A method for dynamic generation and execution of a mobile terminal device refurbishing process route based on multi-dimensional defect features

Abstract

The application discloses a kind of based on multi-dimensional defect feature mobile terminal equipment renovation process route dynamic generation and execution method, it is related to mobile terminal equipment intelligent renovation production technical field.The four-layer system architecture of cloud-edge-end cooperation is constructed in the application, the data of equipment to be renovated are collected by physical perception layer, and standardized multi-dimensional defect feature vector is generated after edge computing preprocessing;Core business logic layer generates initial optimal process route based on process atom library and topological sorting algorithm, and realizes the dynamic update of process route in production through hot reconfiguration mechanism;At the same time, the seed trigger linkage calculation model is matched to realize the collaborative binding of multi-source identity identification, the controllable injection of security key is realized based on multi-dimensional environment fingerprint and dynamic token, and the whole-process closed-loop verification is realized through three-party data comparison and production state machine.The application realizes the flexible control of renovation process, and improves production efficiency, product compliance and quality traceability ability.

Description

Technical Field

[0001] This invention relates to the field of intelligent refurbishment production technology for mobile devices, specifically to a method for dynamically generating and executing a refurbishment process route for mobile devices based on multi-dimensional defect characteristics. Background Technology

[0002] As the global smartphone market enters a period of saturation, the circulation of used mobile devices and the official refurbishment business are experiencing explosive growth. Unlike the standardized production of new devices, the refurbishment business faces the challenge of non-standardized inputs. The failure modes of recycled devices vary widely and are often concealed. At the same time, it must meet stringent compliance requirements such as telecommunications industry standards and network access licenses. Traditional production management systems derived from new device manufacturing, with their fixed processes, standard inputs, and batch processing logic, are ill-suited to the dynamic processes, non-standard inputs, and single-piece flow demands of the refurbishment business, thus hindering industry development.

[0003] The current mainstream refurbishment production solutions in the industry have the following problems: rigid process flow, generally adopting preset fixed linear workflow, unable to adapt to actual equipment defects, requiring fault-free equipment to perform redundant processes, resulting in wasted time and resources, and new hidden defects discovered during production cannot be automatically inserted into the corresponding process; isolated management of multi-source identity identifiers, with key identifiers such as IMEI, SN, and MAC generated and managed independently, lacking collaborative computing and logical binding mechanisms.

[0004] Furthermore, existing technologies have shortcomings in terms of security control and quality traceability. Highly sensitive security key injection often adopts a static distribution method, lacking multi-dimensional verification and dynamic protection of the injection environment, which poses security risks such as key leakage and illegal reuse. The entire production process lacks a real-time closed-loop verification mechanism, and key quality control points rely heavily on manual verification or centralized inspection at the final stage. This results in delayed anomaly detection, making it impossible to immediately block defective products. Moreover, the data throughout the process is fragmented, making it difficult to trace quality issues. Summary of the Invention

[0005] To address the above problems, this invention provides a method for dynamically generating and executing a mobile device refurbishment process route based on multi-dimensional defect features, comprising the following steps: A cloud-edge-device collaborative system architecture is constructed, which is divided into a physical perception layer, an edge computing layer, a core business logic layer, and an application interaction layer from top to bottom, to collaboratively complete the entire process management of device refurbishment. The fault data and unique hardware fingerprint information of the equipment to be refurbished are collected by the physical perception layer, preprocessed by the edge computing layer, and a multi-dimensional defect feature vector based on standardized defect classification is generated and uploaded. The dynamic routing engine of the core business logic layer receives feature vectors and generates an initial optimal process route based on a predefined process atom library and topology sorting algorithm, following the disassembly and assembly order, soft and hard dependencies, and safety pre-constraints. When a new defect is detected in production, a hot reconfiguration mechanism is triggered to dynamically update the process route through local topology rearrangement of the directed acyclic graph. Based on the seed-triggered linkage calculation model, the collaborative generation, logical strong binding and pre-occupancy locking of multi-source identity identifiers of devices are completed; Based on multi-dimensional environmental fingerprint verification and a one-time dynamic token mechanism, the entire process of controllable injection of device security keys is completed; By comparing data at the bit level from the device's underlying layer, physical tags, and the central database, and combining this with production state machine control, a closed-loop verification and anomaly interception can be achieved throughout the entire process.

[0006] Preferably, the process atomic library breaks down the entire mobile device refurbishment process into the smallest indivisible process atomic nodes. Each process atomic node defines clear input preconditions and output post-effects. The input preconditions determine the verification results of the preceding process based on the device's global data snapshot, and the output post-effects determine the execution results of the process based on the job test data log. If they do not match, the process is forced to roll back and an alarm is triggered.

[0007] Preferably, the standardized defect classification includes 4 primary dimensions and 12 secondary fault domains. The primary dimensions are divided into structural and appearance dimensions, electronic and hardware dimensions, display and interaction dimensions, and software and security dimensions. Each primary dimension has 3 secondary fault domains, thereby realizing the standardized hierarchical classification of defect features.

[0008] Preferably, when generating the initial optimal process route, the process atomic library is first traversed to match all process atomic nodes that meet the preconditions, and then a path comprehensive scoring model is constructed through multi-dimensional evaluation indicators to select the path with the smallest comprehensive score from the legal topology paths as the initial optimal process route; the evaluation indicators include total process time, resource switching cost, rework risk coefficient, and bottleneck station load.

[0009] Preferably, the trigger condition for the hot reconfiguration mechanism is that the defect corresponding to the workstation detection result cannot be covered by the subsequent nodes of the current process route; when hot reconfiguration is executed, the completed node set and the indestructible process dependency relationship are locked first, and then the pruning, grafting, fusion and hot loading steps are executed in sequence to complete the seamless update of the process route.

[0010] Preferably, the seed-triggered linkage calculation model uses the device's physical serial number as a unique seed, and calculates the complete set of associated identity data of the device based on preset rule templates for different models and sales locations, forming a strong association cluster at the database level; before the identity data is written, it is pre-occupied and locked in the database, and after successful writing, it is permanently logically bound.

[0011] Preferably, the one-time dynamic token is generated using a time-slice-based hash algorithm and is strongly coupled with a multi-dimensional environmental fingerprint consisting of tool fingerprint, personnel fingerprint, device fingerprint, and network fingerprint. The token is for short-term, one-time use, transmitted through an asymmetric encryption channel, and must pass local bidirectional verification on the terminal before the key stream can be decrypted and injected.

[0012] Preferably, the three-party data bit-level comparison first obtains the database baseline data based on the device physical serial number, and then compares the consistency hash values ​​of the device underlying data with the database baseline data and the physical tag data with the database baseline data respectively. If any one of the comparisons is inconsistent, the device flow is immediately blocked and an alarm is triggered.

[0013] Preferably, the production state machine includes core states such as standby, processing, verification, qualified, locked, and rework. The access transition of the state machine requires environmental fingerprint verification to pass. If the verification is inconsistent, it will immediately jump to the locked state and trigger hardware blocking. In the locked state, an authorized administrator needs to intervene to repair and review before the state can be unlocked.

[0014] Preferably, the physical sensing layer is responsible for the acquisition of raw data at the bottom layer, the edge computing layer is responsible for real-time data preprocessing and low-latency process routing jump response, the core business logic layer has four built-in functional engines: dynamic routing engine, multi-source identifier collaborative generator, security key injection gateway, and real-time closed-loop verification center, and the application interaction layer provides visual management and full lifecycle traceability functions.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: Through dynamic routing algorithms, the refurbishment process was adaptively adjusted, reducing rework rates and wasted time, and improving the production line's ability to handle non-standard refurbishment equipment. By using a multi-source identifier collaborative generation and environmental fingerprint key injection mechanism, the problems of inconsistent device identity data and leakage of security credentials are solved, ensuring that refurbished products meet the requirements; Through real-time closed-loop verification across the entire chain, quality problems can be detected and blocked at the millisecond level, reducing the risk of batch quality accidents and improving the consistency of outgoing products. The digital recording of production data throughout the entire process enables precise traceability of the entire lifecycle, from equipment recycling to finished product shipment, providing a data foundation for quality analysis and liability determination. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall process of the method of the present invention; Figure 2 This is a system architecture diagram based on cloud-edge-device collaboration. Detailed Implementation

[0017] Example 1: Reference Figures 1 to 2 This invention provides a method for dynamically generating and executing mobile device refurbishment process routes based on multi-dimensional defect features. By constructing a closed-loop intelligent system that can sense the real-time status of the device, dynamically adjust the production path, collaboratively generate compliant identity data, and ensure the injection of secure keys, it achieves intelligent, flexible, compliant, and traceable management and control of the entire mobile device refurbishment process.

[0018] The system adopts a cloud-edge-device collaborative architecture, which is divided into four layers from top to bottom: physical perception layer, edge computing layer, core business logic layer and application interaction layer. Each layer works together to complete the control and execution of the entire equipment refurbishment process.

[0019] The physical sensing layer (edge-side) is the system's bottom-level data acquisition layer, including hardware devices such as the mobile devices to be refurbished, automated testing fixtures, smart programming boxes, RFID / barcode scanners, and visual inspection cameras. This layer is primarily responsible for collecting the underlying operational logs and unique hardware fingerprint information of the devices to be refurbished, as well as production environment data from each workstation, providing raw data support for upper-level decision-making.

[0020] The edge computing layer (edge ​​side) is an edge gateway deployed locally on the production line. It performs real-time preprocessing on the massive detection data collected by the physical sensing layer, extracts multi-dimensional defect feature vectors of the equipment to be refurbished, and responds to simple process routing jump commands within milliseconds, ensuring the low latency characteristics of production line control and avoiding the impact of network latency in cloud decision-making on production line cycle time.

[0021] The core business logic layer (cloud / central server) is deployed in the cloud or on a local central server and includes four major functional engines: Dynamic routing engine: A process route generator and reconfigurator based on a directed acyclic graph (DAG) to realize the initial generation and real-time dynamic reconfiguration of process routes; Multi-source identifier co-generator: A rule-based IMEI (International Mobile Equipment Identity) / SN (Device Serial Number) / MAC (Media Access Control Address) / PSN (Physical Serial Number) linkage calculation module, used to realize the collaborative generation and strong binding of multi-source identity identifiers of devices; Security Key Injection Gateway: A dynamic token distribution module based on environmental fingerprint verification, used to implement secure injection control of highly sensitive security keys; Real-time closed-loop verification center: a full-link data comparison and state machine control module used to realize full-process data consistency verification and closed-loop management of production status.

[0022] The application interaction layer serves as the interface between the system and users, providing a visual production dashboard, a process configuration center, abnormal alarm push notifications, and equipment lifecycle traceability query functions, thereby achieving visual control and full-process traceability of the production process.

[0023] The system is based on a cloud-edge-device collaborative architecture. The interaction between modules mainly occurs in three stages: dynamic routing decision-making, collaborative generation of identity data, and security key injection. Among them, the dynamic process routing interaction process is the foundation for realizing one-machine-one-policy refurbishment management. This process realizes full-process management from equipment warehousing to production line receiving customized process routes. The specific steps are as follows: End-side data acquisition: After the equipment to be refurbished is put into the warehouse, the physical serial number of the equipment is read by a barcode scanner, and the fault data of the equipment is collected through the initial inspection station, including visible defect information such as screen cracks and battery bulging. Edge data processing and reporting: The edge gateway packages and preprocesses the collected raw fault data to generate a multi-dimensional defect feature vector for the device. And upload the vector to the core business logic layer in the cloud; Cloud-based process route decision-making: The dynamic routing engine receives multi-dimensional defect feature vectors. The system matches the built-in process atom library, filters available process atom nodes based on preconditions, and runs a topology sorting algorithm to generate an initial process route. Execution instruction issuance: The generated initial process route is pushed to the terminal (HMI) of the corresponding workstation to guide the operators to carry out the refurbishment work according to the customized route.

[0024] The foregoing describes the end-to-end execution chain of the entire device refurbishment process based on the cloud-edge-device collaborative architecture of this invention. To further clarify the core technical means by which this method overcomes the shortcomings of existing technologies, the specific implementation steps of each core mechanism are explained in detail below: S1. Dynamic generation and reconfiguration mechanism of process route based on multi-dimensional defect features The mechanism adopts a flexible control model of process atomic library plus dynamic assembly, realizing customized generation and real-time adjustment of process routes for each piece of equipment to be refurbished. The specific implementation method is as follows: S11, Definition of Process Atomization This invention breaks down the entire refurbishment process of mobile devices into the smallest indivisible process atomic nodes, such as screen removal, battery replacement, IMEI writing, RF calibration, and photosensor testing. Each process atomic node defines clearly defined input preconditions and output post-effects: Input preconditions: Used to determine whether all log files of the preceding process have passed verification. The system maintains a global context object, which contains all current data snapshots of the device, as the benchmark for determining the preconditions. Output post-effect: After the current workstation is completed, test data logs are generated based on various work indicators. If the actual output effect does not match the standard definition (such as the equipment status not changing or the core data not being updated), even if the physical action has been completed, the system will determine that the process logic has failed, force the process to roll back and trigger an alarm; if the actual output effect matches the standard definition, the process verification passes and the equipment flows to the next workstation.

[0025] S12. Definition of Multidimensional Defect Feature Vector After the initial inspection of the equipment upon receipt, the system generates a multi-dimensional defect feature vector of the equipment based on the initial inspection data. ,in These represent the 1st, 2nd, ..., nth defect items of the device. This invention divides defect features into 4 primary dimensions and 12 secondary fault domains, achieving standardized classification of defect features through a hierarchical structure, facilitating subsequent vector calculation and route matching. The specific classification is as follows: Dimension 1: Structure and Appearance (Primary Dimension) Body structure failure area: includes defects such as screen cracks, shell deformation, cracks, paint peeling, water corrosion marks, etc. Mechanical component failure domain: includes defects such as button malfunction (power / volume), damaged card tray, loose interface (charging / earphone), and hinge failure (foldable screen); Dimension 2: Electronics and Hardware Dimension (Primary Dimension) Motherboard core fault domain: includes defects such as motherboard leakage, failure to power on, no baseband, system crash and restart, and overheat protection; Functional module fault domain: includes defects such as camera damage (front and rear), flash not working, earpiece / speaker not working, microphone failure, etc. Sensor system fault domain: includes defects such as fingerprint recognition failure, face unlock failure, and gyroscope / accelerometer calibration abnormalities; Dimension 3: Display and Interaction Dimension (Primary Dimension) Screen display failure domain: includes defects such as black screen, distorted screen, lines, uneven backlight, and touch failure (disconnection / drift); Battery energy failure domain: includes defects such as battery swelling, inability to charge, severe reduction in battery life, and wireless charging failure; Dimension 4: Software and Security (Primary Dimension) System software fault domain: includes defects such as system lag, low-level errors, inability to enter the system, and abnormal pre-installed software; Security credential fault domain: includes defects such as lost / damaged device authentication key, missing digital rights management key, and factory reset protection lock not unlocked; Radio frequency communication fault domain: includes defects such as invalid IMEI / MEID (Mobile Equipment Identifier), lost Wi-Fi / Bluetooth Media Access Control address, weak signal, and no network registration.

[0026] S13. Implementation methods of dynamic routing algorithms The dynamic routing algorithm of this invention is divided into two stages: initial process route generation and real-time hot reconfiguration, and the specific implementation is as follows: S131, Initial Process Route Generation The dynamic routing engine receives the device's multidimensional defect feature vector. Then, iterate through the process atomic library and match all atoms that meet the prerequisites. The process atomic nodes generate initial optimal paths based on a topology sorting algorithm. This topology sorting must adhere to the following core constraints: The disassembly and assembly sequence follows the process logic of disassembly → repair or replacement → assembly. The logic is as follows: the disassembly process must be completed before the screen replacement and other repair processes can be performed; the assembly process must be completed before the overall testing process can be performed. The hardware and software dependency constraints follow the process logic of hardware repair completion → software flashing → functional testing. The logic is as follows: the repair of core hardware such as the motherboard must be completed before the system flashing process can be performed; the system flashing process must be completed before the whole machine functional testing process can be performed. The security pre-constraints follow the process logic of environmental fingerprint verification passing → key injection. The logic is that the security key injection process must be arranged after all high-risk operations that may cause short circuits in the motherboard (such as screen pressing), and must be executed immediately after environmental fingerprint verification passes.

[0027] The dependency verification logic for process atomic nodes is as follows: whether any process atomic node can be executed depends on whether all its preconditions and their corresponding dependent nodes have been successfully completed. For example, the preconditions for the photosensitive calibration node are the completion of screen replacement and system programming. If the system detects that the system programming process has failed or has not been executed, the photosensitive calibration node will be grayed out or marked as not allowed to be executed in the process route.

[0028] S132. Optimal Path Evaluation and Determination This invention constructs a comprehensive path scoring model using multi-dimensional evaluation indicators to select the optimal path from all legal topological paths. The specific evaluation indicators are shown in the table below: Table 1 Evaluation Index of Optimal Process Route Based on the above indicators, define the parameters for calculating the overall path score: Time(P): The cumulative standard working hours of path P; Switch(P): The number of workstation / fixture switching times involved in path P; Risk(P): The weighted sum of the historical failure rates of each process atomic node in path P; Load(P): Real-time queuing index of critical bottleneck nodes in path P.

[0029] The optimal path determination logic is as follows: among all legal topology paths, select the path with the smallest comprehensive score (Score(P)) as the initial process route to be issued to the production line.

[0030] S133, Real-time thermal reconfiguration of process route The core innovation of this invention is its thermal reconfiguration mechanism. During production, if a new defect is discovered in the equipment at a certain stage (such as disassembly), the mechanism can be activated. The edge device immediately reports and updates the multidimensional defect feature vector of the device. The dynamic routing engine triggers a hot reconfiguration mechanism, dynamically inserting missing process atomic nodes after the currently executed node, automatically adjusting subsequent dependent nodes, generating an updated process route, and immediately sending it to the workstation terminal. This hot reconfiguration is not a simple modification of the process list, but a local topology rearrangement based on a directed acyclic graph. The specific implementation process is as follows: Triggering judgment mechanism: The system monitors the defect feature vector flow in the production process in real time. When the defect corresponding to the detection result of a certain station cannot be covered by the subsequent nodes of the current process route, a reconstructing interruption signal is immediately triggered. Boundary constraint algorithm execution: Before performing reconstruction, the algorithm first locks the set of currently completed nodes and loads a global constraint template, which defines the indestructible process dependencies (such as burning before testing) to ensure that the reconstructed process route conforms to the basic process logic. Hot reconfiguration execution steps: Step 1 (Pruning): Remove unnecessary subsequent nodes from the current directed acyclic graph of the process. For example, after defect repair, the originally planned scrapped nodes are removed. Step 2 (grafting): Based on the newly added defects, match the corresponding optimal repair process path; Step 3 (Integration): The newly added defect repair process path is grafted onto the currently executed node, the topology sort is recalculated, and the updated process route is generated. Step 4 (Hot Loading): The state machine of the workstation terminal seamlessly switches to the first node to be executed on the new route without restarting the equipment or resetting the system, ensuring the continuous operation of the production line.

[0031] Through the above mechanism, the present invention can achieve the following: for devices that only require screen replacement, it can automatically skip irrelevant steps such as motherboard repair; for devices with hidden faults discovered during production, it can automatically insert the corresponding repair steps without manual intervention or shutdown to reconfigure the process, truly realizing flexible production with one policy for each machine.

[0032] S2, Multi-source Identifier Collaborative Generation and Strong Binding Model This invention employs a seed-triggered and linkage calculation model to achieve collaborative generation and strong logical binding of multi-source identity identifiers. The specific implementation method is as follows: S21, Pre-set rule template library The system has a built-in rule template library, pre-configured with identifier generation rules for different models and sales regions (such as China, the EU, and North America). These rules include number prefix definitions, verification algorithms (such as the Luhn algorithm), length constraints, and mapping logic between various identifier fields. Specifically, the encoding rules for the device's serial number (SN) are defined as follows: The SN is 13 digits long and consists of 7 parts: year, week, model, version, place of origin, capacity, and serial number. The definitions of each field are as follows: Model number field: occupies 3 digits and is the project code, which is consistent with the internal code of the corresponding product. For example, if the internal code of the product is M95, then the model number field in the SN number is M95; if the internal code of the product is MA02, then the model number field in the SN number is A02. When the SN number changes, the corresponding box number is adjusted synchronously, and the first 6 digits of the box number are consistent with the new SN number rules. Version field: occupies 1 position and is used to identify the device version type. The corresponding relationship is as follows: Q: Full network compatible; A: Alibaba full network compatible; G: Dual 4G open; B: Alibaba dual 4G open; Y: China Mobile open; M: Alibaba China Mobile open; M: China Mobile customized; U: Alibaba China Mobile customized; D: China Unicom customized; C: Alibaba China Telecom customized; H: China Telecom customized; L: Overseas non-Latin American version; L: Overseas Latin American version.

[0033] S22. Collaborative Generation and Strong Binding Implementation Process Seed locking: By scanning the device's physical serial number, it is used as a unique seed for identifier generation, ensuring that each device corresponds to a unique combination of identifiers; Linked calculation: Based on the rule template corresponding to the device, the system calculates the complete set of identity data, including the associated SN, IMEI, MAC-WiFi (Wireless Local Area Network Media Access Control Address), MAC-BT (Bluetooth Media Access Control Address), and electronic identification code, in one go. Logical strong binding: At the database level, the group of identity data is marked as a strongly associated cluster. Any attempt to modify a single field within it will trigger an integrity check failure and be rejected by the system, thus eliminating the risk of single-field tampering from a logical perspective.

[0034] S23, Implementation method of pre-occupancy mechanism To address the identifier concurrency conflict issue in traditional production models, this invention establishes a pre-reservation mechanism: before writing identity data to the device, the system first pre-reserves and locks the set of identifiers in the database, marking them as assigned to the corresponding PSN. In the locked state, this set of identifiers is invisible to requests from any other device, preventing concurrency conflicts. The specific execution process is as follows: before writing to the device, the system sends a pre-reservation lock command to the database, assigning and locking the set of identifiers to the corresponding PSN; after the workstation tool writes the data to the phone's underlying layer, the system reads the phone's underlying data to confirm the writing is correct; after successful writing, the database permanently logically binds the set of identifiers to the corresponding PSN, marking it as completed.

[0035] The above solution eliminates human input errors or system-generated conflicts from the mathematical and logical source, ensuring the absolute consistency and compliance of device identity data and guaranteeing that the device can successfully pass the compliance verification of operators and regulatory agencies.

[0036] S3, Secure Key Injection System Based on Environmental Fingerprint Verification To address the risk of leakage of highly sensitive keys, this invention constructs a zero-trust injection environment to achieve fully controllable injection of secure keys throughout the entire process. It also includes a full-link closed-loop verification mechanism to ensure quality control throughout the entire production process. The specific implementation method is as follows: S31, Multi-dimensional Environmental Fingerprint Acquisition When a workstation terminal initiates a key injection request, the system automatically collects multi-dimensional feature information of the current environment, performs hashing processing, and forms an environmental fingerprint. The specific features collected include: Tool fingerprint: The version number and hash value of the programming software; Personnel fingerprints: the operator's biometric features or access tokens; Device fingerprint: A unique ID (identifier) ​​on the motherboard of the target device; Network fingerprint: Whitelist information of the specific VLAN (Virtual Local Area Network) or IP (address) where the workstation is located.

[0037] S32. Implementation of Dynamic Token Mechanism After the security key injection gateway receives and verifies the environmental fingerprint information, it does not directly issue the key file. Instead, it generates a one-time, short-term (e.g., 30-second) dynamic decryption command, as follows: Dynamic token generation algorithm: Dynamically decrypt the password It is generated using a time-slice-based hash algorithm, and the calculation formula is as follows: in, Represents a hash function. For environmental fingerprinting, The root key The timestamp is sliced ​​into 5-minute segments based on the current time to ensure the token's short expiration time. This represents the XOR operation, ensuring strong coupling between the environmental fingerprint and the key; Environment binding logic: Multi-dimensional environment fingerprints (including tool hash and motherboard ID) are generated during the process. First, a SHA-256 hash operation is performed to form a fingerprint digest, which directly participates in... The calculation is such that even the slightest change in the environment fingerprint (such as a tool version upgrade) will affect the generated fingerprint. Completely different, it mathematically achieves a strong binding between the token and the injection environment; Distribution and verification rules: Distribution process: Transmitted to the workstation terminal via an RSA-based asymmetric encrypted channel to prevent man-in-the-middle eavesdropping; Verification process: The workstation terminal receives... Afterwards, instead of using it directly, the expected value is recalculated using the locally collected environmental fingerprint. Only when received Expected with local computing Decryption of the keystream is only permitted when the keys match, using a two-way verification mechanism to prevent... Intercepted and replayed.

[0038] S33, Third-party data comparison and consistency verification algorithm This invention incorporates a three-way comparison and verification process at key production stages (such as after programming and before packaging). By comparing bit-level data from the device's underlying layer, physical tags, and a central database, data consistency is ensured. The specific implementation is as follows: (1) Definition of third-party data sources Source A (Device Layer): Reads data from the device's underlying registers, denoted as SystemValue; Source B (Physical Label): The label / sticker data on the device obtained through OCR recognition or scanning, denoted as LabelValue; Source C (Database Baseline): Pre-reserved record data in the central database, denoted as DBValue.

[0039] The system performs bit-level comparisons of key fields such as IMEI, SN, MAC, device color, and capacity based on the three data sources mentioned above.

[0040] (2) Consistent hashing comparison algorithm Define device identity consistency hash value : in, , , These are the 1st, 2nd, ..., nth identity feature fields to be verified. The above key fields are selected accordingly: Underlying data: IMEI, SN, MAC-WiFi, MAC-BT, and security key fingerprint read from device registers; Tag data: Body code and color box barcode read by industrial camera optical character recognition (OCR) or barcode scanner; Database data: The pre-occupancy record corresponding to this PSN in the central database.

[0041] (3) Real-time verification execution logic Start verification function Define a functional module called Real-time Verification, which requires three essential pieces of information to start: The physical serial number of the device to be verified (as the sole basis for querying); The actual identity data of the device is read directly from the device's underlying registers; Physical tag data of equipment identified by barcode scanners or industrial cameras.

[0042] Step 1: Obtain baseline data from the database The system first uses the input physical sequence number to query the corresponding pre-occupancy record in the central database.

[0043] If the database does not find any pre-reservation information for this physical serial number, the device is immediately determined to be an illegal device, the device transfer is blocked, and a notification message is sent.

[0044] Step 2: Organize the data sets to be compared among the three parties. The system organizes the core identity fields from the three sources into three sets of data to be compared: The first group (device physical layer data): International Mobile Equipment Identity (IMEI), device serial number, and wireless LAN media access control address actually read from the device's underlying layer; The second group (physical label layer data): scanned equipment body label code and equipment color box barcode; The third group (database logical layer data): the international mobile device identifier, device serial number, and media access control address that are pre-assigned and strongly associated with the physical serial number in the central database.

[0045] Step 3: Perform core bit-level consistency comparison This is the core protection step of the verification process. It uses digital fingerprint generation to accurately compare the three sets of data, with the specific rules as follows: Rule 1: First, compare the digital fingerprints of the device's physical layer data and the database's logical layer data. If they do not match, it indicates a mismatch between the device's underlying data and the database's pre-assigned data, which may be due to a missing or incorrect number being written. Immediately block the device's operation and push the corresponding prompt message. Rule 2: Compare the digital fingerprints of the physical label layer data and the database logic layer data. If they do not match, it means that the physical label affixed to the device does not match the pre-reserved data in the database, which may be due to incorrect labeling or mixed materials on the production line. Immediately stop the device's operation and push the corresponding prompt message. Implicit rule: As long as the first two rules are met, the device physical layer data and the physical tag layer data will naturally be consistent, and there is no need to compare them separately.

[0046] Step 4: Release after verification If all the above comparison rules pass, it means that the three-party identity data of the equipment are completely consistent, the verification is successful, the system returns a release signal, and the control production line state machine allows the equipment to flow to the next workstation.

[0047] S34. Production State Machine and Anomaly Closed-Loop Management This invention defines a production state machine model for controlling the entire process of equipment operation, and is equipped with an abnormal closed-loop process to achieve immediate interception of defective products.

[0048] (1) Definition of core states of the state machine The state machine includes the following core states: Standby (SIdle), Processing (SProcessing), Verifying (SVerifying), Pass (SPass), Lock (SLock), and Rework (SRework).

[0049] (2) State transition rules Admission transition: The state machine is only allowed to transition from SIdle to SProcessing when the environment fingerprint verification is successful; Verification transition: In the SVerifying state, if the three-party data comparison module returns a consistent signal, the state machine transitions to SPass; if it returns an inconsistent signal, the state machine immediately jumps to SLock and triggers a hardware blocking signal. Anomaly Recovery: In SLock state, the system prohibits automatic device transition. An authorized administrator must intervene to fix the anomaly and manually trigger a forced verification. Only after the verification passes can the state machine transition to SRework or SPass.

[0050] (3) Abnormal closed-loop process When the device enters SLock status, the system automatically performs the following closed-loop actions: (a) Lock the corresponding workstation resources to prevent equipment from being misused; (b) Generate an exception work order and record the current environment fingerprint and error code; (c) Awaiting manual repair and review; (d) After the review is approved, update the quality traceability chain and complete the status unlocking.

[0051] This invention, through the aforementioned system, enables the immediate interception of defective products, preventing the accumulation of errors and batch scrapping. At the same time, it automatically saves the logs, screenshots, and operation context during verification, forming an unalterable quality archive and providing accurate evidence for quality traceability.

[0052] Example 2: This example uses the refurbishment of a recycled smartphone as an example to fully illustrate the method of the present invention. The specific implementation steps are as follows: Inbound Inspection and Initial Route Generation: After the equipment is put into the warehouse, the PSN of the equipment is scanned. The initial inspection found two defects: screen damage and loss of security key. The system generates corresponding multi-dimensional defect feature vectors and generates an initial process route through the dynamic routing engine: disassembly, screen replacement, assembly, key injection, and packaging. Dynamic restructuring of process routes: During the disassembly process, operators discover minor damage to the equipment wiring and record it into the system; the system receives new defect features in real time, updates the equipment's multi-dimensional defect feature vector, automatically triggers the hot reconfiguration mechanism, inserts the wiring replacement and flight time distance calibration test process nodes before the assembly process, generates the updated process route and immediately sends it to the workstation terminal. Multi-source identity data collaborative generation: When binding a device PSN, the system automatically calculates and pre-allocates a new set of IMEI, SN, and MAC addresses based on the built-in rule template to ensure that the data complies with the latest network access rules and establishes a strong association cluster in the database to achieve logical strong binding of multi-source identifiers; Secure Key Injection Execution: When performing the key injection process, the system first verifies the multi-dimensional environment fingerprint of the current workstation, such as the tool version and operator permissions. After the verification is successful, a one-time dynamic token is issued. The key is decrypted and written in memory through the token. The key is never written to the ground in plaintext throughout the entire process, ensuring injection security. Closed-loop verification and anomaly interception: Before the packaging process, the system automatically performs a three-party data comparison to verify the consistency between the device label and the underlying data of the equipment and the baseline data of the database; if the last digit of the IMEI is not consistent (human mislabeling), the system immediately locks the device and triggers an alarm to prevent defective products from being shipped. Archived Shipment: After label errors are corrected and re-verification passes, the equipment is released; the system automatically generates a digital passport for the equipment containing full-process production data, completes the equipment archiving, and allows shipment.

[0053] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "include," "contain," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0054] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for dynamically generating and executing a mobile device refurbishment process route based on multi-dimensional defect features, characterized in that, Includes the following steps: A cloud-edge-device collaborative system architecture is constructed, which is divided into a physical perception layer, an edge computing layer, a core business logic layer, and an application interaction layer from top to bottom, to collaboratively complete the entire process management of device refurbishment. The fault data and unique hardware fingerprint information of the equipment to be refurbished are collected by the physical perception layer, preprocessed by the edge computing layer, and a multi-dimensional defect feature vector based on standardized defect classification is generated and uploaded. The dynamic routing engine of the core business logic layer receives feature vectors and generates an initial optimal process route based on a predefined process atom library and topology sorting algorithm, following the disassembly and assembly order, soft and hard dependencies, and safety pre-constraints. When a new defect is detected in production, a hot reconfiguration mechanism is triggered to dynamically update the process route through local topology rearrangement of the directed acyclic graph. Based on the seed-triggered linkage calculation model, the collaborative generation, logical strong binding and pre-occupancy locking of multi-source identity identifiers of devices are completed; Based on multi-dimensional environmental fingerprint verification and a one-time dynamic token mechanism, the entire process of controllable injection of device security keys is completed; By comparing data at the bit level from the device's underlying layer, physical tags, and the central database, and combining this with production state machine control, a closed-loop verification and anomaly interception can be achieved throughout the entire process.

2. The method according to claim 1, characterized in that, The process atomic library breaks down the entire mobile device refurbishment process into the smallest indivisible process atomic nodes, each of which defines clear input preconditions and output post-effects. The input preconditions are determined based on the device's global data snapshot to determine the preprocess verification results, and the output post-effects are determined based on the job test data logs to determine the process execution results. If they do not match, the process will be rolled back and an alarm will be triggered.

3. The method according to claim 1, characterized in that, The standardized defect classification includes 4 primary dimensions and 12 secondary fault domains. The primary dimensions are divided into structure and appearance, electronics and hardware, display and interaction, and software and security. Each primary dimension has 3 secondary fault domains, realizing the standardized hierarchical classification of defect features.

4. The method according to claim 1, characterized in that, When generating the initial optimal process route, the process atomic library is first traversed to match all process atomic nodes that meet the preconditions. Then, a path comprehensive scoring model is constructed through multi-dimensional evaluation indicators to select the path with the smallest comprehensive score from the legal topology paths as the initial optimal process route. The evaluation indicators include total process time, resource switching cost, rework risk coefficient, and bottleneck station load.

5. The method according to claim 1, characterized in that, The trigger condition for the hot reconfiguration mechanism is that the defect corresponding to the workstation detection result cannot be covered by the subsequent nodes of the current process route. When hot reconfiguration is executed, the completed node set and the indestructible process dependency relationship are locked first, and then the pruning, grafting, fusion and hot loading steps are executed in sequence to complete the seamless update of the process route.

6. The method according to claim 1, characterized in that, The seed-triggered linkage calculation model uses the device's physical serial number as a unique seed. Based on pre-set rule templates for different models and sales locations, it calculates the complete set of associated identity data for the device and forms a strong association cluster at the database level. Before the identity data is written, it is pre-locked in the database, and after successful writing, it completes permanent logical binding.

7. The method according to claim 1, characterized in that, The one-time dynamic token is generated using a time-slice-based hash algorithm and is strongly coupled with a multi-dimensional environmental fingerprint consisting of tool fingerprint, personnel fingerprint, device fingerprint, and network fingerprint. The token is for short-term, one-time use and is transmitted through an asymmetric encryption channel. It must pass local bidirectional verification on the terminal before the key stream can be decrypted and injected.

8. The method according to claim 1, characterized in that, The three-party data bit-level comparison first obtains the database baseline data based on the device physical serial number, and then compares the consistency hash values ​​of the device underlying data and the database baseline data, and the physical tag data and the database baseline data respectively. If any of the comparisons are inconsistent, the device flow will be immediately blocked and an alarm will be triggered.

9. The method according to claim 1, characterized in that, The production state machine includes core states such as standby, processing, verification, qualified, locked, and rework. The access transition of the state machine requires environmental fingerprint verification to pass. If the verification is inconsistent, it will immediately jump to the locked state and trigger hardware blocking. In the locked state, an authorized administrator must intervene to repair and verify before the state can be unlocked.

10. The method according to claim 1, characterized in that, The physical sensing layer is responsible for the acquisition of raw data at the bottom layer, the edge computing layer is responsible for real-time data preprocessing and low-latency process routing jump response, the core business logic layer has four built-in functional engines: dynamic routing engine, multi-source identifier collaborative generator, security key injection gateway, and real-time closed-loop verification center, and the application interaction layer provides visual management and full lifecycle traceability functions.

Images