Dynamic security check multi-chain collaborative sorting path adjustment method and system based on logistics system

Abstract

The invention discloses a dynamic security check multi-chain collaborative sorting path adjustment method and system based on a logistics system, and the method comprises the steps: S1, setting a serial multi-type security check machine array on a sorting transmission path, and obtaining multi-dimensional security check data; s2, constructing a contraband feature database through an AI image recognition algorithm, and performing dynamic comparison to output a contraband type judgment result; s3, establishing a layered path planning model, presetting multi-stage conventional sorting paths of non-contraband goods, and correspondingly setting at least three types of exclusive sorting paths according to the types of the contraband goods; s4, when a judgment result triggers a contraband type signal, adjusting a parcel transmission track in real time and labeling; s5, generating an electronic security check report with a timestamp; and S6, broadcasting the information of the prohibited goods to an alliance chain of an express company and government affair chain nodes of related departments. According to the invention, multidimensional detection of prohibited goods, dynamic path optimization and credible collaboration of cross-department data in logistics sorting are realized, and the problem of lagging of cross-department data sharing is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of logistics sorting safety detection, and particularly relates to a dynamic security check multi-chain collaborative sorting path adjustment method and system based on a logistics system. BACKGROUND

[0002] With the explosive growth of the logistics industry in China, the express business volume is expected to break through 200 billion pieces in 2025, and the short board of traditional sorting mode in efficiency and safety is increasingly prominent. The existing sorting system generally has the problem of separation of security check and sorting links, which leads to lagging of prohibited goods interception, high cost of manual intervention, and poor flow of cross-department supervision data, making it difficult to form a linkage effect.

[0003] The traditional logistics sorting technology adopts a serial mode of "security check first, sorting later", mainly relying on manual operation or fixed path sorting equipment. Specifically, the package needs to pass through an independent security check area, and then be transported to the sorting line after being confirmed safe by manual operation. If prohibited goods are detected, manual intervention is required for unpacking and screening. This mode has three major defects: Low efficiency: separation of security check and sorting leads to long process, and manual secondary operation increases time cost, especially when the volume of packages increases, which easily causes sorting congestion; Insufficient safety: the package security check mode has a high missed detection rate, and the positioning of prohibited goods relies on manual unpacking, making it difficult to quickly respond to high-risk item threats; Data island problem: security check information is only stored in the enterprise internal system, lacking a real-time sharing mechanism with regulatory departments, and unable to meet the information registration and linkage disposal requirements required by regulations. In addition, the traditional sorting path is determined at the time of packing and cannot be changed, which cannot adapt to the dynamic adjustment needs of the path by the security check result, resulting in limited application of sorting equipment with security check function.

[0004] Although the existing logistics sorting technology introduces some intelligent equipment, such as the combination of single-type security check machines and fixed path sorting lines, it still cannot break through the inherent framework of "security check-sorting" separation. For example, some enterprises try to install single X-ray security check equipment on the sorting line, but only rough detection of prohibited goods can be achieved, and multi-dimensional identification and hierarchical processing cannot be performed for different types of prohibited goods (such as metal, biological, and explosive). In terms of path planning, the existing system mostly uses static path algorithms based on destination, lacking a dynamic response mechanism for real-time congestion data and prohibited goods danger levels, leading to mixed flow transmission of high-risk items and ordinary packages, and safety hazards.

[0005] At the data coordination level, although the prior art attempts to realize data transmission between enterprises and regulatory departments through a database interface, it lacks the support of blockchain and other trusted evidence storage technologies, and the data is easy to tamper with and difficult to trace, which cannot meet the requirements of cross-department supervision on the integrity of the evidence chain. In addition, traditional electronic report generation relies on manual input and format conversion, and cannot automatically associate sorting logs and security data, resulting in lagging report generation and insufficient credibility. SUMMARY

[0006] Based on the above technical problems, the present application discloses a dynamic security multi-chain collaborative sorting path adjustment method based on a logistics system, comprising: S1. A series of multi-type security machine arrays are arranged on the sorting transmission path, and X-ray image acquisition and two-dimensional code recognition are sequentially performed on the parcels to obtain multi-dimensional security data containing prohibited item feature parameters; S2. A prohibited item feature database is constructed through an AI image recognition algorithm, and real-time multi-dimensional security data is dynamically compared with the database to output a prohibited item type determination result through a convolutional neural network model; S3. A hierarchical path planning model is established, and a multi-level conventional sorting path for non-prohibited items and at least three types of exclusive sorting paths corresponding to prohibited item types are preset; S4. When the determination result triggers a prohibited item type signal, the parcel transmission track is adjusted in real time and labeled, and it is guided to the exclusive compartment corresponding to the corresponding danger level, and the non-prohibited items are dynamically sorted to the destination according to the preset multi-level path; S5. Based on the sorting path log and the security data, an electronic security report with a time stamp is automatically generated, and a blockchain evidence storage number is associated, so that multiple departments can access and verify in real time through a cross-chain query interface; S6. The X-ray image hash value, the face sheet blockchain address, the sorting time stamp and the type label of the prohibited item are simultaneously broadcast to the express company alliance chain and the government chain nodes of the relevant departments.

[0007] Preferably, the multi-dimensional security data containing prohibited item feature parameters obtained in S1 includes density feature parameters, shape contour feature parameters and biological feature parameters, wherein the density feature parameters detect the density value of the internal substance of the parcel through the X-ray security machine, and are used to identify metal prohibited items and liquid hazardous materials; the shape contour feature parameters extract the geometric contour data of the internal goods of the parcel through the visual imaging device, and are used to compare the packaging feature of controlled drugs and the shape feature of explosive components; and the biological feature parameters obtain the DNA fragment information or protein spectrum feature of the internal substance of the parcel through the spectrum analysis device, and detect the biological source prohibited items such as alien species, genetic biological samples and endangered animal products.

[0008] Preferably, the S2 outputs the contraband type determination result by the convolutional neural network model. By inputting the X-ray image data in the real-time multi-dimensional security inspection data into the pre-trained convolutional neural network model, the spatial features in the image are extracted by using the multi-layer convolutional layer in the model, the feature dimension is reduced by the pooling layer, the extracted features are calculated with the preset contraband feature vectors of the metal, biological and explosive contrabands in the contraband feature database by cosine similarity, the density feature parameters, shape contour feature parameters and biological feature parameters are weighted and fused by combining the attention mechanism, and the contraband type determination result including the controlled knives, alien species, genetic biological evidence and explosive categories is output by the Softmax classifier.

[0009] Preferably, the S3 establishes a hierarchical path planning model. The first sorting rule tree of the non-contraband conventional sorting path is constructed by obtaining the package weight and volume parameters, and the coarse-grained shunting is performed according to the destination administrative region. Then, the second path mapping table is constructed according to the danger level of the contraband type, wherein the high-risk contraband corresponds to the exclusive compartment of the independent closed transmission track, the medium-risk contraband corresponds to the isolation compartment with a lead shielding layer, and the low-risk contraband corresponds to the conventional supervision compartment. The real-time congestion index of each path is dynamically calculated by the Dijkstra algorithm, the path priority is optimized by combining the reinforcement learning model, and the hierarchical path planning model containing the multi-level shunting node and the danger level response mechanism is formed.

[0010] Preferably, the real-time congestion index of each path is dynamically calculated by the Dijkstra algorithm, and the path priority is optimized by combining the reinforcement learning model to form the hierarchical path planning model containing the multi-level shunting node and the danger level response mechanism. Specifically, the sorting transmission track is abstracted as a directed graph , wherein the node represents the shunting node and the compartment, and the edge represents the track connection relationship. Each edge is given a real-time weight, and the formula is: , is the traffic weight coefficient, is the time delay weight coefficient, is the real-time package flow of the edge at the moment, is the maximum carrying capacity of the edge, is the average transmission time delay of the package passing through the edge ; the shortest path from the source node to the target node is calculated by the Dijkstra algorithm, and the initial path priority sequence is obtained; the reinforcement learning model is used to take the path congestion index as the state space, the path switching action as the action space, and the danger level punishment factor The reward function is formed, and the formula is: The time difference algorithm is used, and the formula is: , is the learning rate, is the discount factor, the path selection strategy is optimized, and a hierarchical path planning model containing multiple shunt nodes and dangerous level response mechanism is formed.

[0011] Preferably, the S4 adjusts the package transmission track in real time, sets an electromagnetic induction steering system at the shunt node of the sorting transmission track, determines the target package position through package two-dimensional code coordinate matching when receiving the contraband type signal, monitors the package transmission speed and impact force in real time, generates corresponding electromagnetic driving instructions according to the dangerous level parameters, drives the steering plate to adjust the deflection angle of the transmission track by changing the current direction and intensity of the electromagnetic coil, and the steering plate is driven to adjust the deflection angle of the transmission track. , the deflection angle of the steering plate is adjusted by the transmission track, wherein is the package transmission speed, is the track friction coefficient, is the acceleration of gravity, is the effective length of the steering plate, so that the high-risk contraband package is guided along the independent closed track to the special slot of the explosion-proof coating, the medium-risk contraband package is guided along the isolation track with a metal detection ring to the lead shielding slot, the low-risk contraband package is guided along the conventional shunt track to the supervision slot, and the non-contraband package is guided along the conventional electromagnetic lock triggered by the photoelectric sensor to keep the original track transmission.

[0012] Preferably, the S5 automatically generates an electronic security report with a timestamp, extracts the package flow node sequence from the sorting path log through NLP technology, including the package time, security time, shunt time, and package time, and structures the X-ray image hash value, contraband type label, and face sheet desensitization information in the security data to form a report framework containing an event time axis, generates a report content integrity proof through a zero-knowledge proof algorithm, generates a timestamp combined with a UTC time source, binds the timestamp and the block chain storage number into a chain structure that cannot be tampered with, and when multiple departments call cross-chain query interfaces, the identity is verified through a role permission smart contract, and after verification, the cross-chain gateway synchronously extracts associated data from the alliance chain and the government chain to return to the query terminal in the form of an encrypted stream, realizing a cross-department reporting verification mechanism of the data minimum authorization principle and the trusted timestamp technology.

[0013] Preferably, the process of synchronously broadcasting data in S6 is to build a communication channel for the express company alliance chain and the government chain through the blockchain cross-chain gateway, use the SHA-256 algorithm to generate a fixed-length hash value , store the face sheet information on the alliance chain to obtain a unique blockchain address , and combine the sorting timestamp With contraband type label Generate structured data packets , define multi-chain broadcast rules through smart contracts, when the contraband type is high-risk level, trigger the targeted broadcast interface of the relevant department government chain, and encrypt the hash value in and the danger level label to the relevant department government chain node; when biological safety type contraband is involved, through the threshold signature mechanism of the supervision department government chain, at least 3 alliance chain nodes jointly sign and broadcast the biological feature parameter segment in to the supervision department government chain; all types of contraband Full data is verified by the consensus mechanism PoST of the postal government chain and written into the postal chain node to form a cross-chain data synchronization system based on multi-chain heterogeneous architecture and hierarchical authorization mechanism.

[0014] The dynamic security check multi-chain collaborative sorting path adjustment system based on the logistics system includes a multi-dimensional security check module, a judgment module, a path planning module, an execution module, a cross-chain data collaboration module, and a trusted evidence module. The multi-dimensional security check module is composed of an X-ray security check machine, a visual imaging device, a spectrum analysis device, and a two-dimensional code identifier connected in series, which sequentially collects density feature parameters, shape contour feature parameters, biological feature parameters, and identity information of the package, generating a multi-dimensional data chain containing contraband features. The judgment module dynamically compares real-time security check data with contraband feature database through AI image recognition algorithm and convolutional neural network model, extracts image spatial features through multiple convolutional layers, combines attention mechanism to fuse multi-dimensional parameters, and outputs contraband type judgment results through Softmax classifier. The path planning module performs regular sorting path coarse-grained shunting and contraband matching transmission path through the construction of a multi-level path model. The execution module sets up an electromagnetic induction steering system at the shunting node. After receiving the contraband type signal, it locates the target package through two-dimensional code coordinate matching, monitors the transmission speed and impact force in real time, generates electromagnetic drive instructions according to the danger level, adjusts the deflection angle of the deflection plate, guides the contraband to the corresponding compartment, and the non-contraband is transmitted along the regular path. The cross-chain data collaboration module broadcasts the X-ray image hash value, face sheet blockchain address, sorting timestamp, and type label to the express company alliance chain and relevant department government chain nodes according to the danger level through the blockchain cross-chain gateway, forming a multi-chain heterogeneous data synchronization system. The trusted evidence module reorganizes the event timeline based on sorting path logs and security check data through NLP technology, generates an unalterable electronic security check report combined with zero-knowledge proof and UTC time source, associates the blockchain evidence number, and allows multiple departments to verify identity through cross-chain interface and retrieve verification.

[0015] Preferably, the electromagnetic induction steering system comprises a multi-axis force sensor and a visual positioning module; the visual positioning module obtains the coordinate position information of the package in real time by identifying the package two-dimensional code, establishes the spatial coordinate mapping of the package in the sorting track, and quickly locks the specific position of the target package in the transmission track when receiving the contraband type signal; the multi-axis force sensor is integrated on the surface of the transmission track, and real-time monitoring of the speed, impact force and acceleration mechanical parameters of the package during the transmission process generates a dynamic mechanical data set containing the speed vector and impact force; the electromagnetic induction steering system calculates and generates a deflection instruction for the steering plate based on the coordinate information output by the visual positioning module and the mechanical parameters collected by the multi-axis force sensor, specifically: according to the dangerous level parameter, the corresponding electromagnetic coil driving mode is called, the deflection angle of the steering plate matching the motion state of the package is generated by changing the current direction and intensity, wherein the high-risk contraband triggers the maximum deflection angle threshold corresponding to the independent closed track, the medium-risk contraband triggers the medium deflection angle threshold corresponding to the lead shielding layer isolation track, the low-risk contraband triggers the minimum deflection angle threshold corresponding to the conventional supervision track, and the non-contraband maintains the steering plate in the initial position, and the conventional electromagnetic lock triggered by the photoelectric sensor maintains the original track transmission.

[0016] Compared with the prior art, the technical scheme of the present application has the following technical effects: The present application synchronously completes multi-dimensional data acquisition of density, shape, biological characteristics and the like in the sorting process by connecting multiple types of security inspection machines (X-ray, visual imaging, spectral analysis), and realizes accurate identification and classification of contraband; The present application constructs a hierarchical path model based on the dangerous level, dynamically optimizes the path priority by combining the Dijkstra algorithm and reinforcement learning, realizes real-time diversion of contraband and non-contraband, and independently transmits high-risk goods; The present application uses blockchain cross-chain technology to synchronize security inspection data hash values, timestamps and other information to express company alliance chains and government chains, realizes hierarchical authorization and evidence solidification through a smart contract, ensures that the data cannot be tampered with and can be traced across departments.

[0017] The above description is only a summary of the technical scheme of the present application, in order to more clearly understand the technical means of the present application, so as to implement according to the content of the specification, and in order to make the above and other purposes, characteristics and advantages of the present application more obvious and easy to understand, the following will detail the preferred embodiments of the present application with the help of the drawings.

[0018] According to the detailed description of the specific embodiments of the present application in the following text combined with the drawings, those skilled in the art will be more clear about the above and other purposes, advantages and characteristics of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the description of the embodiments or the prior art. Obviously, the drawings described below are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual proportion.

[0020] Figure 1 Flow chart of the dynamic security multi-chain collaborative sorting path adjustment method based on the logistics system of the present application; Figure 2 Structural diagram of the hierarchical path planning model of the dynamic security multi-chain collaborative sorting path adjustment method of the logistics system of the present application; Figure 3 Structural diagram of the security device array in series of the sorting transmission path of the present application; Figure 4 Structural diagram of the security device array in series of the sorting transmission path of the present application; Figure 5 Structural diagram of the security device array in parallel of the sorting transmission path of the present application; Figure 6 Structural diagram of the security device array in series and parallel of the sorting transmission path of the present application; Figure 7 Structural diagram of the dynamic security multi-chain collaborative sorting path adjustment system based on the logistics system of the present application.

[0021] Reference numerals: 100, security device, 200, upper package section, 300, security section, 400, sorting section; I, sorting unit; 3, sorting platform; II, track. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all the embodiments. In the following description, specific details such as specific configurations and components are provided only to help understand the embodiments of the present application. Therefore, those skilled in the art should understand that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. In addition, in order to be clear and concise, the description of known functions and structures is omitted in the embodiments.

[0023] It should be understood that every reference made throughout this specification to "one embodiment" or "the embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the application. Thus, appearances of the phrases "in one embodiment" or "in the embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics can be combined in any suitable manner in one or more embodiments.

[0024] In addition, reference numerals and / or letters can be repeated in different instances throughout the specification. Such repetition is for the purpose of simplicity and clarity and does not in itself dictate a relationship between the various embodiments and / or configurations discussed.

[0025] The term "and / or", merely describes an associated relationship between associated objects, which means that there can be three relationships, for example, A and / or B, which means that there are three cases of A alone, B alone, and A and B together. The term "and" in this paper is to describe another relationship between associated objects, which means that there can be two relationships, for example, A and B, which means that there are two cases of A alone and A and B together. In addition, the character " / " in this paper generally represents an "or" relationship between the associated objects before and after it.

[0026] The term "at least one" in this paper merely describes the relationship between associated objects, which means that there can be three relationships, for example, at least one of A and B, which means that there are three cases of A alone, A and B together, and B alone.

[0027] It should also be noted that the relationship terms such as first and second, etc. used in this paper are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion.

[0028] Embodiment 1 This embodiment mainly describes a dynamic security check multi-chain collaborative sorting path adjustment method based on a logistics system, as shown in Figure 1 Specifically, it includes: S1, setting a series of multi-type security check machine arrays on the sorting transmission path, sequentially collecting X-ray image and two-dimensional code of the package, and obtaining multi-dimensional security check data containing prohibited item feature parameters; S2, constructing a prohibited item feature database through an AI image recognition algorithm, dynamically comparing real-time multi-dimensional security check data with the database, and outputting a prohibited item type determination result through a convolutional neural network model; S3, establish a hierarchical path planning model, preset multi-level conventional sorting path of non-prohibited goods and set at least three types of exclusive sorting path corresponding to the type of prohibited goods; S4, when the determination result triggers the prohibited goods type signal, real-time adjust the package transmission track and label, guide it to the corresponding exclusive dangerous level compartment, and non-prohibited goods are dynamically sorted to the destination according to the preset multi-level path; S5, based on the sorting path log and security data, automatically generate a time-stamped electronic security report, associate the blockchain storage number, and provide multiple departments with real-time access through cross-chain query interface; S6, the X-ray image hash value of the prohibited goods, the face sheet blockchain address, the sorting timestamp and the type label are broadcasted to the express company alliance chain and the government chain node of the relevant departments.

[0029] Further, the multi-dimensional security data containing the feature parameters of the prohibited goods obtained in S1 includes density feature parameters, shape contour feature parameters and biological feature parameters, wherein the density feature parameters are detected by the X-ray security inspection machine to identify metal prohibited goods and liquid hazardous goods; the shape contour feature parameters are extracted by the visual imaging device to compare the geometric contour data of the goods inside the package with the packaging features of controlled drugs and the shape features of explosive components; the biological feature parameters are obtained by the spectral analysis device to detect the DNA fragment information or protein spectrum features of the substances inside the package, and the biological source prohibited goods such as alien species, genetic biological samples and endangered animal products.

[0030] Further, the type of prohibited goods is determined by the convolutional neural network model in S2, the X-ray image data in the real-time multi-dimensional security data is input into the pre-trained convolutional neural network model, the spatial features in the image are extracted by using the multi-layer convolutional layer in the model, the feature dimension is reduced by the pooling layer, and the extracted features are calculated by the cosine similarity with the preset prohibited goods feature vectors of metal, biological and explosive in the prohibited goods feature database through the full connection layer. The density feature parameters, shape contour feature parameters and biological feature parameters are weighted and fused by combining the attention mechanism, and the type of prohibited goods including controlled cutters, alien species, genetic biological samples and explosive categories is output by the Softmax classifier.

[0031] Further, as Figure 2As shown, a hierarchical path planning model is established in S3, a first sorting rule tree of a non-prohibited regular sorting path is constructed by obtaining the weight and volume parameters of the package, and coarse-grained shunting is performed according to the destination administrative region; then a second path mapping table is constructed according to the danger level of the prohibited item type, wherein the high-risk prohibited item corresponds to a special compartment of an independent closed transmission track, the medium-risk prohibited item corresponds to an isolated compartment with a lead shielding layer, and the low-risk prohibited item corresponds to a regular supervision compartment; the real-time congestion index of each path is dynamically calculated by Dijkstra algorithm, the path priority is optimized by combining the reinforcement learning model, and a hierarchical path planning model containing multi-level shunting nodes and danger level response mechanism is formed.

[0032] Further, the real-time congestion index of each path is dynamically calculated by Dijkstra algorithm, and the path priority is optimized by combining the reinforcement learning model, to form a hierarchical path planning model containing multi-level shunting nodes and danger level response mechanism, specifically: the sorting transmission track is abstracted as a directed graph , wherein the node represents a shunting node and a compartment, and the edge represents a track connection relationship, each edge is given a real-time weight, and the formula is: , is a traffic weight coefficient, is a time delay weight coefficient, is the real-time package flow of edge at time t, is the maximum carrying capacity of the edge, is the average transmission time delay of the package passing through edge ; the shortest path from the source node to the target node is calculated by Dijkstra algorithm, and the initial path priority sequence is obtained; the reinforcement learning model is used to take the path congestion index as the state space, the path switching action as the action space, and the danger level penalty factor as the reward function, and the formula is: , the time difference algorithm is used, and the formula is: , is the learning rate, is the discount factor, the path selection strategy is optimized, and a hierarchical path planning model containing multi-level shunting nodes and danger level response mechanism is formed.

[0033] Furthermore, in S4, the package transmission track is adjusted in real time, and an electromagnetic induction steering system is set at the diversion node of the sorting transmission track. When a prohibited item type signal is received, the target package position is determined by matching the package QR code coordinates, and the package transmission speed and impact force are monitored in real time. The corresponding electromagnetic drive instruction is generated according to the hazard level parameters, and the steering plate is driven by changing the current direction and intensity of the electromagnetic coil to The deflection angle adjusts the transmission track, where is the package transmission speed, is the rail friction coefficient, is the acceleration due to gravity, The effective working length of the steering plate is to guide high-risk contraband packages along an independent closed track to an exclusive grid with an explosion-proof coating, medium-risk contraband packages enter the lead shielded grid via an isolation track with a metal detection ring, low-risk contraband packages go through a conventional diversion track to the supervision grid, and non-contraband packages maintain the original track transmission through a conventional electromagnetic lock triggered by a photoelectric sensor.

[0034] Furthermore, S6 automatically generates an electronic security inspection report with a timestamp. It uses NLP technology to extract the package flow node sequence from the sorting path log, including the loading time, security inspection time, diversion time, and unloading time. It is then structured and reorganized with the X-ray image hash value, contraband type label, and desensitized information of the delivery note in the security inspection data to form a report framework containing an event timeline. A proof of integrity of the report content is generated through a zero-knowledge proof algorithm, and a timestamp is generated in combination with the UTC time source. The timestamp is bound to the blockchain evidence number into an unalterable chain structure. When multiple departments call the cross-chain query interface, the identity is verified through the role permission smart contract. After the verification is passed, the cross-chain gateway will synchronously extract the related data from the alliance chain and the government chain, and return it to the query terminal in the form of an encrypted stream, realizing the cross-departmental report verification mechanism of the principle of minimum data authorization and trusted timestamp technology.

[0035] Furthermore, the process of synchronous broadcasting data in S5 is to build a communication channel between the express company alliance chain and the government chain through the blockchain cross-chain gateway, and use the SHA-256 algorithm to generate a fixed-length hash value for the X-ray image. , store the waybill information on the alliance chain to obtain a unique blockchain address , combined with the sorting timestamp Labels with prohibited item types Generate structured data groups , define multi-chain broadcast rules through smart contracts. When the type of contraband is high-risk, it triggers the directional broadcast interface of the government chain of the relevant departments. the hash value in the X-ray image and the danger level label are encrypted and transmitted to the relevant department government chain node; when biological safety type contraband is involved, the biological feature parameter segment in the X-ray image is broadcast to the regulatory department government chain; and the full amount of data of all types of contraband are written into the postal chain node through the consensus mechanism PoST of the postal government chain after verification, forming a cross-chain data synchronization system based on a multi-chain heterogeneous architecture and a hierarchical authorization mechanism.

[0036] This embodiment describes in detail that the present application realizes multi-dimensional contraband detection through the series connection of multiple types of security inspection machines, improves the judgment accuracy by combining AI and convolutional neural networks, realizes dynamic diversion of contraband by using hierarchical path planning and electromagnetic induction steering, and realizes data credible sharing through blockchain cross-chain technology, thereby improving the sorting efficiency and contraband interception rate, shortening the regulatory response time to the minute level, helping enterprises to comply with regulations, and providing an intelligent solution for the whole process of traceability for logistics safety.

[0037] Based on embodiment 1, this embodiment describes in detail the construction of a contraband feature database through an AI image recognition algorithm, specifically: Two-dimensional / three-dimensional image data of the internal substances of the package are obtained through an X-ray security inspection machine, contour images of the appearance and internal items of the package are collected by a visual imaging device, spectral feature data of biological samples are provided by a spectral analysis device, and at the same time, combined with the labeled data in the historical contraband case library (such as the metal density threshold of controlled knives and the DNA feature sequence of alien species), an original data set containing multi-dimensional parameters of density, shape, and biological features is formed; for the noise and interference in the original data, the X-ray image is denoised by Gaussian filtering, image enhancement, and other algorithms, the visual image is standardized and corrected by contour detection and feature point extraction technology, and the spectral data is converted to the frequency domain by Fourier transform, generating a uniform format feature vector data set; ​Through the deep learning framework, the feature extraction and modeling mechanism is constructed, the pre-processed image data is learned by the convolutional neural network (CNN) architecture, and the spatial features of different scales (such as edges, textures, and geometric shapes) are automatically extracted by using multiple convolutional layers. The pooling layer is used to reduce the feature dimension and enhance the translation invariance. For spectral data, the Transformer model is used for sequence feature modeling, and the feature correlation in different wavelength intervals is captured through the self-attention mechanism. After the multi-dimensional feature vector is input into the fully connected layer for fusion, the K-means clustering algorithm is used for unsupervised classification of the prohibited item features, forming metal, biological, and explosive basic category clusters; then, through the semi-supervised learning algorithm (such as LabelPropagation), combined with a small amount of manually labeled data, the category clusters are finely labeled to generate a feature database containing typical feature parameters of each type of prohibited item (such as the density range of controlled knives , the shape entropy threshold of explosives > 0.6). The database supports real-time updating, and when the system detects a new type of prohibited item, the feature parameters are automatically extracted through the online learning algorithm and integrated into the existing classification system to ensure the dynamic adaptability of the database and the continuous improvement of the detection accuracy.

[0038] This embodiment describes in detail the construction of a prohibited item feature database through an AI image recognition algorithm, which can automatically extract multi-dimensional features such as X-ray images and spectral data, realize unsupervised clustering and semi-supervised labeling of prohibited items such as metals, biological, and explosives, dynamically update the features of new prohibited items, and adapt to changes in prohibited item types, providing data support for real-time and accurate judgment.

[0039] Based on embodiment 1, this embodiment describes in detail the setting of a series of multi-type security inspection machine arrays on the sorting transmission path in S1 for image acquisition and two-dimensional code recognition. It can further change the series to parallel, and also can adjust to the mode of setting both series and parallel according to the detection accuracy, specifically: In the logistics sorting system, according to the package type, detection demand, and site conditions, the series, parallel, or series-parallel hybrid mode is flexibly selected to realize the dynamic balance of multi-dimensional detection capability and sorting efficiency; As shown in Figure 3 , the security inspection device 100 is used for security inspection scanning of the articles to detect whether there are security risks; the upper package section 200 is a starting area where the articles to be inspected are placed into the system manually or automatically; after the articles enter, they are comprehensively checked by the security inspection device 100 in the security inspection section 300; After security inspection, the articles move to the sorting section 400 along the track II, which is equipped with a sorting unit I and a sorting platform 3. According to the security inspection results, the sorting unit I can sort the articles to the corresponding outlets through the sorting platform 3 according to different categories or security states, realizing automatic classification and processing of the articles. like Figure 4 As shown, the tandem mode is suitable for single-path, full-volume screening with a priority on in-depth inspection. This mode sequentially deploys X-ray security inspection equipment, visual imaging equipment, spectral analysis equipment, and QR code readers along the sorting and transport path, forming a linear inspection chain. Packages pass through each device in a fixed order. First, the QR code reader reads the package label information and establishes a package ID index. The X-ray security inspection equipment then acquires density characteristics (such as high grayscale areas of metal parts or the layered density distribution of liquid hazardous materials). The visual imaging equipment then extracts shape and contour features (such as the rectangular packaging outline of controlled drugs or the irregular geometric shape of explosive components). Finally, the spectral analysis equipment collects biometric parameters (such as the DNA spectrum of exotic species or the protein absorption peaks of endangered animal products). The key advantage of this mode lies in the strong sequential nature of the inspection process and the correlation of data. The inspection results of preceding devices can serve as triggers for subsequent devices (for example, if a metallic foreign object is detected by X-ray, the visual imaging resolution is automatically increased for detailed verification). This mode is suitable for scenarios with a high proportion of high-risk packages and the need for in-depth, full-volume inspection (such as international logistics customs clearance and sorting). The key technical point lies in the timing synchronization mechanism between devices. Through closed-loop control of photoelectric sensors and conveyor belt speed, it ensures that the residence time of the package in the field of view of each device meets the detection accuracy requirements, avoiding feature omissions caused by fluctuations in transmission speed.

[0040] like Figure 5 As shown, the parallel mode is suitable for multi-path parallel inspections that prioritize efficiency. This mode deploys different types of security inspection machines at the same inspection station along the sorting path, triggering simultaneous inspections on multiple devices as packages pass through. For example, an X-ray security inspection machine (for density detection), a line array camera (for shape extraction), and a Raman spectrometer (for biological analysis) are placed side by side at the same location, with a QR code scanning module integrated below. This multi-sensor array enables simultaneous acquisition of multi-dimensional data. This mode breaks through the timing bottleneck of the serial mode through spatial multiplexing and parallel computing. The time consumption of single package detection is equivalent to that of single device detection, which significantly improves sorting efficiency. It is suitable for large-scale distribution centers with an average daily package volume of over one million pieces. The technical implementation needs to solve the problem of spatiotemporal alignment of multi-source data. A high-precision clock synchronization module is used to ensure that the sampling time error of each device is less than 1ms. The package coordinate mapping algorithm (such as three-dimensional positioning based on lidar) is used to associate the feature parameters collected by different devices with the same spatial coordinate system to avoid feature dislocation caused by detection position deviation. In addition, the parallel mode supports dynamic load balancing. When a certain type of equipment (such as a spectrometer) fails, the system automatically switches to other equipment combinations to complete the detection, thereby improving the robustness of the system.

[0041] like Figure 6As shown, the series-parallel hybrid mode is suitable for a hierarchical detection architecture that optimizes precision and efficiency, and the series-parallel hybrid mode combines the advantages of the first two modes to build a "rough screening-accurate detection" hierarchical detection link; the first stage uses a parallel array to achieve rapid preliminary screening, including a low-cost X-ray machine and a two-dimensional code identifier, to perform density threshold detection and face sheet risk rating (such as marking high-risk packages according to the sender's historical records) on the package; for packages found to be abnormal in preliminary screening (such as density exceeding the metal threshold, sender located on the risk list), automatically import the second-stage serial detection channel, and enable high-precision spectrometer and 3D visual imaging equipment for in-depth feature analysis, while normal packages quickly pass along the conventional parallel path; this mode uses a hierarchical filtering mechanism to divert more than 90% of low-risk packages to an efficient channel, and only 5%-10% of suspicious packages are subjected to in-depth detection, which improves the overall sorting efficiency by more than 40% while ensuring the interception rate of prohibited items. The core technology lies in the dynamic adjustment of the hierarchical strategy: through a reinforcement learning model to analyze historical detection data, automatically optimize the preliminary screening threshold and accurate detection trigger conditions, such as increasing the preliminary screening sensitivity of explosive-related features during certain periods (such as holidays), and dynamically allocating detection resources.

[0042] Intelligent selection of security inspection machine topology is achieved through a decision-making model, including: Package type distribution: when biological prohibited items account for > 30%, prefer the series mode to ensure the integrity of spectral detection; Site space constraints: in scenarios where the sorting line length is limited (such as urban distribution centers), use the parallel mode to compress the detection station footprint; Policy compliance requirements: for cross-border packages that require full biological safety detection, the series-parallel hybrid mode is mandatory to ensure full coverage of DNA feature collection; Real-time computing load: when the edge computing node's computing power is insufficient, use the lightweight detection algorithm (such as a simplified CNN model) of the parallel mode to reduce computing pressure.

[0043] Through this decision-making model, topology mode switching can be completed within 50ms, while maintaining the dynamic optimal solution of detection accuracy and sorting efficiency, providing customized security inspection and sorting solutions for different logistics scenarios.

[0044] This embodiment describes that the security inspection machine array uses series, parallel, and series-parallel hybrid modes, which can flexibly adapt to different scene requirements. The series mode achieves full-depth detection, the parallel mode improves sorting efficiency, and the series-parallel hybrid mode balances precision and efficiency. Coupled with an intelligent decision-making model to dynamically switch the topology structure, it helps logistics enterprises efficiently and accurately complete prohibited item screening.

[0045] Embodiment 2 This embodiment describes a dynamic security inspection multi-chain collaborative sorting path adjustment system based on the logistics system, such as Figure 7As shown, including multi-dimensional security module, decision module, path planning module, execution module, cross-chain data collaboration module and trusted evidence module; The multi-dimensional security module is composed of an X-ray security inspection machine, a visual imaging device, a spectral analysis device and a two-dimensional code identifier connected in series, and sequentially collects density characteristic parameters, shape contour characteristic parameters, biological characteristic parameters and identity information of the package, to generate a multi-dimensional data chain containing prohibited item features; The decision module dynamically compares real-time security inspection data with a prohibited item feature database through an AI image recognition algorithm and a convolutional neural network model, extracts image spatial features through multiple convolutional layers, combines attention mechanism to fuse multi-dimensional parameters, and outputs prohibited item type determination results through a Softmax classifier; The path planning module performs coarse-grained shunting of conventional sorting paths and matching transmission paths of prohibited items by constructing a multi-level path model; The execution module sets an electromagnetic induction steering system at the shunting node, after receiving the prohibited item type signal, locates the target package through two-dimensional code coordinate matching, monitors the transmission speed and impact force in real time, generates electromagnetic drive instructions according to the danger level, adjusts the deflection angle of the steering plate, guides the prohibited items to the corresponding compartment, and the non-prohibited items are transmitted along the conventional path; The cross-chain data collaboration module broadcasts the X-ray image hash value, the face sheet blockchain address, the sorting timestamp and the type label to the express company alliance chain and the relevant department government chain node according to the danger level through the blockchain cross-chain gateway, to form a multi-chain heterogeneous data synchronization system; The trusted evidence module generates an unalterable electronic security inspection report based on the sorting path log and the security inspection data through NLP technology, combines zero-knowledge proof and UTC time source, associates the blockchain evidence number, and provides multiple departments to verify the identity through the cross-link interface to call the verification.

[0046] The electromagnetic induction steering system comprises a multi-axis force sensor and a visual positioning module; the visual positioning module obtains the coordinate position information of the package in real time by identifying the package two-dimensional code, establishes the spatial coordinate mapping of the package in the sorting track, and quickly locks the specific position of the target package in the transmission track when receiving the contraband type signal; the multi-axis force sensor is integrated on the surface of the transmission track, and real-time monitoring of the speed, impact force and acceleration mechanical parameters in the package transmission process generates a dynamic mechanical data set containing the speed vector and impact force; the electromagnetic induction steering system calculates and generates a deflection instruction for the steering plate based on the coordinate information output by the visual positioning module and the mechanical parameters collected by the multi-axis force sensor, specifically: according to the dangerous level parameter, the corresponding electromagnetic coil driving mode is called, the deflection angle of the steering plate matching the motion state of the package is generated by changing the current direction and intensity, wherein the high-risk contraband triggers the maximum deflection angle threshold of the corresponding independent closed track, the medium-risk contraband triggers the medium deflection angle threshold of the corresponding lead shielding layer isolation track, the low-risk contraband triggers the minimum deflection angle threshold of the corresponding regular supervision track, and the non-contraband maintains the steering plate in the initial position, and the regular electromagnetic lock triggered by the photoelectric sensor keeps the original track transmission.

[0047] The embodiment describes in detail that the multi-dimensional security inspection module realizes accurate identification of contraband, hierarchical path planning combined with dynamic adjustment improves sorting efficiency, cross-chain data collaboration ensures that information is tamper-proof and shared in real time, the trusted evidence module supports minute-level cross-department verification, improves the overall contraband interception rate, and builds a safe and efficient intelligent logistics sorting system.

[0048] The above is only a preferred embodiment of the present application, and does not limit the protection scope of the present application. For those skilled in the art, the present application can have various changes and variations; any change, modification, replacement, integration and parameter change of these embodiments within the spirit and principles of the present application, which can realize the same function without departing from the principles and spirit of the present application, falls within the protection scope of the present application.

Claims

1. A dynamic security inspection multi-chain collaborative sorting path adjustment method based on a logistics system, characterized in that: include: S1. One or more security inspection machine arrays are set up in series on the sorting and transmission path to perform X-ray image acquisition and QR code recognition on the packages in sequence to obtain multi-dimensional security inspection data including characteristic parameters of prohibited items; S2. Build a database of prohibited items using AI image recognition algorithms, dynamically compare real-time multi-dimensional security inspection data with the database, and output the prohibited item type determination results using a convolutional neural network model. S3. Establish a hierarchical path planning model, pre-set multiple levels of conventional sorting paths for non-prohibited items, and set at least three types of dedicated sorting paths corresponding to the types of prohibited items; S4. When the judgment result triggers the prohibited item type signal, the package transmission track is adjusted in real time and the label is affixed, directing it to the dedicated slot corresponding to the danger level. Non-prohibited items are dynamically sorted to the destination according to the preset multi-level path; S5. Based on the sorting path log and security inspection data, automatically generate a time-stamped electronic security inspection report, associate it with the blockchain evidence number, and make it available for real-time retrieval and verification by multiple departments through the cross-chain query interface; S6. The X-ray image hash value, bill blockchain address, sorting timestamp and type label of the prohibited items are synchronously broadcast to the express company alliance chain and the government chain nodes of relevant departments.

2. The method for adjusting the dynamic security inspection multi-chain collaborative sorting path based on the logistics system according to claim 1 is characterized in that: The multi-dimensional security inspection data containing characteristic parameters of prohibited items obtained in S1 include density characteristic parameters, shape contour characteristic parameters and biometric characteristic parameters, among which the density characteristic parameters are detected by an X-ray security inspection machine to detect the density value of the material inside the package, which is used to identify metal contraband and liquid dangerous goods; the shape contour characteristic parameters are extracted by a visual imaging device to extract the geometric contour data of the items inside the package, which is used to compare the packaging characteristics of controlled drugs and the morphological characteristics of explosive components; the biometric characteristic parameters are obtained by a spectral analysis device to obtain the DNA fragment information or protein spectral characteristics of the material in the package, and detect biological contraband such as alien species, genetic biological samples and endangered animal products.

3. The method for adjusting the dynamic security inspection multi-chain collaborative sorting path based on the logistics system according to claim 1, characterized in that: In the S2, the contraband type determination result is output through a convolutional neural network model. The X-ray image data in the real-time multi-dimensional security inspection data is input into the pre-trained convolutional neural network model, and the spatial features in the image are extracted using the multi-layer convolution layer in the model. The feature dimension is reduced through the pooling layer. The extracted features are calculated by cosine similarity with the prohibited item feature vectors of metal, biological, and explosive categories preset in the prohibited item feature database through the fully connected layer. The density feature parameters, shape contour feature parameters, and biological feature parameters are weightedly fused in combination with the attention mechanism. The prohibited item type determination result including controlled knives, alien species, genetic biological samples, and explosives categories is output through the Softmax classifier.

4. The method for adjusting the dynamic security inspection multi-chain collaborative sorting path based on the logistics system according to claim 1, characterized in that: A hierarchical path planning model is established in S3. By obtaining the weight and volume parameters of the package, a first-level sorting rule tree for the conventional sorting path of non-prohibited items is constructed, and coarse-grained diversion is performed according to the destination administrative area. A second-level path mapping table is then constructed according to the hazard level of the prohibited item type, where high-risk contraband corresponds to a dedicated grid on an independent closed transmission track, medium-risk contraband corresponds to an isolated grid with a lead shielding layer, and low-risk contraband corresponds to a conventional supervision grid. The real-time congestion index of each path is dynamically calculated using the Dijkstra algorithm, and the path priority is optimized in combination with the reinforcement learning model, forming a hierarchical path planning model that includes multi-level diversion nodes and a hazard level response mechanism.

5. The method for adjusting the dynamic security inspection multi-chain collaborative sorting path based on the logistics system according to claim 4 is characterized in that: The Dijkstra algorithm is used to dynamically calculate the real-time congestion index of each path and optimize the path priority in combination with the reinforcement learning model to form a hierarchical path planning model including multi-level diversion nodes and a hazard level response mechanism. Specifically, the sorting transmission track is abstracted into a directed graph. , where the nodes Indicates the diversion node and the grid, the edge Represents the track connection relationship, and each edge is assigned a real-time weight. The formula is: , is the flow weight coefficient, is the delay weight coefficient, for Time Edge Real-time package traffic, is the maximum carrying capacity of the edge, For packages to pass through the edge The average transmission delay of the source node is calculated by Dijkstra algorithm. To the target node The shortest path , get the initial path priority sequence; use the reinforcement learning model to calculate the path congestion index is the state space, with path switching action is the action space, with the danger level penalty factor The reward function is formed as follows: , through the time series difference algorithm, the formula is: , is the learning rate, The path selection strategy is optimized based on the discount factor, and a hierarchical path planning model is formed, which includes multi-level diversion nodes and hazard level response mechanism.

6. The method for adjusting the dynamic security inspection multi-chain collaborative sorting path based on the logistics system according to claim 1, characterized in that: In the said S4, the package transmission track is adjusted in real time, and an electromagnetic induction steering system is set at the diversion node of the sorting transmission track. When a prohibited item type signal is received, the target package position is determined by matching the package QR code coordinates, and the package transmission speed and impact force are monitored in real time. The corresponding electromagnetic drive instruction is generated according to the hazard level parameter, and the steering plate is driven by changing the current direction and intensity of the electromagnetic coil to The deflection angle adjusts the transmission track, where is the package transmission speed, is the rail friction coefficient, is the acceleration due to gravity, The effective working length of the steering plate is to guide high-risk contraband packages along an independent closed track to an exclusive grid with an explosion-proof coating, medium-risk contraband packages enter the lead shielded grid via an isolation track with a metal detection ring, low-risk contraband packages go through a conventional diversion track to the supervision grid, and non-contraband packages maintain the original track transmission through a conventional electromagnetic lock triggered by a photoelectric sensor.

7. The method for adjusting the dynamic security inspection multi-chain collaborative sorting path based on the logistics system according to claim 1, characterized in that: The S5 automatically generates an electronic security inspection report with a timestamp, and uses NLP technology to extract the package flow node sequence from the sorting path log, including the package loading time, security inspection time, diversion time, and package unloading time. It is structurally reorganized with the X-ray image hash value, contraband type label, and desensitized information of the waybill in the security inspection data to form a report framework containing an event timeline. The report content integrity proof is generated through a zero-knowledge proof algorithm, and a timestamp is generated in combination with the UTC time source. The timestamp and the blockchain evidence number are bound to form an unalterable chain structure. When multiple departments call the cross-chain query interface, the identity is verified through the role permission smart contract. After the verification is passed, the cross-chain gateway synchronously extracts the related data from the alliance chain and the government chain, and returns it to the query terminal in the form of an encrypted stream, realizing the cross-departmental report verification mechanism of the principle of minimum data authorization and trusted timestamp technology.

8. The method for adjusting the dynamic security inspection multi-chain collaborative sorting path based on the logistics system according to claim 1, characterized in that: The process of synchronous broadcasting data in S6 is to build a communication channel between the express company alliance chain and the government chain through the blockchain cross-chain gateway, and use the SHA-256 algorithm to generate a fixed-length hash value for the X-ray image. , store the waybill information on the alliance chain to obtain a unique blockchain address , combined with the sorting timestamp Labels with prohibited item types Generate structured data groups , define multi-chain broadcast rules through smart contracts. When the type of contraband is high-risk, it triggers the directional broadcast interface of the government chain of the relevant departments. The hash value and hazard level mark in the encrypted data are transmitted to the government chain node of the relevant department; when it comes to biosafety contraband, the threshold signature mechanism of the government chain of the regulatory department is used and signed by at least three alliance chain nodes. The biometric parameter fragments in the data are broadcast to the government chain of the regulatory department; the full data of all types of contraband All of them are verified by the consensus mechanism PoST of the Postal Government Affairs Chain and then written into the Postal Chain node, forming a cross-chain data synchronization system based on multi-chain heterogeneous architecture and hierarchical authorization mechanism.

9. A dynamic security inspection multi-chain collaborative sorting path adjustment system based on the logistics system is characterized by: It includes multi-dimensional security inspection module, judgment module, path planning module, execution module, cross-chain data collaboration module and trusted evidence module; The multi-dimensional security inspection module consists of a serially connected X-ray security inspection machine, a visual imaging device, a spectrum analysis device, and a QR code reader. It collects density characteristic parameters, shape and contour characteristic parameters, biometric characteristic parameters, and identity information of the package in sequence, generating a multi-dimensional data chain containing the characteristics of prohibited items. The determination module uses AI image recognition algorithms and convolutional neural network models to dynamically compare real-time security inspection data with a database of prohibited items. It extracts image spatial features through multi-layer convolutional layers, fuses multi-dimensional parameters with an attention mechanism, and outputs the prohibited item type determination result through a Softmax classifier. The path planning module constructs a multi-level path model to perform coarse-grained diversion of conventional sorting paths and matching transmission paths for contraband; The execution module sets up an electromagnetic induction steering system at the diversion node. After receiving the prohibited item type signal, it locates the target package through QR code coordinate matching, monitors the transmission speed and impact force in real time, generates electromagnetic drive instructions based on the danger level, adjusts the deflection angle of the steering plate, and guides the prohibited items to the corresponding slot. Non-prohibited items are transported along the regular path. The cross-chain data collaboration module broadcasts the X-ray image hash value, bill blockchain address, sorting timestamp and type label to the express company alliance chain and relevant department government chain nodes according to the risk level through the blockchain cross-chain gateway, forming a multi-chain heterogeneous data synchronization system; The trusted evidence module is based on sorting path logs and security inspection data, reorganizes the event timeline through NLP technology, combines zero-knowledge proof with UTC time source to generate an unalterable electronic security inspection report, and associates the blockchain evidence number for multiple departments to verify identity through cross-chain interfaces and then retrieve and verify.

10. The dynamic security inspection multi-chain collaborative sorting path adjustment system based on the logistics system according to claim 9 is characterized in that: The electromagnetic induction steering system includes a multi-axis force sensor and a visual positioning module. The visual positioning module obtains the package's coordinate position information in real time by identifying the package's QR code, establishing a spatial coordinate mapping of the package within the sorting track. Upon receiving a prohibited item type signal, it quickly locates the target package's specific location within the transport track. The multi-axis force sensor, integrated into the transport track surface, monitors the package's velocity, impact force, and acceleration during transport in real time, generating a dynamic mechanical data set containing velocity vectors and impact force. Based on the coordinate information output by the visual positioning module and the mechanical parameters collected by the multi-axis force sensor, the electromagnetic induction steering system calculates and generates steering plate deflection instructions. Specifically, the system invokes the corresponding electromagnetic coil drive mode based on the hazard level parameter, altering the current direction and intensity to cause the steering plate to deflect at an angle that matches the package's motion state. High-risk prohibited items trigger the maximum deflection angle threshold corresponding to an independent closed track, medium-risk prohibited items trigger the medium deflection angle threshold corresponding to an isolated track with a lead shield, and low-risk prohibited items trigger the minimum deflection angle threshold corresponding to a conventionally supervised track. For non-prohibited items, the steering plate maintains its initial position, and conventional electromagnetic locks triggered by photoelectric sensors maintain transport along the original track.

Images