is input into the first decoder to obtain the sixth feature representation.

[0048] Exemplarily, the loss calculation module utilizes the difference layer of the target anomaly recognition model and uses methods such as mean square error (MSE) and cross entropy loss to perform difference calculations on the fifth feature representation and the sixth feature representation to obtain a target difference value between the abnormal image data and the normal image data, thereby determining the loss function between the abnormal image data and the normal image data based on the target difference value, and obtaining the corresponding loss value under the loss function.

[0049] Exemplarily, the parameter adjustment module uses the stochastic gradient descent method combined with the loss value to adjust the parameters of the target anomaly recognition model. In each iteration, the value of the loss function is calculated until the preset conditions are met (such as the loss function converges to a smaller value, reaches the maximum number of iterations, etc.), thereby obtaining the corresponding target anomaly recognition model that meets the preset conditions.

[0050] Specifically, by encoding, reconstructing, compressing and decoding normal image data and abnormal image data, the model can learn the difference features between normal images and abnormal images more deeply. These difference features can help the model identify abnormal images more accurately and improve the accuracy of abnormal identification.

[0051] In some embodiments, the data analysis module includes: a window segmentation module, which is used to determine a target window and cut the second data according to the target window to obtain multiple initial window data; a matrix conversion module, which is used to determine a first matrix corresponding to the normal sensor data under multidimensional data and a second matrix corresponding to the initial window data under the same multidimensional data; an association processing module, which is used to perform matrix similarity calculation based on the first matrix and the second matrix to obtain the data association degree between the normal sensor data and the initial window data, and determine the data difference degree between the normal sensor data and the initial window data according to the data association degree; a distance calculation module, which is used to calculate the normal sensor data. a data distance between the sensing data and the initial window data, and then fusing the data distance, the data correlation and the data difference to determine the target correlation between the normal sensing data and the initial window data; a distribution analysis module, used to perform data distribution analysis on the normal sensing data to obtain first distribution information and to perform data distribution analysis on the initial window data to obtain second distribution information; a difference determination module, used to fuse the first distribution information and the second distribution information according to the target correlation to determine the correlation difference value between the initial window data and the normal sensing data; a data screening module, used to obtain the second abnormal data from the initial window data according to the correlation difference value.

[0052] Exemplarily, the window segmentation module determines the size and step length of the target window based on expert experience or historical experience, and cuts the second data based on the determined target window. Starting from the starting position of the data, the data is intercepted in sequence according to the size and step length of the target window, thereby obtaining multiple initial window data.

[0053] Exemplarily, the multi-dimensional data in the matrix conversion module may include different sensing indicators such as temperature, humidity, and pressure, and then the normal sensing data is sorted and arranged according to the determined multi-dimensional data dimensions to construct a corresponding first matrix. Similarly, each initial window data is also constructed into a second matrix according to the same multi-dimensional data dimensions.

[0054] Exemplarily, the association processing module calculates the first matrix and the second matrix using a matrix similarity calculation method such as cosine similarity, Euclidean distance, etc. to obtain the data association degree between the normal sensor data and the initial window data. Thus, the data difference degree between the normal sensor data and the initial window data is determined according to the calculated data association degree. For example, the data difference degree can be obtained by subtracting the data association degree from 1.

[0055] Exemplarily, the distance calculation module calculates the data distance between the normal sensing data and the initial window data using a data distance measurement method such as Manhattan distance. Thus, different weights are assigned to the data distance, data relevance, and data difference according to the importance and actual situation of the data, and then they are fused by weighted summation and the like to determine the target relevance between the normal sensing data and the initial window data.

[0056] Exemplarily, the distribution analysis module obtains first distribution information corresponding to the normal sensor data and second distribution information corresponding to the initial window data by performing histogram analysis or kernel density estimation on the normal sensor data or the initial window data. Exemplarily, the difference determination module performs weighted averaging on the first distribution information and the second distribution information in combination with the target correlation degree, thereby calculating the correlation difference value between the initial window data and the normal sensing data through the weighted average data, which value can comprehensively reflect the degree of difference between the two.

[0057] Exemplarily, the data screening module sets an outlier threshold according to historical experience, and then compares the associated difference value of each initial window data with the outlier threshold, and screens out the initial window data whose associated difference value exceeds the outlier threshold as the second abnormal data.

[0058] In some embodiments, the difference determination module includes: a distribution difference analysis module, which is used to calculate the difference between the first distribution information and the second distribution information to obtain a target distribution difference; a spatial analysis module, which is used to determine the spatial distribution difference between the initial window data and the normal sensor data according to the target distribution difference combined with the target association degree and the data distance; a difference calculation module, which is used to determine the association difference value between the initial window data and the normal sensor data according to the spatial distribution difference and the target distribution difference; wherein the association difference value is obtained according to the following formula: ; ; in, represents the spatial distribution difference between the initial window data and the normal sensing data, represents the target distribution difference between the first distribution information and the second distribution information, represents the target relevance, Indicates the normal sensing data and the initial window data The data distance between them, exp represents the exponential function, and R1 represents the normal sensing data The corresponding first matrix is Represents the initial window data The corresponding second matrix is, represents the correlation difference value between the initial window data and the normal sensor data, n represents the total data volume corresponding to the union result obtained by performing union calculation on the normal sensor data and the initial window data, abs represents taking the absolute value, represents the first distribution information corresponding to the i-th union result under the normal sensing data, Represents the second distribution information corresponding to the i-th union result under the initial window data.

[0059] Exemplarily, the distribution difference analysis module calculates the absolute value or square value of the difference between the first distribution information and the second distribution information to obtain a target distribution difference, which can reflect the degree of difference between the two data distributions in terms of values.

[0060] Exemplarily, the spatial analysis module calculates the spatial distribution difference between the initial window data and the normal sensing data by fusing the target distribution difference, target correlation and data distance according to the following formula. This difference can reflect the difference between the two sets of data from a more comprehensive spatial perspective: ; in, represents the spatial distribution difference between the initial window data and the normal sensing data, represents the target distribution difference between the first distribution information and the second distribution information, represents the target relevance, Indicates the normal sensing data and the initial window data The data distance between them, exp represents the exponential function, and R1 represents the normal sensing data The corresponding first matrix is Represents the initial window data The corresponding second matrix.

[0061] Exemplarily, the calculation difference module integrates the spatial distribution difference and the target distribution difference according to the influence mechanism of the spatial distribution difference and the target distribution difference on the associated difference value using the following formula, and finally determines the associated difference value between the initial window data and the normal sensing data: ; in, represents the spatial distribution difference between the initial window data and the normal sensing data, represents the correlation difference value between the initial window data and the normal sensor data, n represents the total data volume corresponding to the union result obtained by performing union calculation on the normal sensor data and the initial window data, abs represents taking the absolute value, represents the first distribution information corresponding to the i-th union result under the normal sensing data, Represents the second distribution information corresponding to the i-th union result under the initial window data.

[0062] Specifically, the distribution difference analysis module directly compares the distribution information of the two sets of data, and can keenly capture the differences in data distribution patterns, which is crucial for discovering potential abnormal patterns. Combined with the spatial analysis module to consider data differences from a comprehensive spatial perspective, and the calculation difference module to integrate multiple difference information, the associated difference value can fully and accurately reflect the difference between the initial window data and the normal sensor data, thereby greatly improving the accuracy of anomaly detection. Therefore, by analyzing the target distribution difference and spatial distribution difference separately, and then integrating them into the associated difference value, we can gain an in-depth understanding of the changes in the distribution and spatial structure of the data. This helps analysts discover potential laws, trends, and abnormal characteristics in the data, and provide more valuable information for further data mining and business decisions.

[0063] In some embodiments, the false detection module includes: a similarity calculation module, which is used to eliminate the first abnormal data in the first data to obtain third data, and then perform similarity calculation on any two data in the third data to obtain an image similarity value; a data arrangement module, which is used to perform duplicate data screening on the third data according to the image similarity value to obtain fourth data, and obtain a target reference image from the fourth data, and then arrange the similarity between the fourth data and the target reference image to obtain initial arrangement data; an arrangement screening module, which is used to perform mutual correlation information calculation on two adjacent data in the initial arrangement data to obtain a target correlation value, and obtain a correlation value sequence corresponding to the initial arrangement data according to the calculation order based on the target correlation value; an arrangement merging module, which is used to determine a preset value, and merge the initial arrangement data according to the preset value and the correlation value sequence to obtain the target arrangement data; a data classification module, which is used to perform target classification on each sub-data in the target arrangement data according to a target classification model to obtain the initial data type corresponding to the sub-data; a classification merging module, which is used to merge the initial data type to obtain the target data type corresponding to the first data, and perform false detection on the first information according to the target data type to obtain the target detection result.

[0064] Exemplarily, the similarity calculation module first removes the first abnormal data from the first data to obtain the third data, and then calculates the image similarity value between any two data in the third data according to the structural similarity index.

[0065] Exemplarily, the data arrangement module sets a similarity threshold value according to the calculated image similarity value. When the image similarity value of two data exceeds the threshold value, they are determined to be duplicate data, and one of the data is removed to obtain fourth data. A representative data is selected from the fourth data as a target reference image. The most typical, most common or most characteristic data in the data set can be selected as a reference image, and then the similarity between each data in the fourth data and the target reference image is calculated, and these data are arranged in order from high to low or from low to high in similarity to obtain initial arrangement data.

[0066] Exemplarily, the arrangement and screening module uses a cross-correlation function (such as cross-correlation coefficient calculation) to calculate the target correlation value between two adjacent data in the initial arrangement data. The cross-correlation information can reflect the correlation between the two data in time or space, so that the target correlation values ​​of each adjacent data pair are arranged in sequence according to the calculation order to form a correlation value sequence corresponding to the initial arrangement data.

[0067] Exemplarily, the arrangement and merging module sets a preset value. This preset value can be used as a threshold for determining whether adjacent data can be merged. The correlation value sequence is traversed, and when the target correlation value of two adjacent data is greater than the preset value, the two adjacent data are merged into one data. This process is repeated until all data that meets the conditions are merged, and finally the target arrangement data is obtained.

[0068] Exemplarily, the data classification module uses the labeled training data to train the target classification model according to the classification model, such as convolutional neural network (CNN), support vector machine (SVM), etc., so that it can accurately classify the data. Thus, each sub-data in the target arrangement data is input into the trained target classification model, and the model will output the initial data type corresponding to the sub-data, and the initial data type can be the corresponding attribute type of the target object under a certain attribute information.

[0069] Exemplarily, the classification and merging module uses the majority voting method to merge the initial data types of each sub-data to obtain the target data type corresponding to the first data. Thus, according to the target data type, combined with the pre-set false judgment rules (such as the data type does not match the business scenario, the data type is abnormal, etc.), the first information is subjected to false detection, and finally the target detection result is obtained.

[0070] Specifically, comprehensive consideration of multiple features such as image similarity and mutual correlation can more comprehensively describe the relationship between data, thereby improving the accuracy of false detection. Different features can reflect the authenticity of data from different angles, and multi-dimensional analysis can reduce misjudgments and missed judgments. In addition, arranging and merging data can reduce the amount of data and reduce the complexity and amount of calculation for subsequent processing.

[0071] In some embodiments, after obtaining the first data and the second data, the system further includes: a data saving module, used to store the first data and the second data in the database using the perception layer of the Internet of Things, so that the platform layer of the Internet of Things obtains the first data and the second data from the database for data analysis.

[0072] Exemplarily, after obtaining the first data and the second data, the system also selects the corresponding database according to the data storage module, and then establishes a connection between the perception layer and the database, and then stores the first data and the second data in the database, so that the platform layer of the Internet of Things obtains the first data and the second data from the database for data analysis.

[0073] In some embodiments, the system also includes: an exception reminder module, which is used to generate a target reminder text according to the target detection result when the target detection result meets the preset result, and send the target reminder text to the target terminal, so that the target user corresponding to the target terminal performs exception processing according to the target reminder text.

[0074] Exemplarily, the preset result is that when the target detection result is abnormal content, the abnormal reminder module generates a target reminder text based on the target detection result combined with the text generation rule, and then sends the target reminder text to the target terminal. After the target terminal receives the target reminder text, it reminds the target user in an obvious way, such as sound prompts, vibrations, pop-up windows, etc. The target user opens the reminder information to view a detailed description of the abnormal situation. Therefore, the target user formulates a corresponding abnormality handling plan based on the abnormal information provided in the target reminder text and his own professional knowledge and experience, and then according to the formulated abnormality handling plan, the target user performs corresponding operations on the target terminal.

[0075] See also Figure 2 , Figure 2 A product anti-counterfeiting traceability monitoring method based on the Internet of Things is provided in an embodiment of the present application. The product anti-counterfeiting traceability monitoring method based on the Internet of Things includes steps S201 to S206: Step S201: using the perception layer of the Internet of Things to determine first information according to the target identifier of the target object, and collecting first data of the target object in the production process and second data of the target object in the transportation process according to the perception layer; Step S202: using the platform layer of the Internet of Things to perform abnormality identification on the first data to obtain first abnormal data; performing abnormality identification on the second data to obtain second abnormal data, and determining the target quality type corresponding to the target object according to the first abnormal data and the second abnormal data; Step S203: using the platform layer to perform false detection on the first information according to the first data to obtain a target detection result; Step S204: using the platform layer to modify the first information according to the target quality type and the target detection result to obtain second information, and correlating the first abnormal data with the second abnormal data according to the second information to obtain relevant traceability information corresponding to the second information; Step S205: using the platform layer to associate the target identifier, the second information and the related traceability information and store them in a database; Step S206: using the application layer of the Internet of Things to obtain a target query identifier corresponding to the target user, and obtaining third information and target traceability information corresponding to the target query identifier in the database according to the target query identifier.

[0076] In some embodiments, the product anti-counterfeiting traceability monitoring method based on the Internet of Things can be applied to terminal devices.

[0077] It should be noted that technical personnel in the relevant field can clearly understand that, for the convenience and conciseness of description, the specific working process of the product anti-counterfeiting and traceability monitoring method based on the Internet of Things described above can refer to the corresponding process in the aforementioned embodiment of the product anti-counterfeiting and traceability monitoring system based on the Internet of Things, and will not be repeated here.

[0078] An embodiment of the present invention also provides a storage medium for computer-readable storage, wherein the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement any step of a product anti-counterfeiting and traceability monitoring system based on the Internet of Things as provided in the description of the embodiment of the present invention.

[0079] The storage medium may be an internal storage unit of the terminal device described in the foregoing embodiment, such as a hard disk or memory of the terminal device. The storage medium may also be an external storage device of the terminal device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc., equipped on the terminal device.

[0080] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware embodiment, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0081] It should be understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations. It should be noted that, in this article, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "including a..." does not exclude the presence of other identical elements in the process, method, article or system including the element.

[0082] The serial numbers of the embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. The above description is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A product anti-counterfeiting traceability monitoring system based on the Internet of Things, characterized in that: The system comprises: A data collection module, used to determine first information according to a target identifier of a target object using a perception layer of the Internet of Things, and to collect first data of the target object in a production process and second data of the target object in a transportation process according to the perception layer; an exception handling module, configured to use the platform layer of the Internet of Things to perform exception identification on the first data to obtain first exception data; perform exception identification on the second data to obtain second exception data, and determine the target quality type corresponding to the target object according to the first exception data and the second exception data; a false detection module, configured to use the platform layer to perform false detection on the first information according to the first data to obtain a target detection result; a data association module, configured to use the platform layer to modify the first information according to the target quality type and the target detection result to obtain second information, and to associate the first abnormal data with the second abnormal data according to the second information to obtain relevant traceability information corresponding to the second information; A data storage module, used to associate the target identifier, the second information and the related traceability information with each other using the platform layer and store them in a database; The data monitoring module is used to obtain a target query identifier corresponding to a target user by using the application layer of the Internet of Things, and obtain third information and target traceability information corresponding to the target query identifier in the database according to the target query identifier.

2. The system according to claim 1, characterized in that The exception handling module comprises: A data acquisition module, used to collect normal image data corresponding to the target object during the production process using the platform layer, and obtain abnormal description text corresponding to the target object based on expert experience; An image generation module, used to use the platform layer to perform abnormal adjustment on the normal image data according to the abnormal description text to obtain abnormal image data corresponding to the target object; A model determination module, used to determine a target abnormality recognition model corresponding to the target object according to the normal image data and the abnormal image data using the platform layer; an abnormality identification module, used to use the platform layer to perform abnormality identification on the first data according to the target abnormality identification model to obtain the first abnormal data; A data analysis module, used to obtain normal sensor data corresponding to the target object during transportation using the platform layer, and to obtain the second abnormal data by performing data association analysis based on the normal sensor data and the second data; a damage determination module, configured to obtain frequency information and time information corresponding to the second abnormal data by using the platform layer, and determine a target damage parameter corresponding to the target object during transportation according to the second abnormal data, the frequency information and the time information; A quality determination module is used to determine the target quality type corresponding to the target object according to the first abnormal data and the target damage parameter using the platform layer.

3. The system according to claim 2, characterized in that The image generation module comprises: A text feature extraction module, used to extract features from the abnormal description text using a text feature extraction network of the image generation model in the platform layer to obtain first text feature information; An image feature extraction module, used for extracting features from the normal image data using the image feature extraction network of the image generation model to obtain first image feature information; A feature fusion module, used for performing feature fusion on the first text feature information and the first image feature information by using a parameter sharing network of the image generation model to obtain second text feature information and second image feature information; an association analysis module, configured to use the image-text analysis network of the image generation model to perform an association degree analysis on the second text feature information and the second image feature information to obtain an image-text association value, and determine a corresponding image-text association matrix according to the image-text association value; An anomaly detection module, used to use the anomaly recognition network of the image generation model to determine a segmentation threshold using an optimization algorithm according to the image-text association matrix, and determine the image anomaly features corresponding to the normal image data under the anomaly description text and the text anomaly features corresponding to the anomaly description text under the normal image data according to the segmentation threshold and the image-text association matrix; An initial generation module, used for performing image generation according to the abnormal text features and the abnormal image features using the image generation network of the image generation model to obtain initial image data; The image fusion module is used to fuse the initial image data with the normal image data to obtain the abnormal image data corresponding to the target object.

4. The system according to claim 2, characterized in that The model determination module comprises: A data encoding module, configured to encode the normal image data using a first encoder of the target abnormality recognition model to obtain a first feature representation, and simultaneously encode the abnormal image data using a second encoder of the target abnormality recognition model to obtain a second feature representation, wherein the first encoder and the second encoder have the same network structure and share network parameters; an information reconstruction module, for using the information reconstruction layer of the target anomaly recognition model to perform feature dimensionality reduction on the first feature representation to obtain a first low-dimensional representation and to perform feature dimensionality reduction on the second feature representation to obtain a second low-dimensional representation, then perform a high-dimensional restoration operation on the first low-dimensional representation to obtain a first high-dimensional representation and a high-dimensional restoration operation on the second low-dimensional representation to obtain a second high-dimensional representation, and then perform a similarity calculation on the first feature representation and the first high-dimensional representation to obtain a first similarity value and a similarity calculation on the first feature representation and the second high-dimensional representation to obtain a second similarity value, train the information reconstruction layer according to the first similarity value and the second similarity value to obtain information reconstruction parameters corresponding to the information reconstruction layer, and adjust the first high-dimensional representation according to the information reconstruction parameters to obtain a third high-dimensional representation and adjust the second high-dimensional representation to obtain a fourth high-dimensional representation; A feature processing module, configured to use the feature processing layer of the target anomaly recognition model to perform feature compression on the third high-dimensional representation to obtain a third feature representation and to perform feature compression on the fourth high-dimensional representation to obtain a fourth feature representation; A data decoding module, configured to use a first decoder of the target anomaly recognition model to decode the third feature representation to obtain a fifth feature representation and to decode the fourth feature representation to obtain a sixth feature representation; a loss calculation module, configured to perform difference calculation on the fifth feature representation and the sixth feature representation using the difference calculation layer of the target abnormality recognition model to obtain a target difference value corresponding to the abnormal image data and the normal image data, and determine a loss function between the abnormal image data and the normal image data according to the target difference value; The parameter adjustment module is used to adjust the parameters of the target anomaly recognition model according to the loss function until the target anomaly recognition model corresponding to the preset conditions is obtained.

5. The system according to claim 2, characterized in that The data analysis module comprises: A window segmentation module, used for determining a target window, and segmenting the second data according to the target window to obtain a plurality of initial window data; A matrix conversion module, used to determine a first matrix corresponding to the normal sensing data under multi-dimensional data and a second matrix corresponding to the initial window data under the same multi-dimensional data; an association processing module, configured to perform matrix similarity calculation according to the first matrix and the second matrix to obtain a data association degree between the normal sensing data and the initial window data, and determine a data difference degree between the normal sensing data and the initial window data according to the data association degree; A distance calculation module, used for calculating the data distance between the normal sensing data and the initial window data, and then fusing the data distance, the data correlation degree and the data difference degree to determine the target correlation degree between the normal sensing data and the initial window data; A distribution analysis module, configured to perform data distribution analysis on the normal sensing data to obtain first distribution information and to perform data distribution analysis on the initial window data to obtain second distribution information; a difference determination module, configured to determine a correlation difference value between the initial window data and the normal sensing data by fusing the first distribution information and the second distribution information according to the target correlation degree; A data screening module is used to obtain the second abnormal data from the initial window data according to the associated difference value.

6. The system according to claim 5, characterized in that The difference determination module comprises: A distribution difference analysis module, used for calculating the difference between the first distribution information and the second distribution information to obtain a target distribution difference; A spatial analysis module, configured to determine a spatial distribution difference between the initial window data and the normal sensing data according to the target distribution difference combined with the target association degree and the data distance; a calculation difference module, configured to determine the correlation difference value between the initial window data and the normal sensing data according to the spatial distribution difference and the target distribution difference; The correlation difference value is obtained according to the following formula: ; ; in, represents the spatial distribution difference between the initial window data and the normal sensing data, represents the target distribution difference between the first distribution information and the second distribution information, represents the target relevance, Indicates the normal sensing data and the initial window data The data distance between them, exp represents the exponential function, and R1 represents the normal sensing data The corresponding first matrix is Represents the initial window data The corresponding second matrix is, represents the correlation difference value between the initial window data and the normal sensor data, n represents the total data volume corresponding to the union result obtained by performing union calculation on the normal sensor data and the initial window data, abs represents taking the absolute value, represents the first distribution information corresponding to the i-th union result under the normal sensing data, Represents the second distribution information corresponding to the i-th union result under the initial window data.

7. The system according to claim 1, characterized in that The false detection module comprises: a similarity calculation module, used for removing the first abnormal data from the first data to obtain third data, and then performing similarity calculation on any two data in the third data to obtain an image similarity value; a data arrangement module, configured to perform duplicate data screening on the third data according to the image similarity value to obtain fourth data, obtain a target reference image from the fourth data, and further arrange the similarity between the fourth data and the target reference image to obtain initial arrangement data; An arrangement and screening module, used for calculating the mutual correlation information of two consecutively adjacent data in the initial arrangement data to obtain a target correlation value, and obtaining a correlation value sequence corresponding to the initial arrangement data in a calculation order according to the target correlation value; an arrangement merging module, used to determine a preset value, and merge the initial arrangement data according to the preset value and the correlation value sequence to obtain target arrangement data; A data classification module, used for performing target classification on each sub-data in the target arrangement data according to a target classification model to obtain an initial data type corresponding to the sub-data; A classification and merging module is used to fuse the initial data type to obtain a target data type corresponding to the first data, and to perform false detection on the first information according to the target data type to obtain the target detection result.

8. The system according to any one of claims 1 to 7, characterized in that: After obtaining the first data and the second data, the system further includes: A data storage module is used to store the first data and the second data in the database by using the perception layer of the Internet of Things, so that the platform layer of the Internet of Things obtains the first data and the second data from the database for data analysis.

9. The system according to claim 8, characterized in that The system further comprises: The abnormality reminder module is used to generate a target reminder text according to the target detection result when the target detection result meets the preset result, and send the target reminder text to the target terminal, so that the target user corresponding to the target terminal performs abnormal processing according to the target reminder text.

10. A product anti-counterfeiting traceability monitoring method based on the Internet of Things, characterized in that: The method comprises: Determine first information according to a target identifier of a target object using a perception layer of the Internet of Things, and collect first data of the target object in a production process and second data of the target object in a transportation process according to the perception layer; Using the platform layer of the Internet of Things to perform abnormality identification on the first data to obtain first abnormal data; performing abnormality identification on the second data to obtain second abnormal data, and determining the target quality type corresponding to the target object according to the first abnormal data and the second abnormal data; Using the platform layer to perform false detection on the first information according to the first data to obtain a target detection result; Using the platform layer to modify the first information according to the target quality type and the target detection result to obtain second information, and correlating the first abnormal data with the second abnormal data according to the second information to obtain relevant traceability information corresponding to the second information; Using the platform layer, the target identifier, the second information and the related traceability information are associated and stored in a database; The target query identifier corresponding to the target user is obtained by utilizing the application layer of the Internet of Things, and the third information and target tracing information corresponding to the target query identifier are obtained in the database according to the target query identifier.

Images