Power electronic component sales management system and method based on Internet of Things

Abstract

The invention discloses a power electronic component sales management system and method based on the Internet of Things, and relates to the technical field of sales management. The system comprises a sensing layer, a communication layer, a calculation layer, a block chain collaboration layer and a security layer, and the sensing layer is used for collecting full-process data of power electronic components and generating a unique industrial chain code; the communication layer is used for realizing data transmission among the sensing layer, the computing layer and the block chain collaboration layer; the calculation layer is integrated with a federated learning module, and the federated learning module is used for receiving local training data of a plurality of subjects; the block chain collaboration layer comprises a plurality of alliance chain nodes and at least three types of smart contracts, and the security layer is used for guaranteeing system data security and privacy protection. According to the method, the problems of data islands and responsibility prevarication in traditional management are solved, and the industrial chain collaboration efficiency and decision scientificity are improved.

Description

Technical Field

[0001] This invention relates to the field of sales management technology, specifically to a sales management system and method for power electronic components based on the Internet of Things. Background Technology

[0002] In the current power electronic components industry, with the rapid development of downstream fields such as new energy, photovoltaics, and industrial control, the models and specifications of components are becoming increasingly complex, and application scenarios are becoming increasingly segmented. Sales management involves multiple entities such as manufacturers, distributors, customers, and logistics providers, and is highly dependent on the full lifecycle data of components (such as production parameters, inventory environment, transportation status, and fault information). On the one hand, all entities need to collaborate on production planning, inventory allocation, and order fulfillment based on real data. On the other hand, downstream customers' needs for component traceability and after-sales responsibility definition have also increased significantly. The industry as a whole has placed higher demands on the "data collaboration" and "credibility" of sales management, which constitutes the technical background of this solution.

[0003] In existing technologies, the sales management of power electronic components mostly adopts a traditional decentralized model. Data from each entity (such as manufacturer production plans, distributor inventory, and logistics records) is stored in its own system, lacking a reliable data sharing mechanism across entities. Furthermore, after-sales liability determination relies on manual evidence gathering and negotiation. This model not only leads to large deviations in demand forecasting and low inventory allocation efficiency due to "data silos," but also, due to insufficient data transparency, makes it easy for entities to shift blame when components malfunction due to insufficient evidence. This results in long liability determination cycles, high dispute rates, and severely impacts the efficiency and trust of supply chain collaboration, making it difficult to meet the industry's demand for "reliable collaboration throughout the entire sales process." Summary of the Invention

[0004] The purpose of this invention is to provide a power electronic component sales management system and method based on the Internet of Things to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a power electronic component sales management system based on the Internet of Things, comprising the following steps: It includes a perception layer, a communication layer, a computing layer, a blockchain collaboration layer, and a security layer; The perception layer is used to collect data on the entire process of power electronic components and generate a unique supply chain code. The data on the entire process includes data generated by power electronic components during the manufacturing, inventory, transportation and use stages. The supply chain code is used to identify the unique identity and circulation node of the power electronic components. The communication layer is used to realize data transmission between the perception layer, the computing layer and the blockchain collaboration layer, including a general transmission channel for transmitting ordinary sales data, a dedicated transmission channel for transmitting on-chain shared data, and a scheduling unit for scheduling data transmission priorities. The computing layer integrates a federated learning module, which receives local training data from multiple entities, generates an industry chain-level demand prediction model through local training, gradient uploading, and gradient aggregation, and distributes the optimized model parameters to each entity. The computing layer also includes a database for storing data, which is divided into a private database for storing private data and a shared database for storing public data. The blockchain collaboration layer includes multiple consortium blockchain nodes and at least three types of smart contracts. The consortium blockchain nodes include at least manufacturer nodes, distributor nodes, customer nodes, and logistics provider nodes. Each consortium blockchain node is used to upload and share power electronic component-related data of the corresponding entity. The smart contracts include inventory collaboration contracts for inventory collaboration, order tracking contracts for order tracking, and after-sales responsibility contracts for after-sales responsibility determination. The security layer is used to ensure system data security and privacy protection, including a terminal security unit for terminal protection, an on-chain security unit for on-chain data encryption, and an access control unit for controlling access permissions. The on-chain security unit integrates a zero-knowledge proof module and a quantum random number generation module.

[0006] Preferably, the sensing layer includes an Internet of Things (IoT) chip, at least three types of smart sensors, and RFID tags; The IoT chip includes an MCU chip and a SoC chip. The MCU chip is used to perform local preprocessing of the sensing layer data, and the SoC chip integrates RFID read / write and positioning functions to read the RFID tag data and collect the location data of power electronic components. The intelligent sensor is used to collect environmental parameters and quantity data of power electronic components at different stages, and transmit the collected data to the communication layer. The RFID tag is used to store the basic parameters of the power electronic components and the supply chain code. The basic parameters include the model, production batch, production parameters and quality inspection data of the power electronic components. The supply chain code format includes at least the manufacturer identifier, batch identifier and individual item identifier.

[0007] Preferably, the smart sensor includes a temperature and humidity sensor, a vibration sensor, and an infrared counting sensor; The temperature and humidity sensor is used to collect temperature and humidity data of power electronic components during the storage and transportation stages. The temperature and humidity data are used to determine whether the storage and transportation environment meets the storage requirements of power electronic components. The vibration sensor is used to collect vibration amplitude data of power electronic components during transportation, and the vibration amplitude data is used for subsequent after-sales liability determination. The infrared counting sensor is used to collect quantity data of power electronic components during the warehousing and outbound stages, and the quantity data is used to update the inventory data in the computing layer.

[0008] Preferably, the general transmission channel of the communication layer includes a WiFi module and a Bluetooth module. The WiFi module adopts the WiFi 6 protocol, and the Bluetooth module adopts the Bluetooth 5.2 protocol. The general transmission channel is used to transmit the order amount, customer contact information, order delivery deadline, and order payment status of power electronic components. The dedicated transmission channel includes an NB-IoT module and a 5G industrial private network module. The dedicated transmission channel is used to transmit inventory status data, fault summary data, transportation trajectory data, inventory warning threshold, and order fulfillment progress of power electronic components. The scheduling unit is used to mark the fault data of power electronic components as high-priority data and transmit it to the blockchain collaboration layer through the dedicated transmission channel first, so as to ensure the real-time response of fault data.

[0009] Preferably, the computing layer further includes an edge gateway and a cloud platform; The edge gateway is equipped with an IoT operating system and integrates a local training unit of the federated learning module, which is used to receive local data from the corresponding subject and perform local model training to generate model gradients. The cloud platform includes a federated learning scheduling center and a digital twin collaboration module. The federated learning scheduling center is used to receive model gradients uploaded by each edge gateway, generate the industry chain-level demand prediction model through a gradient aggregation algorithm, and distribute the parameters of the industry chain-level demand prediction model to each edge gateway. The digital twin collaboration module is used to construct a digital twin mapping between power electronic components and the manufacturer's production line, distributor's warehouse, and customer's workshop, so as to realize cross-entity visual collaborative scheduling. The private library is used to store the privacy data of each entity, including the manufacturer's production process parameters, the distributor's customer details, and the customer's production plan data; the shared library is used to store publicly available on-chain data, including the inventory quantity, transportation trajectory, and fault summary of power electronic components.

[0010] Preferably, the consortium blockchain nodes of the blockchain collaboration layer further include supervisory nodes, which are used to verify the identity and legitimacy of each consortium blockchain node and supervise the authenticity of the on-chain data; The inventory coordination contract is used to automatically send a replenishment request to the manufacturer node when the dealer's inventory is lower than a preset safety threshold, and to trigger direct supply scheduling from the manufacturer to the customer based on the manufacturer node's in-transit inventory data. The order traceability contract is used to record the entire process node data of power electronic components from the factory to the customer's receipt. The entire process node data includes data on the factory departure node, warehousing node, delivery node, transportation node, and receipt node. The after-sales liability contract is used to automatically retrieve the full-process data on the chain after receiving fault data uploaded by the customer node, determine the responsible party for the fault according to the preset liability determination rules, and generate a liability handling agreement. The zero-knowledge proof module is used to process the privacy data uploaded by each subject, meaning that each subject can use the statistical characteristics of the privacy data but cannot obtain the original content of the privacy data.

[0011] Preferably, the terminal security unit of the security layer includes an antivirus module and a malware protection module. The antivirus module is used to detect and remove viruses from the IoT gateway and terminal devices, and the malware protection module is used to intercept malicious attacks targeting the terminal devices. In addition to the zero-knowledge proof module and the quantum random number generation module, the on-chain security unit also includes an AES-256 encryption module. The quantum random number generation module is used to generate unpredictable encryption keys, and the AES-256 encryption module is used to encrypt private data transmitted on-chain. The access control unit adopts a role-based access control policy to assign different access permissions to each consortium blockchain node. Specifically, the manufacturer node can only access the inventory data of the distributor node and the purchase plan data of the customer node, but cannot access the privacy data of the customer node; the distributor node can only access the production plan data of the manufacturer node and the order data of the customer node, but cannot access the production process parameters of the manufacturer node.

[0012] This invention also provides a sales management method for power electronic components based on the Internet of Things, comprising the following steps: Data collection is synchronized with the blockchain. Data on the entire process of power electronic components during the manufacturing, inventory, transportation and use stages is collected to generate a unique industry chain code. The entire process data and industry chain code are transmitted to the consortium blockchain, where they are uploaded and shared on the blockchain by the nodes of the corresponding entities. Federated learning demand prediction receives local training data from each entity, generates model gradients through local model training, aggregates all model gradients to generate an industry chain-level demand prediction model, and distributes the optimization parameters of the model to each entity. Orders and inventory are coordinated. After receiving a customer's order request, the inventory data on the consortium blockchain is queried. If the distributor's inventory is lower than the preset threshold, a replenishment request is automatically triggered and inventory is allocated according to the manufacturer's in-transit inventory data, and the inventory information of each entity on the chain is updated. After-sales responsibility determination: Receive power electronic component failure data uploaded by customers, retrieve full-process data on the consortium blockchain, determine the responsible party for the failure according to preset rules, and generate and push responsibility handling agreements; After-sales data closed loop: optimize production process based on on-chain fault data, conduct risk investigation on power electronic components of the same batch, and upload the process optimization results and investigation results to the consortium blockchain to update the public data repository.

[0013] The present invention also provides an electronic device, which is a physical device, comprising: The processor and the memory are communicatively connected. The memory is used to store at least one executable instruction executed by the processor, which executes the executable instruction to implement the IoT-based power electronic component sales management system as described above.

[0014] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described Internet of Things-based power electronic component sales management system.

[0015] Compared with the prior art, the beneficial effects of the present invention are: By collecting full-process data of power electronic components through the perception layer and generating unique industry chain codes, and combining this with the blockchain collaboration layer to achieve trusted data sharing, the system achieves traceability throughout the entire lifecycle of components and prevents data silos across different entities. Through the federated learning module in the computing layer, multi-entity collaborative modeling is achieved, integrating multi-dimensional data while protecting data privacy, thus improving the accuracy of industry chain demand forecasting and assisting various entities in making scientific decisions. Through channel-based transmission and priority scheduling in the communication layer, ordinary sales data is distinguished from on-chain shared data, ensuring real-time transmission of core data and reducing the cost of conventional data transmission. Through terminal protection, on-chain encryption, and access control in the security layer, malicious attacks are resisted and data access is standardized, achieving system data security and controllability while preventing the leakage of privacy for each entity. Through smart contracts in the blockchain collaboration layer, inventory coordination and after-sales responsibility determination are automatically executed, reducing manual intervention and improving industry chain collaboration efficiency and quickly defining cross-entity responsibilities. Attached Figure Description

[0016] Figure 1 A schematic diagram of the structure of a power electronic component sales management system based on the Internet of Things provided in an embodiment of the present invention; Figure 2 The main flowchart of a power electronic component sales management method based on the Internet of Things provided in this embodiment of the invention; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] The method in this embodiment is executed by a terminal, which can be a mobile phone, tablet computer, PDA, laptop or desktop computer, etc. Of course, it can also be other devices with similar functions, and this embodiment does not limit them.

[0019] Please see Figure 1 This invention provides a power electronic component sales management system based on the Internet of Things, including a perception layer, a communication layer, a computing layer, a blockchain collaboration layer, and a security layer.

[0020] The perception layer is used to collect data on the entire process of power electronic components and generate a unique supply chain code. The data on the entire process includes data generated by power electronic components during the manufacturing, inventory, transportation and use stages. The supply chain code is used to identify the unique identity and circulation node of the power electronic components. It should be noted that, in this embodiment, "full-process data" refers to all data generated throughout the entire lifecycle of a power electronic component, from its production and delivery to the customer and its use, encompassing the component's status, environment, and operational information at different stages. "Supply chain code" refers to a unique identifier for each power electronic component and its distribution path. This code is globally unique and can be linked to all information throughout the component's lifecycle, ensuring accurate traceability. "Manufacturing stage data" refers to data generated when a power electronic component leaves the factory after production, including but not limited to production parameters (such as silicon wafer doping concentration and packaging process parameters), quality inspection results, manufacturing time, and information on the personnel conducting the manufacturing inspection. "Inventory stage data" refers to data generated during the storage of power electronic components in a warehouse, including but not limited to storage location (such as warehouse number and shelf number). Storage environment parameters (such as temperature and humidity), inventory quantity change records (inbound quantity, outbound quantity), and inventory operation personnel information; transportation stage data refers to the data generated during the logistics transportation of power electronic components, including but not limited to transportation carrier information (such as logistics vehicle number, driver information), real-time transportation location, transportation environment parameters (such as vibration amplitude, temperature), and time nodes of each transportation segment (departure time, transfer time, arrival time); usage stage data refers to the data generated during the use of power electronic components after they are received by the customer, including but not limited to operating condition parameters (such as operating current, operating temperature), fault feedback information (fault type, fault occurrence time), and maintenance records; circulation nodes refer to the entities or locations corresponding to the key links that power electronic components pass through throughout their entire life cycle, including but not limited to the manufacturer's production workshop, distributor's warehouse, logistics transportation vehicles, and customer's production workshop.

[0021] The purpose of the perception layer is to collect full-lifecycle data of power electronic components and assign a unique supply chain code to each component. This enables the unique identification of the component and the binding of full-process data, laying the data foundation for subsequent cross-entity data sharing, order collaboration, and after-sales traceability. Its effects include ensuring the traceability of power electronic component lifecycle data, avoiding undetermined responsibility or demand forecasting errors due to data gaps, and preventing component identity confusion or information tampering through the uniqueness of the supply chain code, thus improving data credibility. The perception layer is implemented by acquiring data at each stage using hardware acquisition devices (such as sensors and code generation devices), and then linking the scattered data at each stage through the supply chain code to form a complete component data archive, ensuring that the data is "one code per person, one code to the end."

[0022] In one possible implementation, the hardware devices for collecting end-to-end data in the perception layer include, but are not limited to, a SoC chip with integrated RFID read / write functionality, a temperature and humidity sensor, a vibration sensor, an infrared counting sensor, and a BeiDou positioning module. The SoC chip is used to generate and write supply chain codes, and simultaneously read basic component information stored in RFID tags. The temperature and humidity sensor is used to collect ambient temperature and humidity data during storage and transportation, with accuracy controlled within ±0.5℃ and ±2%RH. The vibration sensor is used to collect vibration amplitude data during transportation, with a measurement range covering 0-200 m / s. 2 Accuracy ±0.01m / s 2 Infrared counting sensors are used to collect the quantity of components entering and leaving the warehouse during the inventory stage, with a counting error ≤0.1%; Beidou positioning modules are used to collect real-time location during the transportation stage, with a positioning accuracy ≤1 meter; The generation rules for the supply chain code include "manufacturer identifier - batch identifier - individual item identifier - production timestamp", where the manufacturer identifier is the manufacturer's unique code in the system (e.g., "IF" for Infineon), the batch identifier is the unique code for the component's production batch (e.g., "202405" for the batch produced in May 2024), the individual item identifier is the unique serial number of each component within the same batch (e.g., 001-1000), and the production timestamp is the specific time the component leaves the factory (e.g., 202405101430 represents 14:30 on May 10, 2024). This rule ensures the global uniqueness of the supply chain code.

[0023] The communication layer is used to realize data transmission between the perception layer, the computing layer and the blockchain collaboration layer, including a general transmission channel for transmitting ordinary sales data, a dedicated transmission channel for transmitting on-chain shared data, and a scheduling unit for scheduling data transmission priorities. It should be noted that, in this application embodiment, ordinary sales data refers to non-core data generated during the sales process of power electronic components that does not need to be publicly shared on the consortium blockchain, including but not limited to order amount, customer contact information, order delivery deadline, order payment status, and customer purchasing preference records; the general transmission channel refers to the channel in the communication layer used to transmit ordinary sales data. This channel is adapted to conventional data transmission needs, focusing on the universality of transmission and cost control, and does not need to meet high real-time or high security requirements; on-chain shared data refers to core data that needs to be publicly shared among the nodes of the consortium blockchain during the sales process of power electronic components, including but not limited to inventory status data (real-time inventory quantity, inventory location) and fault summary data (fault type, fault-related industries). The communication layer includes: chain coding, transportation trajectory data (real-time location, transportation time nodes), inventory warning thresholds, and order fulfillment progress (order placement time, outbound time, and receipt time); a dedicated transmission channel refers to the channel in the communication layer used to transmit shared data on the chain. This channel is adapted to the real-time and security requirements of on-chain data, focusing on transmission stability and low latency, and can ensure reliable data transmission between entities; data transmission priority refers to the sorting rules of the communication layer for different types of data transmission. Among them, data related to fault handling and security warnings have higher priority than regular data because they require rapid response; the scheduling unit refers to the functional unit in the communication layer used to execute data transmission priority sorting and channel allocation. It can automatically complete data transmission scheduling according to preset rules without manual intervention.

[0024] The purpose of the communication layer is to achieve efficient and orderly data transmission between the perception layer, computing layer, and blockchain collaboration layer. By differentiating data types and setting up different transmission channels, and prioritizing transmission based on data importance, it ensures that different types of data can be matched with appropriate transmission resources. The benefits include avoiding transmission congestion or delays caused by mixed data types, ensuring real-time transmission of critical data (such as fault data), reducing waiting time for after-sales responsibility determination, and lowering the transmission cost of routine data (such as using low-cost WiFi protocols to transmit ordinary sales data). The communication layer is implemented by classifying data based on the "privacy-importance" dimension: ordinary sales data, with "high privacy and low importance," is allocated to a general transmission channel; on-chain shared data, with "low privacy and high importance," is allocated to a dedicated transmission channel. The transmission order is then dynamically adjusted through a scheduling unit to ensure that high-priority data occupies transmission resources first, maximizing transmission efficiency.

[0025] In one possible implementation, the communication protocols used in the general transmission channel include, but are not limited to, WiFi 6 and Bluetooth 5.2. WiFi 6 is suitable for short-range, high-bandwidth general sales data transmission within the warehouse (such as order amount statistics and customer contact information entry within the distributor's system), with a transmission rate of up to 1.2Gbps and a coverage range of ≤100 meters. Bluetooth 5.2 is suitable for low-power general sales data transmission between terminal devices and the local gateway (such as purchasing preference records submitted by customers' mobile phones), with a transmission rate of up to 2Mbps, a coverage range of ≤30 meters, and power consumption reduced by 30% compared to Bluetooth 5.0. The communication protocols used in the dedicated transmission channel include, but are not limited to, NB-IoT and 5G industrial private network protocols. NB-IoT is suitable for low-power, long-range... On-chain shared data transmission (such as uploading inventory status in remote warehouses) can reach a transmission rate of 250kbps, with a coverage range of ≤10 kilometers and power consumption of only 1 / 10 of traditional GSM modules; the 5G industrial private network protocol is suitable for high-bandwidth, low-latency on-chain shared data transmission (such as real-time updates of transportation trajectories and emergency transmission of fault data), with a transmission rate of up to 10Gbps and a latency of ≤10ms; the transmission priority rules of the scheduling unit are set as follows: fault data > inventory warning data > order fulfillment progress data > regular sales data, and when the load rate of the dedicated transmission channel exceeds 80%, the scheduling unit can temporarily store non-urgent on-chain shared data (such as inventory data in non-warning states) in a local cache (cache capacity ≥10GB), and transmit it when the channel load rate is lower than 50%, thus avoiding channel congestion.

[0026] The computing layer integrates a federated learning module, which receives local training data from multiple entities, generates an industry chain-level demand prediction model through local training, gradient uploading, and gradient aggregation, and distributes the optimized model parameters to each entity. The computing layer also includes a database for storing data, which is divided into a private database for storing private data and a shared database for storing public data. It should be noted that, in this embodiment, the federated learning module refers to a functional module integrated into the computing layer for implementing multi-agent distributed model training. It supports joint model training by various agents without sharing raw data, thus avoiding raw data leakage. Multiple agents refer to different roles participating in the power electronic component sales industry chain, including but not limited to manufacturers, distributors, purchasing customers, and logistics service providers. Each agent possesses independent business data and training resources. Local training data refers to the raw data generated by each agent within its own business scenario for model training. The types of local training data differ between agents (e.g., a manufacturer's local training data includes production capacity data, raw material price data, and production contract data). Local training data for distributors includes inventory data, historical sales data, and inventory turnover rate data; local training data for customers includes purchase volume data, purchase cycle data, and downstream industry demand fluctuation data. Local training refers to the process where each entity trains its model on its own local devices (such as edge gateways or local servers) based on local training data. During training, the raw data is not uploaded to the cloud; only the model gradient is output after training is complete. Gradient upload refers to the process where each entity, after completing local training, uploads the trained model gradient (i.e., the change in model parameters, excluding raw data information) to the cloud platform at the computing layer. Gradient aggregation refers to the process where the cloud platform at the computing layer receives the model gradients uploaded by each entity and aggregates them using... The process of using a pre-defined algorithm (such as the federated average algorithm) to fuse and calculate all gradients to generate global model parameters; a supply chain-level demand forecasting model refers to a model obtained through gradient aggregation that can be used to predict the demand trend of the entire power electronic component sales supply chain. This model integrates the local training data features of each entity and can adapt to the overall demand fluctuations of the supply chain; model parameters refer to the set of parameters that constitute the supply chain-level demand forecasting model, including but not limited to weight coefficients, bias terms, activation function parameters, etc. These parameters determine the model's prediction accuracy; a database refers to a storage unit integrated into the computing layer for storing various types of data, enabling data classification management and secure storage; privacy data refers to data that entities do not wish to disclose, involving trade secrets or... Sensitive information includes, but is not limited to, manufacturers' production process parameters and cost data, distributors' detailed customer information and profit data, and customers' production plan data and procurement budget data; private databases refer to sub-databases in the database used to store private data, which are only accessible to the data owner and cannot be viewed or modified by other entities; public data refers to data that all entities agree to share within the industry chain and does not involve sensitive information, including but not limited to the inventory quantity, transportation trajectory, fault summary, and publicly disclosed production plans (such as the quarterly production capacity announced by manufacturers); shared databases refer to sub-databases in the database used to store public data, which are accessible to all consortium blockchain nodes, and each node can query or use the data in the shared database.

[0027] The purpose of the computation layer is to achieve cross-entity collaborative modeling while protecting the data privacy of each entity, generating accurate supply chain-level demand forecasting models to provide data support for the production, inventory, and procurement decisions of each entity. Simultaneously, it achieves secure data management through distributed storage. The benefits include preventing the leakage of trade secrets due to data sharing (e.g., manufacturers can participate in model training without disclosing their production processes), improving the adaptability of the demand forecasting model to the overall trends of the supply chain (by integrating data features from multiple entities), and reducing the risk of inventory backlog or stockouts (e.g., distributors can prepare inventory in advance based on model forecasts). The computation layer is implemented based on the "data remains still, model moves" concept of federated learning, allowing each entity to complete training locally and only upload model gradients. A global model is then generated through cloud aggregation, ensuring data privacy. Simultaneously, data is divided into private and public data based on "privacy," stored in private and shared repositories respectively, achieving hierarchical data management that protects privacy while meeting sharing needs.

[0028] In one possible implementation, the gradient aggregation algorithm used by the federated learning module includes, but is not limited to, the federated averaging algorithm (FedAvg) and the federated momentum algorithm (FedMomentum). The federated averaging algorithm is suitable for scenarios with relatively uniform data distribution (such as consistent fluctuation trends in sales data across distributors), obtaining the global gradient by weighting the model gradients of all subjects according to the number of samples. The federated momentum algorithm is suitable for scenarios with uneven data distribution (such as significant differences in purchasing needs among customers in different regions), improving the stability of gradient aggregation and reducing the impact of abnormal gradients on the global model by introducing a momentum factor (e.g., 0.9). The training cycle of the federated learning module is set to once a week, with each training session... The training duration is ≤2 hours to avoid consuming excessive computing resources. The database adopts a distributed storage architecture, where the storage nodes of the private database are deployed on the local servers of each entity (e.g., the manufacturer's private database is deployed on the manufacturer's headquarters server, and the distributor's private database is deployed on the distributor's local data center). It interacts with the cloud platform only through an encrypted interface (using AES-256 encryption) to ensure that private data is not remotely stolen. The storage nodes of the shared database are deployed on the cloud platform and adopt a multi-replica backup mechanism (the number of replicas is ≥3, stored on servers in different regions) to ensure the reliability of public data. Furthermore, data updates in the shared database must be verified by the digital signatures of at least two consortium blockchain nodes before they take effect to prevent data tampering.

[0029] The blockchain collaboration layer includes multiple consortium blockchain nodes and at least three types of smart contracts. The consortium blockchain nodes include at least manufacturer nodes, distributor nodes, customer nodes, and logistics provider nodes. Each consortium blockchain node is used to upload and share power electronic component-related data of the corresponding entity. The smart contracts include inventory collaboration contracts for inventory collaboration, order tracking contracts for order tracking, and after-sales responsibility contracts for after-sales responsibility determination. It should be noted that, in this embodiment, a consortium blockchain node refers to a main node with independent data storage and interaction functions that joins the power electronic component sales industry chain consortium blockchain. Each node must undergo identity verification before joining the consortium blockchain, and the node data has the characteristic of being tamper-proof. A manufacturer node refers to a consortium blockchain node operated by a power electronic component manufacturer, mainly responsible for uploading manufacturer-related power electronic component data (such as production plans, quality inspection reports, factory data, and process improvement plans), and can also query public data of other nodes (such as distributor inventory and customer procurement plans). A distributor node refers to a consortium blockchain node operated by a power electronic component sales distributor, mainly responsible for uploading distributor-related power electronic component data (such as real-time inventory, sales trends, inventory warnings, and customer order summaries), and can also query manufacturer production plans and logistics provider transportation data. A customer node refers to a consortium blockchain node operated by a power electronic component purchasing customer, mainly responsible for uploading customer-related power electronic component data (such as procurement plans, fault feedback, usage condition summaries, and receipt confirmations), and can also query distributor inventory and logistics provider transportation tracks. A logistics provider node refers to a consortium blockchain operated by a power electronic component logistics service provider. Nodes are primarily responsible for uploading logistics-related data on power electronic components (such as transport carrier information, real-time transport location, transport environment parameters, and transport time points). They can also query order delivery and receipt addresses and component storage requirements. Smart contracts refer to code programs deployed on the consortium blockchain that can automatically execute based on preset conditions, completing specific functions without human intervention, and their execution results are immutable. Inventory coordination contracts are subroutines within smart contracts used to achieve supply chain inventory coordination. They can automatically trigger replenishment requests or inventory allocation logic based on inventory thresholds to ensure a balance between supply and demand. Order traceability contracts refer to subroutines within smart contracts that use… The subroutine records and queries the entire process data of power electronic component orders, enabling full traceability from order placement to receipt, facilitating order progress tracking for all parties involved. The after-sales responsibility contract refers to the subroutine in the smart contract used to determine the responsible party for after-sales failures of power electronic components. It can automatically match responsibility determination rules based on the full process data on the blockchain and generate a responsibility handling agreement. Power electronic component related data refers to all data generated by each consortium blockchain node during operation that is related to the sales of power electronic components, including but not limited to production data, inventory data, order data, logistics data, failure data, and process improvement data.

[0030] The purpose of setting up the blockchain collaboration layer is to achieve trusted data sharing and automated collaboration among all entities in the power electronic component sales industry chain. It leverages the decentralized nature of the consortium blockchain to ensure data immutability (data uploads require verification by multiple nodes; tampering necessitates cracking encryption on most nodes, resulting in extremely high costs). Smart contracts enable automated processing of inventory, orders, and after-sales service, resolving issues of data silos and blame-shifting across entities. The benefits include increased credibility and transparency of industry chain data (each node can query on-chain data without fear of data falsification), reduced manual collaboration costs (e.g., inventory replenishment requires no manual communication, contracts are automatically triggered), and shortened order processing cycles (from 3-5 days to 1-2 days) and after-sales responsibility determination cycles (from 72 hours to 1 hour). The blockchain collaboration layer's implementation involves building a "multi-entity co-governance" consortium blockchain network, clearly defining each node's data upload responsibilities and access permissions. Simultaneously, it solidifies the core collaboration logic of the industry chain (e.g., inventory alerts, order traceability, responsibility determination) into smart contracts, ensuring unified execution of collaboration rules through automated code execution, avoiding the subjectivity and errors of manual operations.

[0031] In one possible implementation, the access review process for consortium blockchain nodes includes: node applicants submit business licenses, industry qualification certificates (e.g., manufacturers need to provide production licenses, distributors need to provide sales qualifications), legal representative identity certificates, and other materials to a regulatory node (e.g., an industry association). The regulatory node verifies the authenticity of the materials (including online verification and offline spot checks). Upon successful verification, a unique on-chain digital certificate (containing the node's public key and identity information) is generated for the applicant. The applicant activates the node and joins the consortium blockchain using the digital certificate. If a node is found to have falsified data or failed to upload data as required, the regulatory node can suspend or revoke its digital certificate and terminate its access rights. The triggering conditions for the inventory coordination contract include: the real-time inventory of the distributor node is lower than a preset safety threshold (safety threshold = average daily sales over the past 30 days × 7), and the manufacturer node's in-transit inventory can cover the stockout (in-transit inventory ≥ stockout). When both conditions are met simultaneously... The contract automatically sends replenishment requests to the manufacturer node and generates direct supply scheduling instructions from the manufacturer to the customer (including component model, quantity, and delivery address). Once confirmed by the manufacturer node, the goods can be shipped directly to the customer without going through a distributor. The order traceability contract records full-process node data, including: factory exit node (factory exit time, factory inspection personnel, quality inspection report number), warehousing node (warehousing time, warehouse quantity, warehouse review personnel), outbound node (outbound time, outbound quantity, outbound review personnel, order number), transportation node (transport vehicle number, driver information, real-time location update records, transportation environment parameter records), and receipt node (receipt time, recipient information, receipt confirmation status). Each node must attach a digital signature when uploading data to ensure data traceability. The after-sales liability contract's liability determination rules include: if on-chain transportation data shows that transportation environment parameters exceed standards (e.g., vibration amplitude > 50m / s)... 2 If the temperature is >40℃ or <0℃, the logistics provider node is deemed responsible and must bear the cost of component repair or replacement; if on-chain inventory data shows that storage environment parameters exceed the standard (e.g., temperature >40℃ or humidity >60%), the distributor node is deemed responsible; if on-chain factory data shows that quality inspection is unqualified (e.g., parameters do not meet standards, appearance defects), the manufacturer node is deemed responsible; if on-chain usage data shows that operating conditions exceed the standard (e.g., operating current exceeds the rated value, operating temperature exceeds the upper limit), the customer node is deemed responsible.

[0032] The security layer is used to ensure system data security and privacy protection, including a terminal security unit for terminal protection, an on-chain security unit for on-chain data encryption, and an access control unit for controlling access permissions. The on-chain security unit integrates a zero-knowledge proof module and a quantum random number generation module. It should be noted that, in this embodiment, the terminal security unit refers to the functional unit in the security layer used to protect the security of terminal devices such as IoT gateways, terminal acquisition devices (such as sensors, RFID readers), and client terminals (such as computers, mobile phones), and can prevent malicious attacks (such as virus infection and malware intrusion) faced by terminal devices; the on-chain security unit refers to the functional unit in the security layer used to protect the security of data transmission and storage on the consortium blockchain, and can realize data encryption protection and privacy hiding to prevent data from being tampered with or leaked during transmission or storage; the access control unit refers to the functional unit in the security layer used to control the access permissions of each consortium blockchain node, and can allocate differentiated access permissions based on the subject role to ensure that each node can only access data within its authorized scope; the zero-knowledge proof module refers to the module integrated into the on-chain security unit used to realize "data..." The "available but not visible" functional module allows verifiers to verify data authenticity without obtaining the original data (e.g., verifying whether a customer's procurement plan meets the target without needing to see the specific procurement amount); the quantum random number generation module refers to a functional module integrated into the on-chain security unit used to generate unpredictable encryption keys. Based on the quantum no-cloning principle (quantum states cannot be precisely copied), it ensures the randomness of the keys and avoids brute-force attacks; data security refers to ensuring that all data in the system is not accessed, tampered with, or leaked without authorization during collection, transmission, storage, and use, covering the entire data lifecycle; privacy protection refers to the mechanism that protects the sensitive information of each entity in the system (such as customer contact information, manufacturer production processes, and distributor profit data) from being obtained by unauthorized entities, ensuring that sensitive information is only accessible to authorized entities.

[0033] The purpose of the security layer is to build a comprehensive security protection system for the power electronic component sales management system. This system provides multi-dimensional security protection from terminal devices and on-chain data to access permissions, while balancing the needs of data sharing and privacy protection (e.g., allowing shared inventory data while protecting customer contact information). Its effects include preventing malicious control of terminal devices (e.g., sensor intrusion leading to environmental data tampering), preventing on-chain data tampering or leakage (e.g., order amount theft, inventory data alteration), and preventing unauthorized access to sensitive information (e.g., competitors obtaining manufacturer production processes). This enhances the overall credibility and security of the system and strengthens the willingness of various stakeholders to use it. The security layer is implemented using a "layered protection + technology integration" strategy: the terminal layer uses antivirus and malware protection modules to defend against attacks and ensure the authenticity of collected data; the on-chain layer uses AES-256 encryption, quantum random number generation, and zero-knowledge proof modules to protect data and ensure secure transmission and storage; and the access control layer restricts access through role management modules to ensure data is only used by authorized parties, forming a complete security protection chain from "terminal-transmission-storage-access".

[0034] In one possible implementation, the antivirus module of the terminal security unit employs the following virus detection and removal strategies: periodically (e.g., at 2 AM daily, when terminal device usage is low) performing a full virus scan on the IoT gateway and terminal devices, covering system files, applications, and stored data; real-time monitoring of device process running status and network connection status, immediately terminating processes, disconnecting network connections, and isolating related files (e.g., moving suspected infected files to an encrypted quarantine area) upon detecting malicious processes (e.g., unauthorized data theft processes, processes abnormally consuming CPU) or abnormal network connections (e.g., connecting to a known malicious IP address); and simultaneously sending an alert notification to the administrator (including alert type, device number, and abnormal time); and preventing malicious software from entering the network. The software protection module employs the following interception strategies: static interception based on a known malware signature database (updated weekly, containing the hash values ​​and behavioral characteristics of the latest malware); dynamic interception based on behavioral analysis (such as abnormal data transmission behavior and frequent reading / writing of sensitive files); and sandboxing of unknown software (testing software behavior in an isolated environment and allowing it to run only after confirming its security). The AES-256 encryption module in the on-chain security unit is used to encrypt private data transmitted on-chain (such as customer contact information, order amounts, and manufacturer production costs). The encryption key is generated by a quantum random number generation module, with a key length of 256 bits. It is not stored in any physical medium but is only temporarily generated and securely passed during data transmission. The channel (such as a quantum key distribution channel) is transmitted to the receiver, and the key is destroyed immediately after transmission. The zero-knowledge proof module adopts a "non-interactive zero-knowledge proof" implementation method. The verifier does not need to interact with the prover multiple times. The authenticity of the data can be verified only through the proof document provided by the prover. For example, if a distributor needs to verify whether a customer's purchase plan meets the requirement of "monthly purchase quantity ≥ 100 units", the customer does not need to provide specific purchase quantities. They only need to generate a proof document, and the distributor can verify it through the proof document. The role access control policy of the permission management unit includes: the access permissions of the manufacturer node are "can view the inventory data and inventory warning data of the distributor node, the purchase plan data of the customer node (only the total amount, no details), and the transportation data of the logistics provider node, but cannot view..." The access permissions for customer nodes are as follows: contact information for customer nodes, customer details and profit data for distributor nodes, and production process data for other manufacturers. For distributor nodes, access permissions are: "Can view production plan data and publicly available process data for manufacturer nodes, order data (order quantity and delivery date only, no budget) for customer nodes, and transportation trajectory data for logistics providers; cannot view production cost data for manufacturer nodes, production plan details for customer nodes, or inventory and sales data for other distributors." For customer nodes, access permissions are: "Can view inventory data and price data for distributor nodes, transportation trajectory and environmental data for logistics providers, and publicly available production plans for manufacturer nodes; cannot view production process and cost data for manufacturer nodes, or other customer data and profit data for distributor nodes."The access permissions for logistics provider nodes are: "Can view order delivery and receipt addresses, component storage requirements, and customer node receipt information; cannot view order amounts, customer contact information, or internal data of manufacturers and distributors."

[0035] In one optional embodiment, the sensing layer includes an Internet of Things (IoT) chip, at least three types of smart sensors, and RFID tags; The IoT chip includes an MCU chip and a SoC chip. The MCU chip is used to perform local preprocessing of the sensing layer data, and the SoC chip integrates RFID read / write and positioning functions to read the RFID tag data and collect the location data of power electronic components. The intelligent sensor is used to collect environmental parameters and quantity data of power electronic components at different stages, and transmit the collected data to the communication layer. The RFID tag is used to store the basic parameters of the power electronic components and the supply chain code. The basic parameters include the model, production batch, production parameters and quality inspection data of the power electronic components. The supply chain code format includes at least the manufacturer identifier, batch identifier and individual item identifier.

[0036] It should be noted that, in this embodiment, the IoT chip refers to the core hardware in the sensing layer used for data processing and functional integration. The MCU chip (microcontroller unit) is a chip with data processing and peripheral control capabilities, used to perform preliminary preprocessing on the temperature, humidity, quantity, and other data collected by the sensing layer (such as filtering outliers and format conversion) to ensure direct data transmission. The SoC chip (system-on-a-chip) is a chip integrating multiple functional modules. Here, the integrated RFID read / write function is used to read component information stored in RFID tags, while the positioning function collects component location data through BeiDou or GPS modules, achieving integrated "information reading + location tracking." The smart sensor refers to a dedicated device in the sensing layer used to collect physical environment or quantity information, capable of converting physical quantities (such as temperature, vibration, and quantity) into electrical signals and transmitting them; it is the direct execution component for acquiring data throughout the entire process. The RFID tag refers to... These are electronic tags attached to power electronic components to store information. They interact with the SoC chip via radio frequency signals to write and read information. Basic parameters refer to the core attribute information of power electronic components, including model (e.g., FF200R12KT4 for IGBT modules), production batch (e.g., 202405), production parameters (e.g., silicon wafer thickness, packaging pressure), and quality inspection data (e.g., inspection results, inspection personnel number). These parameters are the basic identifiers of component identity and quality. The manufacturer identifier in the supply chain coding format refers to the manufacturer's unique abbreviation in the system (e.g., "IF" for Infineon), the batch identifier refers to the time or sequence code of the component's production batch (e.g., the batch in May 2024 is "202405"), and the single item identifier refers to the unique serial number of each component within the same batch (e.g., 001-1000). The combination of these three ensures the uniqueness of the supply chain coding. As a further extension of the perception layer, the core is to achieve accurate collection and unique identification of data from the entire process of power electronic components through a hardware combination of "IoT chip + smart sensor + RFID tag". The purpose of this design is to solve the problem that a single collection device cannot cover the multiple needs of "data processing, environmental collection, and information storage". The effects achieved include more accurate data collection (preprocessing reduces outliers), tighter information association (RFID tags bind full-process data), and higher functional integration (SoC chip reduces hardware redundancy). The implementation idea is to divide the hardware components according to the logic of "data collection-processing-storage-identification": smart sensors are responsible for "collection", MCU chips are responsible for "processing", RFID tags are responsible for "storage", and SoC chips are responsible for "integrated control and interaction", forming a closed loop of data collection in the perception layer.

[0037] In one possible implementation, the MCU chip is selected from the STM32L4 series, which features low power consumption (sleep current ≤1μA), adapting to the long-term power consumption requirements of warehouse or transportation scenarios. It can filter outliers in the collected temperature and humidity data (e.g., removing invalid data exceeding the -40℃~85℃ range). The SoC chip is an RFID reader / writer chip with integrated BeiDou positioning, achieving a positioning accuracy ≤1 meter, an RFID read / write distance ≤10 cm, and supporting the ISO15693 protocol, ensuring compatibility with mainstream RFID tags. The number of smart sensors is configured according to the scenario, with one temperature sensor deployed for every 20㎡ in the warehouse. Humidity sensors and one infrared counting sensor are deployed per row of shelves, and one vibration sensor is deployed per vehicle (such as a truck) for logistics transportation. The RFID tags are ultra-high frequency (UHF) tags with a storage capacity of ≥512 bytes, capable of storing at least 10 historical data records (such as multiple inventory location change records), and an IP68 protection rating, suitable for humid warehouse or dusty transportation environments. The generation of supply chain codes is automatically completed by the SoC chip when the components leave the factory. The coding format strictly follows "manufacturer identification - batch identification - individual product identification", and after generation, it is encrypted and stored in the RFID tag using a hash algorithm to prevent the code from being tampered with.

[0038] In one optional embodiment, the smart sensor includes a temperature and humidity sensor, a vibration sensor, and an infrared counting sensor; The temperature and humidity sensor is used to collect temperature and humidity data of power electronic components during the storage and transportation stages. The temperature and humidity data are used to determine whether the storage and transportation environment meets the storage requirements of power electronic components. The vibration sensor is used to collect vibration amplitude data of power electronic components during transportation, and the vibration amplitude data is used for subsequent after-sales liability determination. The infrared counting sensor is used to collect quantity data of power electronic components during the warehousing and outbound stages, and the quantity data is used to update the inventory data in the computing layer.

[0039] It should be noted that, in this embodiment, the temperature and humidity sensor refers to a device in the smart sensor suite used to collect the ambient temperature and humidity of power electronic components during storage and transportation. The data it collects directly reflects whether the storage environment of the components meets the requirements (e.g., IGBT modules require 0-40℃ and 30%-60%RH); the vibration sensor refers to a device in the smart sensor suite used to collect the vibration amplitude of the components during transportation. Vibration amplitude exceeding the standard (e.g., >50m / s) is considered a violation. 2This may cause component solder joints to detach or the package to be damaged; Infrared counting sensors refer to devices in smart sensors used to collect the quantity of components during the warehousing and outbound stages. They achieve non-contact counting through infrared sensing technology, avoiding errors from manual counting; Temperature data refers to the ambient temperature value collected by the temperature and humidity sensor, and humidity data refers to the ambient relative humidity value collected by the sensor. Together, they constitute the storage environment parameters of the components; Vibration amplitude data refers to the acceleration value corresponding to the maximum displacement of vibration per unit time collected by the vibration sensor, used to assess the impact of transportation on components; Quantity data refers to the number of components collected by the infrared counting sensor for warehousing or outbound, which is the core basis for updating inventory data. As a further subdivision and expansion of intelligent sensors, the core is to clarify the data collection objects and functional positioning of different types of sensors. The purpose of setting them up is to provide accurate scenario-based data collection solutions for the pain points of power electronic components, such as "susceptibility to abnormal environment during storage, vibration during transportation, and counting errors during warehousing and outgoing". The effects that can be achieved include avoiding component failure caused by abnormal environment (such as early warning of excessive temperature and humidity), accurately locating responsibility for transportation damage (such as vibration data tracing), and reducing inventory counting errors (such as infrared non-contact counting). The implementation idea is to match the corresponding sensor type according to the scenario requirements of "storage environment - transportation safety - quantity statistics" to ensure that each key link is covered by dedicated data collection equipment.

[0040] In one possible implementation, the temperature and humidity sensor is model SHT31, which has a response time ≤8 seconds, annual drift ≤0.1℃ / year and ≤0.5%RH / year, supports the I2C communication protocol, can communicate directly with the MCU chip of the sensing layer, and has a self-calibration function, automatically calibrating once a month to ensure data accuracy; the vibration sensor is model ADXL345, with a measurement range of ±16g (1g≈9.8m / s). 2 It features a resolution of 0.018 g / LSB, supports the SPI communication protocol, and allows setting a vibration threshold (e.g., 50 m / s). 2 When the threshold is exceeded, a local alarm is automatically triggered (e.g., indicator light flashing), and warning data is sent to the MCU chip at the same time. The infrared counting sensor is model E3Z-LS63, with a detection distance of 5-30 cm, a response time of ≤1 ms, and anti-light interference capability (can work normally under warehouse LED lights). When a component passes through the sensor detection area, it automatically outputs a counting signal and supports continuous counting (up to 1000 counting records can be stored) to prevent data loss. In addition, all three sensors have low power consumption characteristics, with an operating current of ≤10mA, suitable for battery-powered scenarios (such as vibration sensors in transportation), and a battery life of ≥30 days.

[0041] In an optional embodiment, the general transmission channel of the communication layer includes a WiFi module and a Bluetooth module. The WiFi module adopts the WiFi 6 protocol, and the Bluetooth module adopts the Bluetooth 5.2 protocol. The general transmission channel is used to transmit the order amount, customer contact information, order delivery deadline, and order payment status of power electronic components. The dedicated transmission channel includes an NB-IoT module and a 5G industrial private network module. The dedicated transmission channel is used to transmit inventory status data, fault summary data, transportation trajectory data, inventory warning threshold, and order fulfillment progress of power electronic components. The scheduling unit is used to mark the fault data of power electronic components as high-priority data and transmit it to the blockchain collaboration layer through the dedicated transmission channel first, so as to ensure the real-time response of fault data.

[0042] It should be noted that in this embodiment, the WiFi module refers to a communication component using the WiFi 6 protocol in the general transmission channel. WiFi 6 is the sixth-generation wireless local area network protocol, characterized by high bandwidth and low latency. The Bluetooth module refers to a communication component using the Bluetooth 5.2 protocol in the general transmission channel. Bluetooth 5.2 supports long-range, low-power communication. The NB-IoT module refers to a communication component using a narrowband IoT protocol in the dedicated transmission channel. The NB-IoT protocol is suitable for low-power wide-area communication. The 5G industrial private network module refers to a communication component using a 5G industrial private network protocol in the dedicated transmission channel. 5G industrial private networks feature low latency and high reliability. Ordinary sales data refers to data that does not require on-chain sharing. Non-core sales information includes order amount (e.g., 100,000 yuan), customer contact information (e.g., 138XXXX1234), order delivery period (e.g., 7 days), and order payment status (e.g., paid). On-chain shared data refers to core data that needs to be shared among consortium blockchain nodes, including inventory status data (e.g., real-time inventory of 800 units), fault summary data (e.g., fault type "no output when powered on"), transportation trajectory data (e.g., Shanghai → Shenzhen), inventory warning threshold (e.g., 1050 units), and order fulfillment progress (order placed → outbound → signed for). The scheduling unit refers to the functional component in the communication layer used to schedule data transmission according to priority. High-priority data refers to data that needs urgent processing (e.g., fault data), which occupies transmission resources first. As a further extension of the communication layer, the core is to achieve efficient and orderly data transmission through "channel-specific transmission + priority scheduling". The purpose of this design is to solve the congestion or delay problems caused by the mixed transmission of different types of data (such as ordinary sales data occupying bandwidth and causing delays in faulty data). The effects achieved include reducing the cost of ordinary data transmission (such as the low cost of WiFi 6), ensuring the real-time performance of core data (such as the low latency of 5G industrial private networks), and avoiding waste of transmission resources. The implementation idea is to divide data types according to "data privacy - real-time performance": ordinary sales data has "high privacy and low real-time performance" and uses a general transmission channel; on-chain shared data has "low privacy and high real-time performance" and uses a dedicated transmission channel. Then, the scheduling unit ensures that high-priority data is transmitted first.

[0043] In one possible implementation, the WiFi module is the AX200 model, supporting the WiFi 6 protocol, with a transmission rate of up to 2.4Gbps and a coverage range of ≤100 meters. It is suitable for short-range general sales data transmission within the warehouse (such as order amount statistics within the distributor's premises). Two WiFi routers are deployed in each warehouse to ensure no dead zones. The Bluetooth module is the CSR8675 model, supporting the Bluetooth 5.2 protocol, with a transmission rate of up to 2Mbps, a coverage range of ≤30 meters, and power consumption of ≤5mA. It is suitable for low-power data transmission between the customer's mobile phone and the local gateway (such as customer-submitted purchase preference records). The NB-IoT module is the BC95-B5 model, supporting the NB-IoT protocol, with a transmission rate of up to 250kbps and a coverage range of ≤10 kilometers. With a power consumption of ≤1μA (sleep mode), it is suitable for on-chain shared data transmission in remote warehouses (such as uploading inventory status in a mountainous warehouse); the 5G industrial private network module uses the ME909S-521 model, which supports 5G SA standalone networking, with a transmission rate of up to 1Gbps and a latency of ≤10ms, suitable for core data transmission such as fault data and transportation trajectories; the priority rules of the scheduling unit are set as follows: fault data (priority 1) > inventory warning data (priority 2) > order fulfillment progress data (priority 3) > ordinary sales data (priority 4), and when the load rate of the dedicated transmission channel exceeds 80%, the scheduling unit will temporarily store the on-chain shared data of priority 3 and below in the local cache (cache capacity 10GB), and transmit it when the load rate is lower than 50%, so as to avoid channel congestion.

[0044] In one alternative embodiment, the computing layer further includes an edge gateway and a cloud platform; The edge gateway is equipped with an IoT operating system and integrates a local training unit of the federated learning module, which is used to receive local data from the corresponding subject and perform local model training to generate model gradients. The cloud platform includes a federated learning scheduling center and a digital twin collaboration module. The federated learning scheduling center is used to receive model gradients uploaded by each edge gateway, generate the industry chain-level demand prediction model through a gradient aggregation algorithm, and distribute the parameters of the industry chain-level demand prediction model to each edge gateway. The digital twin collaboration module is used to construct a digital twin mapping between power electronic components and the manufacturer's production line, distributor's warehouse, and customer's workshop, so as to realize cross-entity visual collaborative scheduling. The private library is used to store the privacy data of each entity, including the manufacturer's production process parameters, the distributor's customer details, and the customer's production plan data; the shared library is used to store publicly available on-chain data, including the inventory quantity, transportation trajectory, and fault summary of power electronic components.

[0045] It should be noted that in this embodiment, "edge gateway" refers to the hardware device deployed locally on each entity in the computing layer for performing local data processing and model training; "IoT operating system" refers to the operating system (such as LiteOS) suitable for IoT devices mounted on the edge gateway, which has lightweight and low power consumption characteristics; "federated learning local training unit" refers to the functional component in the edge gateway for performing local model training, which can train models based on local data and generate gradients; "cloud platform" refers to the platform deployed in the cloud in the computing layer for aggregating model gradients and implementing collaborative scheduling; "federated learning scheduling center" refers to the functional component in the cloud platform for receiving and aggregating model gradients from each edge gateway; and "gradient aggregation algorithm" refers to... It is an algorithm that integrates the gradients of multiple subject models (such as the federated average algorithm); the digital twin collaboration module refers to the component in the cloud platform used to build a digital mapping of physical scenes, which can realize the real-time synchronization of "physical scene-digital model"; the private library refers to the sub-library in the database that stores the privacy data of each subject, including manufacturer production process parameters (such as wafer annealing temperature of 1200℃), detailed information of distributor customers (such as customer procurement budget), and customer production plan data (such as 2024Q4 capacity plan); the shared library refers to the sub-library in the database that stores public data, and the publicly available on-chain data includes the quantity of components in stock (such as 800 pieces), transportation trajectory (such as Shanghai → Shenzhen), and fault summary (such as "no output when powered on"). As a further extension of the computing layer, the core is to achieve a balance between privacy protection and collaborative modeling through "local processing + cloud collaboration + sharded storage". Its purpose is to solve the privacy leakage problem caused by traditional centralized modeling (such as uploading production data from manufacturers to the cloud). The effects it can achieve include protecting data privacy (local training does not leak the original data), improving model accuracy (integrating data from multiple subjects), and realizing hierarchical data management (separation of privacy and public data). The implementation idea is to allow each subject to complete model training locally, only uploading gradients to the cloud for aggregation, while storing data in sharded databases according to data privacy, taking into account both collaboration and security.

[0046] In one possible implementation, the edge gateway is selected from the Huawei AR551F model, equipped with the LiteOS IoT operating system. This operating system occupies ≤128MB of memory and supports multi-protocol access (such as WiFi, NB-IoT). The local training unit of federated learning adopts the TensorFlow Lite framework, which can train the model locally on the edge gateway, with a training time of ≤30 minutes (based on 100,000 historical data). The cloud platform is deployed on Alibaba Cloud ECS server. The federated learning scheduling center adopts the FedAvg federated averaging algorithm, aggregating the model gradients of 10 subjects each time. The time to generate the industry chain-level demand prediction model is ≤1 hour, and it supports incremental updates of model parameters (only updating the changed parameters). The digital twin collaboration module uses the Unity3D engine to build digital scenes, enabling real-time synchronization of manufacturer production lines (such as the equipment status of Infineon's production line), dealer warehouses (such as the shelving layout of the Suzhou warehouse), and customer workshops (such as the equipment location in BYD's workshop), with a synchronization latency of ≤500ms. The database adopts a MySQL distributed storage architecture, with private databases deployed on local servers of each entity (such as the manufacturer's private database deployed at Infineon's German headquarters), using AES-256 encrypted storage, and interacting with the cloud only through encrypted APIs. The shared database is deployed on Alibaba Cloud OSS storage service, using three-replica backup (stored separately in Shanghai, Guangzhou, and Beijing nodes), and data updates require digital signature verification from two consortium blockchain nodes to prevent tampering.

[0047] In one optional embodiment, the consortium blockchain nodes of the blockchain collaboration layer further include a supervisory node, which is used to verify the identity and legitimacy of each consortium blockchain node and supervise the authenticity of the on-chain data; The inventory coordination contract is used to automatically send a replenishment request to the manufacturer node when the dealer's inventory is lower than a preset safety threshold, and to trigger direct supply scheduling from the manufacturer to the customer based on the manufacturer node's in-transit inventory data. The order traceability contract is used to record the entire process node data of power electronic components from the factory to the customer's receipt. The entire process node data includes data on the factory departure node, warehousing node, delivery node, transportation node, and receipt node. The after-sales liability contract is used to automatically retrieve the full-process data on the chain after receiving fault data uploaded by the customer node, determine the responsible party for the fault according to the preset liability determination rules, and generate a liability handling agreement. The zero-knowledge proof module is used to process the privacy data uploaded by each subject, so as to achieve "usable but not visible" privacy data, that is, each subject can use the statistical features of the privacy data but cannot obtain the original content of the privacy data.

[0048] It should be noted that, in this embodiment, the regulatory node refers to a node in the blockchain collaboration layer operated by an industry regulatory agency (such as the China Power Electronics Industry Association), responsible for verifying the identity legitimacy and data authenticity of other nodes; identity legitimacy refers to whether the node applicant possesses the corresponding industry qualifications (e.g., manufacturers need a production license); the replenishment request of the inventory collaboration contract refers to the replenishment notification automatically sent to the manufacturer by the contract when the distributor's inventory falls below a threshold; the in-transit inventory data refers to the quantity of components that the manufacturer has produced but not yet delivered; the full-process node data of the order traceability contract refers to the key stage data of the components from leaving the factory to being signed for, including the factory entry... The points include (manufacture date, test report number), warehousing node (warehousing time, quantity), outbound node (outbound time, order number), transportation node (carrier information, location record), and signing node (signing time, signatory); the preset rules of the after-sales liability contract refer to the basis for determining the responsible party for the fault (e.g., transportation environment exceeding standards → logistics provider's responsibility), and the liability handling agreement refers to the liability assumption plan generated by the contract (e.g., repair, compensation); the "usable but invisible" nature of the zero-knowledge proof module means that the verifying party does not need to obtain the original data, but can verify the authenticity of the data only through the proof documents (e.g., verifying whether the manufacturer's production capacity meets the standard without having to view the specific production capacity data). As a further extension of the blockchain collaboration layer, its core is to enhance the credibility and security of on-chain collaboration by "adding regulatory nodes, refining smart contracts, and strengthening privacy protection." Its purpose is to address issues such as disordered node access in consortium blockchains (e.g., unqualified nodes joining), rudimentary smart contract functionality (e.g., unclear replenishment logic), and risks associated with privacy data sharing (e.g., leakage of customer procurement plans). The effects include standardizing node access (regulatory audits prevent fraudulent nodes), refining collaboration logic (clear contract rules), and protecting shared privacy (zero-knowledge proofs hide original data). The implementation involves introducing third-party regulatory nodes to control access, refining the collaboration logic into executable contract rules, and using zero-knowledge proofs to balance sharing and privacy.

[0049] In one possible implementation, the access review process for regulatory nodes includes: The node applicant submits a business license, industry qualification certificate (manufacturers need to provide a "Power Electronic Components Production License," and distributors need to provide an "Electronic Product Sales Qualification Certificate"), and a scanned copy of the legal representative's ID card to the regulatory node; the regulatory node verifies the authenticity of the materials through "online verification (connecting to the National Enterprise Credit Information Publicity System) + offline random inspection (random on-site inspection)," with a review period of ≤7 working days; after approval, the regulatory node generates an on-chain digital certificate for the applicant (containing the node's public key, identity information, and a validity period of 1 year), which the applicant uses to activate the node; if a node is found to have committed data fraud (such as falsifying inventory data), the regulatory node can suspend its digital certificate. During the suspension period, nodes cannot upload or query data and can only resume operations after rectification is completed. The replenishment rules for the inventory coordination contract are set as follows: when the distributor's inventory is less than the safety threshold (safety threshold = average daily sales over the past 30 days × 7) and the manufacturer's in-transit inventory is greater than or equal to the stock shortage, the contract automatically generates a replenishment request (including component model, quantity, and delivery address). The manufacturer node must confirm within 24 hours; if confirmation is not received within this timeframe, the supervisory node will intervene to coordinate. The node data upload requirements for the order traceability contract are: each node must upload data within one hour of completing the process, and the data must be accompanied by a node digital signature (to prevent forgery). For example, the manufacturer must upload and sign the factory data within one hour of the component leaving the factory. The rules for the after-sales responsibility contract are detailed as follows: transportation vibration > 50m / s. 2 Temperature > 40℃ / < 0℃ → Logistics provider's responsibility; Storage temperature > 40℃ or humidity > 60% → Distributor's responsibility; Failure to pass factory inspection → Manufacturer's responsibility; Exceeding rated current / temperature limits → Customer's responsibility. The zero-knowledge proof module uses the Groth16 algorithm to achieve non-interactive proof, with a verification time of ≤100ms. If a customer needs to prove "monthly purchase volume ≥ 100 units", they only need to generate a proof document, and the manufacturer will verify it through the document without needing to view the specific purchase details.

[0050] In an optional embodiment, the terminal security unit of the security layer includes an antivirus module and a malware protection module. The antivirus module is used to detect and remove viruses from the IoT gateway and terminal devices, and the malware protection module is used to intercept malicious attacks targeting the terminal devices. In addition to the zero-knowledge proof module and the quantum random number generation module, the on-chain security unit also includes an AES-256 encryption module. The quantum random number generation module is used to generate unpredictable encryption keys, and the AES-256 encryption module is used to encrypt private data transmitted on-chain. The access control unit adopts a role-based access control policy to assign different access permissions to each consortium blockchain node. Specifically, the manufacturer node can only access the inventory data of the distributor node and the purchase plan data of the customer node, but cannot access the privacy data of the customer node; the distributor node can only access the production plan data of the manufacturer node and the order data of the customer node, but cannot access the production process parameters of the manufacturer node.

[0051] It should be noted that, in this embodiment, the antivirus module of the terminal security unit refers to a functional component used to scan and kill viruses on IoT gateways and terminal devices; the malware protection module refers to a component used to intercept malicious attacks (such as Trojans and ransomware) targeting terminal devices; the AES-256 encryption module of the on-chain security unit refers to a component that uses the AES-256 algorithm to encrypt on-chain data. AES-256 is a symmetric encryption algorithm with a key length of 256 bits, offering high security; the quantum random number generation module refers to a component that generates encryption keys based on quantum physics principles, with the quantum no-cloning principle ensuring that the key cannot be predicted; the role-based access control policy of the permission management unit refers to rules that allocate different access permissions according to the roles of each entity (manufacturer, distributor, customer), ensuring that each role can only access data within its access permissions; terminal devices refer to hardware devices such as IoT gateways, sensors, and customer computers / mobile phones; privacy data refers to sensitive information of each entity (such as manufacturer production processes, distributor profits, and customer contact information). As a further extension of the security layer, the core is to build a full-chain security system through "endpoint protection + on-chain encryption + access control". Its purpose is to solve the security risks faced by the system, such as endpoint attacks (e.g., sensor intrusion), data leakage (e.g., order amount theft), and unauthorized access (e.g., distributors viewing manufacturer processes). The effects include resisting malicious endpoint attacks (anti-virus module detection rate ≥99.9%), protecting on-chain data security (AES-256 encryption cannot be cracked), and standardizing data access (access control prevents unauthorized access). The implementation approach is to match the corresponding protection components according to the security requirements of "endpoint-transmission-access": endpoint layer to prevent attacks, on-chain layer to ensure encryption, and access control layer to control access, forming a multi-dimensional protection.

[0052] In one possible implementation, the antivirus module of the terminal security unit uses Qi An Xin IoT antivirus software, which supports a full system scan at 2:00 AM daily (during low terminal load periods). The scan scope includes system files, applications, and stored data, with a virus detection rate of ≥99.9%. Upon detection, the virus is immediately isolated and an SMS alert is sent to the administrator. The malware protection module uses behavioral analysis technology to monitor terminal processes (e.g., whether there are abnormal data theft behaviors) and network connections (e.g., whether there are malicious IP connections) in real time, with an interception success rate of ≥99%. Unknown software is run in a sandbox (isolated environment testing) and allowed to execute only after security is confirmed. The AES-256 encryption module of the on-chain security unit is used to encrypt private data such as order amounts and customer contact information. The encryption key is generated by a quantum random number generation module (key length 256 bits, generation time ≤1ms), and the key is not stored physically. The medium is generated only temporarily during transmission and transmitted through the quantum key distribution channel, and destroyed immediately after transmission is completed. The quantum random number generation module is based on the quantum tunneling effect, and the randomness of the random numbers conforms to the NISTSP800-22 standard, preventing the key from being brute-forced. The role permissions of the access control unit are refined as follows: Manufacturer nodes can access distributor inventory data, total customer purchase plan, and logistics transportation data, but cannot access customer contact information, distributor profits, or other manufacturers' processes; Distributor nodes can access manufacturer production plans, customer order volume / delivery period, and logistics trajectory, but cannot access manufacturer processes / costs or customer production plan details; Customer nodes can access distributor inventory / prices, logistics trajectory, and manufacturer's publicly available production capacity, but cannot access manufacturer processes or other customer data of distributors; Permission changes must be reviewed by the supervisory node (e.g., if the manufacturer applies to access customer purchase details), and adjustments can only be made after the review is approved.

[0053] In this embodiment, the perception layer collects data from the entire process of power electronic components and generates a unique supply chain code. Combined with the blockchain collaboration layer, this enables trusted data sharing, achieving traceability throughout the entire lifecycle of components and preventing data silos across different entities. The computing layer's federated learning module enables multi-entity collaborative modeling, integrating multi-dimensional data while protecting data privacy, thus improving the accuracy of supply chain demand forecasting and assisting various entities in making scientific decisions. The communication layer uses segmented transmission and priority scheduling to distinguish between ordinary sales data and on-chain shared data, ensuring real-time transmission of core data and reducing the cost of conventional data transmission. The security layer employs terminal protection, on-chain encryption, and access control to resist malicious attacks and regulate data access, achieving system data security and controllability while preventing the leakage of privacy for each entity. Finally, the blockchain collaboration layer's smart contracts automatically execute inventory coordination and after-sales responsibility determination, reducing manual intervention and improving supply chain collaboration efficiency and quickly defining cross-entity responsibilities.

[0054] Based on the above embodiments, such as Figure 2As shown, the present invention also provides a sales management method for power electronic components based on the Internet of Things, including: Step 100: Data collection and on-chain synchronization. Collect full-process data of power electronic components in the stages of manufacturing, inventory, transportation and use, generate a unique industry chain code, and transmit the full-process data and industry chain code to the consortium blockchain, where the corresponding entity's nodes upload and share it on the chain. Step 100 is the foundational data support for the entire sales management methodology. Its core lies in acquiring end-to-end information on power electronic components from manufacturing to use through system sensing capabilities, and leveraging a consortium blockchain to achieve reliable data storage and sharing. During execution, no manual intervention is required for data collection and uploading. The sensing layer hardware automatically captures component production parameters, inventory environment, transportation status, and operating conditions, ensuring real-time and accurate data collection. Simultaneously, unique industry chain codes link and bind scattered data from different stages, preventing data fragmentation. On-chain synchronization further guarantees data immutability, allowing each node to obtain real-time component data, providing a unified and reliable data source for all subsequent management stages, thus addressing the problems of "data silos" and "data fraud" in traditional management from the source.

[0055] Step 200, Federated learning demand prediction: Receive local training data from each entity, generate model gradients through local model training, aggregate all model gradients to generate an industry chain-level demand prediction model, and distribute the optimization parameters of the model to each entity. Step 200 focuses on breaking down data privacy barriers to achieve collaborative demand forecasting across stakeholders. Unlike traditional centralized modeling that requires collecting raw data from each stakeholder, this step allows manufacturers, distributors, and customers to train their models locally, uploading only the model gradients (without raw data information) to the cloud for aggregation. This "data remains still, model moves" approach protects the business privacy of each stakeholder (e.g., manufacturer production processes, distributor customer details) while integrating multi-dimensional data features (e.g., manufacturer capacity, distributor inventory, customer procurement plans) to generate a predictive model that better meets the actual needs of the industry chain. Furthermore, once the model parameters are distributed to each stakeholder, they can guide them to adjust their production, inventory, or procurement strategies, avoiding "blind production" or "supply shortages" and improving the resource allocation efficiency of the entire industry chain.

[0056] Step 300: Order and inventory coordination. After receiving the customer's order request, query the inventory data on the consortium blockchain. If the distributor's inventory is lower than the preset threshold, automatically trigger a replenishment request and allocate inventory according to the manufacturer's in-transit inventory data, and update the inventory information of each entity on the chain. Step 300 is a core execution step connecting customer demand and supply chain resources. It aims to achieve dynamic inventory optimization and efficient order fulfillment through on-chain data linkage. When a customer initiates an order request, the system can directly query real-time inventory data on the consortium blockchain, eliminating the need for manual cross-entity communication and significantly shortening order response time. If the distributor's inventory is insufficient, the system can automatically trigger a replenishment request and retrieve the manufacturer's in-transit inventory data, flexibly realizing cross-entity inventory allocation models such as "direct supply from manufacturer to customer," reducing intermediate circulation links. Simultaneously, after order processing, the inventory data of all entities on the blockchain will be updated synchronously, ensuring real-time consistency of inventory information across the entire industry chain. This avoids duplicate stocking or inventory omissions due to information asynchrony, improving order fulfillment efficiency and inventory turnover rate.

[0057] Step 400: Defining after-sales responsibility. Receive fault data of power electronic components uploaded by customers, retrieve full-process data on the consortium blockchain, determine the responsible party for the fault according to preset rules, and generate and push a responsibility handling agreement. Step 400 efficiently resolves cross-entity after-sales disputes by combining automated rules with on-chain data. When a customer reports a component malfunction, there's no need for manual evidence collection at each stage. The system automatically retrieves full-process data (such as transportation environment, inventory status, and usage conditions) stored on the consortium blockchain and quickly matches the responsible party based on preset liability determination rules. The entire process eliminates the need for repeated evidence presentation or negotiation among the parties. The smart contract automatically generates a liability handling agreement and pushes it to the relevant parties, avoiding excessively long processing times caused by "responsibility shirking" and ensuring the fairness and transparency of liability determination. This reduces the impact of after-sales disputes on sales cooperation and enhances the trust of all parties in the system.

[0058] Step 500: After-sales data closed loop, optimize production process based on on-chain fault data, conduct risk investigation on power electronic components of the same batch, upload process optimization results and investigation results to consortium blockchain, and update public data repository; Step 500 is a crucial step in achieving continuous optimization of the supply chain, transforming after-sales fault data into a basis for operational improvements. Manufacturers can use on-chain fault data to pinpoint production process weaknesses (such as a batch of components failing due to packaging issues) and adjust production parameters accordingly. Distributors can conduct risk assessments on the same batch of components to proactively mitigate potential fault risks. Simultaneously, the process optimization results and assessment results are uploaded back to the consortium blockchain, forming a closed loop of "fault discovery - problem analysis - improvement implementation - data feedback." This closed-loop mechanism allows sales management to go beyond simply "problem handling," driving continuous iteration and upgrading across all links of the supply chain, thereby improving the product quality and sales management level of power electronic components.

[0059] Based on the above embodiments, such as Figure 3As shown, the present invention also provides an electronic device, the electronic device comprising: The processor 22 includes at least one processor 22, at least one memory 21, a communication interface 23, and a communication bus 24, wherein the processor 22 is communicatively connected to the memory 21. In this embodiment, the memory 21 can be implemented in any suitable manner, for example, the memory 21 can be a read-only memory, a hard disk drive, a solid-state drive, or a USB flash drive, etc.; the memory 21 is used to store at least one executable instruction executed by the processor; In this embodiment, the processor 22 can be implemented in any suitable manner. For example, the processor 22 can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) that can be executed by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers, etc.; the processor is used to execute the executable instructions to implement the IoT-based power electronic component sales management method as described above.

[0060] Based on the above embodiments, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described Internet of Things-based power electronic component sales management method.

[0061] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A power electronic component sales management system based on the Internet of Things, characterized in that, It includes a perception layer, a communication layer, a computing layer, a blockchain collaboration layer, and a security layer; The perception layer is used to collect data on the entire process of power electronic components and generate a unique supply chain code. The data on the entire process includes data generated by power electronic components during the manufacturing, inventory, transportation and use stages. The supply chain code is used to identify the unique identity and circulation node of the power electronic components. The communication layer is used to realize data transmission between the perception layer, the computing layer and the blockchain collaboration layer, including a general transmission channel for transmitting ordinary sales data, a dedicated transmission channel for transmitting on-chain shared data, and a scheduling unit for scheduling data transmission priorities. The computing layer integrates a federated learning module, which receives local training data from multiple entities, generates an industry chain-level demand prediction model through local training, gradient uploading, and gradient aggregation, and distributes the optimized model parameters to each entity. The computing layer also includes a database for storing data, which is divided into a private database for storing private data and a shared database for storing public data. The blockchain collaboration layer includes multiple consortium blockchain nodes and at least three types of smart contracts. The consortium blockchain nodes include at least manufacturer nodes, distributor nodes, customer nodes, and logistics provider nodes. Each consortium blockchain node is used to upload and share power electronic component-related data of the corresponding entity. The smart contracts include inventory collaboration contracts for inventory collaboration, order tracking contracts for order tracking, and after-sales responsibility contracts for after-sales responsibility determination. The security layer is used to ensure system data security and privacy protection, including a terminal security unit for terminal protection, an on-chain security unit for on-chain data encryption, and an access control unit for controlling access permissions. The on-chain security unit integrates a zero-knowledge proof module and a quantum random number generation module.

2. The IoT-based power electronic component sales management system according to claim 1, characterized in that, The perception layer includes an IoT chip, at least three types of smart sensors, and RFID tags; The IoT chip includes an MCU chip and a SoC chip. The MCU chip is used to perform local preprocessing of the sensing layer data, and the SoC chip integrates RFID read / write and positioning functions to read the RFID tag data and collect the location data of power electronic components. The intelligent sensor is used to collect environmental parameters and quantity data of power electronic components at different stages, and transmit the collected data to the communication layer. The RFID tag is used to store the basic parameters of the power electronic components and the supply chain code. The basic parameters include the model, production batch, production parameters and quality inspection data of the power electronic components. The supply chain code format includes at least the manufacturer identifier, batch identifier and individual item identifier.

3. The IoT-based power electronic component sales management system according to claim 2, characterized in that, The intelligent sensor includes a temperature and humidity sensor, a vibration sensor, and an infrared counting sensor; The temperature and humidity sensor is used to collect temperature and humidity data of power electronic components during the storage and transportation stages. The temperature and humidity data are used to determine whether the storage and transportation environment meets the storage requirements of power electronic components. The vibration sensor is used to collect vibration amplitude data of power electronic components during transportation, and the vibration amplitude data is used for subsequent after-sales liability determination. The infrared counting sensor is used to collect quantity data of power electronic components during the warehousing and outbound stages, and the quantity data is used to update the inventory data in the computing layer.

4. The IoT-based power electronic component sales management system according to claim 1, characterized in that, The general transmission channel of the communication layer includes a WiFi module and a Bluetooth module. The WiFi module adopts the WiFi 6 protocol, and the Bluetooth module adopts the Bluetooth 5.2 protocol. The general transmission channel is used to transmit the order amount, customer contact information, order delivery deadline, and order payment status of power electronic components. The dedicated transmission channel includes an NB-IoT module and a 5G industrial private network module. The dedicated transmission channel is used to transmit inventory status data, fault summary data, transportation trajectory data, inventory warning threshold, and order fulfillment progress of power electronic components. The scheduling unit is used to mark the fault data of power electronic components as high-priority data and transmit it to the blockchain collaboration layer through the dedicated transmission channel first, so as to ensure the real-time response of fault data.

5. The IoT-based power electronic component sales management system according to claim 1, characterized in that, The computing layer also includes edge gateways and a cloud platform; The edge gateway is equipped with an IoT operating system and integrates a local training unit of the federated learning module, which is used to receive local data from the corresponding subject and perform local model training to generate model gradients. The cloud platform includes a federated learning scheduling center and a digital twin collaboration module. The federated learning scheduling center is used to receive model gradients uploaded by each edge gateway, generate the industry chain-level demand prediction model through a gradient aggregation algorithm, and distribute the parameters of the industry chain-level demand prediction model to each edge gateway. The digital twin collaboration module is used to construct a digital twin mapping between power electronic components and the manufacturer's production line, distributor's warehouse, and customer's workshop, so as to realize cross-entity visual collaborative scheduling. The private library is used to store the privacy data of each entity, including the manufacturer's production process parameters, the distributor's customer details, and the customer's production plan data; the shared library is used to store publicly available on-chain data, including the inventory quantity, transportation trajectory, and fault summary of power electronic components.

6. The IoT-based power electronic component sales management system according to claim 1, characterized in that, The consortium blockchain nodes in the blockchain collaboration layer also include supervisory nodes, which are used to verify the identity and legitimacy of each consortium blockchain node and supervise the authenticity of the on-chain data; The inventory coordination contract is used to automatically send a replenishment request to the manufacturer node when the dealer's inventory is lower than a preset safety threshold, and to trigger direct supply scheduling from the manufacturer to the customer based on the manufacturer node's in-transit inventory data. The order traceability contract is used to record the entire process node data of power electronic components from the factory to the customer's receipt. The entire process node data includes data on the factory departure node, warehousing node, delivery node, transportation node, and receipt node. The after-sales liability contract is used to automatically retrieve the full-process data on the chain after receiving fault data uploaded by the customer node, determine the responsible party for the fault according to the preset liability determination rules, and generate a liability handling agreement. The zero-knowledge proof module is used to process the privacy data uploaded by each subject, meaning that each subject can use the statistical characteristics of the privacy data but cannot obtain the original content of the privacy data.

7. The IoT-based power electronic component sales management system according to claim 1, characterized in that, The terminal security unit of the security layer includes an antivirus module and a malware protection module. The antivirus module is used to detect and remove viruses from IoT gateways and terminal devices, and the malware protection module is used to intercept malicious attacks against terminal devices. In addition to the zero-knowledge proof module and the quantum random number generation module, the on-chain security unit also includes an AES-256 encryption module. The quantum random number generation module is used to generate unpredictable encryption keys, and the AES-256 encryption module is used to encrypt private data transmitted on-chain. The access control unit adopts a role-based access control policy to assign different access permissions to each consortium blockchain node. Specifically, the manufacturer node can only access the inventory data of the distributor node and the purchase plan data of the customer node, but cannot access the privacy data of the customer node; the distributor node can only access the production plan data of the manufacturer node and the order data of the customer node, but cannot access the production process parameters of the manufacturer node.

8. A sales management method for power electronic components based on the Internet of Things, characterized in that, Includes the following steps: Data collection is synchronized with the blockchain. Data on the entire process of power electronic components during the manufacturing, inventory, transportation and use stages is collected to generate a unique industry chain code. The entire process data and industry chain code are transmitted to the consortium blockchain, where they are uploaded and shared on the blockchain by the nodes of the corresponding entities. Federated learning demand prediction receives local training data from each entity, generates model gradients through local model training, aggregates all model gradients to generate an industry chain-level demand prediction model, and distributes the optimization parameters of the model to each entity. Orders and inventory are coordinated. After receiving a customer's order request, the inventory data on the consortium blockchain is queried. If the distributor's inventory is lower than the preset threshold, a replenishment request is automatically triggered and inventory is allocated according to the manufacturer's in-transit inventory data, and the inventory information of each entity on the chain is updated. After-sales responsibility determination: Receive power electronic component failure data uploaded by customers, retrieve full-process data on the consortium blockchain, determine the responsible party for the failure according to preset rules, and generate and push responsibility handling agreements; After-sales data closed loop: optimize production process based on on-chain fault data, conduct risk investigation on power electronic components of the same batch, and upload the process optimization results and investigation results to the consortium blockchain to update the public data repository.

9. An electronic device, characterized in that, The electronic device includes: The processor and the memory are communicatively connected. The memory is used to store at least one executable instruction executed by the processor, which executes the executable instruction to implement the Internet of Things-based power electronic component sales management method as described in claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the Internet of Things-based power electronic component sales management method as described in claim 8.

Images