CN-121981516-A - Airport grading security inspection method and system based on blockchain and federal learning

CN121981516ACN 121981516 ACN121981516 ACN 121981516ACN-121981516-A

Abstract

The invention belongs to the technical field of airport grading security inspection, and discloses an airport grading security inspection method and system based on blockchain and federal learning. The method comprises the steps of initiating a cross-link request, generating a passenger encryption feature, generating zk-SNARK certification, generating a passenger encryption behavior abstract, pushing flight status information in real time through an airport ACDM system, obtaining a passenger risk level in a federal learning layer by adjusting federal learning weight based on the generated and pushed information, and dynamically adjusting a security channel through real-time aviation security information. The invention introduces federal learning and combines multiparty data training risk assessment models on the premise of protecting privacy. And designing a dynamic weight mechanism, and adjusting the priority of the risk factors according to different scenes. Different from the airport security inspection of the same type, the invention provides different application scenes, improves the travel efficiency of passengers and can realize quick security inspection.

Inventors

WANG TIANFEI
LIU FENGLING
LI YUNLU
SUN YONG
XUE LINGXIANG

Assignees

青岛民航凯亚系统集成有限公司

Dates

Publication Date: 20260505
Application Date: 20251208

Claims (10)

1. The airport grading security check method based on blockchain and federal learning is characterized by obtaining multiparty data through cross-chain, calculating the risk level of passengers by using a federal learning model, and dynamically adjusting security check channel allocation strategy according to real-time flight information, and specifically comprises the following steps: s1, accessing airport flight real-time data in a data access layer, configuring air traffic, airports and user information, communicating with a data service layer through a TCP/IP network protocol, and initiating a cross-link request; S2, receiving a request result, generating a passenger encryption feature in a data service layer, generating zk-SNARK evidence, generating a passenger encryption behavior abstract, integrating the passenger encryption behavior abstract into an airport chain, acquiring the travel behavior of a passenger at the airport, encrypting and then using the passenger encryption behavior abstract as the input of a dynamic weight contract, pushing flight state information in real time through an airport ACDM system, establishing an interface based on a acdm system of each airport, acquiring the flight dynamic information of the airport in real time, adjusting federal learning parameters based on the flight state information and other chain data, and acquiring the risk level of the passenger in the federal learning layer by adjusting the weight of federal learning based on the generated and pushed information; and S3, dynamically adjusting a security inspection channel by the edge decision terminal according to the passenger risk level and real-time avionic security information, storing the adjustment result, and transmitting the stored data information to an application layer for display.
2. The airport hierarchical security method based on blockchain and federal learning of claim 1, wherein in step S2, the data service layer comprises a blockchain physical layer, a federal learning layer, a blockchain layer, a cross-chain gateway layer, and an intelligent contract layer; The block chain layer consists of airport chains and airline department chains and integrates an organic airport node and an airline department node respectively, the airline department node is integrated with different airline department information to provide flight numbers, flight dates, destinations and luggage information, and the airport node consists of all airports and adopts HYPERLEDGERFARIC alliance chains to provide passenger behaviors, equipment states and flight dynamic information.
3. The airport hierarchical security inspection method based on blockchain and federal learning according to claim 2, wherein in the cross-chain gateway layer, the cross-chain data acquisition process adopts a relay bridge mode, firstly, the airport chain transmits a data request to the relay, after receiving the request, the relay transmits an encryption request and zk-SNARK request to the airline chain, and after receiving homomorphic encryption feature vectors and proof results of the airline chain, the relay transmits the aggregated data back to the airport chain.
4. The airport hierarchical security method based on blockchain and federal learning of claim 2, wherein the intelligent contract layer comprises a flight status classification contract, a dynamic weight adjustment contract, a certification deposit contract and a timeout callback contract of the home airport chain; The ProofVerifier.sol contract is used for verifying the passenger-grade ciphertext information and generating zk-SNARK certification; The flightadapter. Sol contract of the home airline chain is used to generate the passenger encryption feature.
5. The airport hierarchical security method based on blockchain and federal learning of claim 4, wherein the flight status classification contract uses a dynamic policy adjustment mechanism to access airport ACDM system flight information in real time and return the flight status information, comprising: An interface is established with an airport ACDM system, the flight status is divided into normal, delayed, cancelled and urgent, the flight information of the day is obtained according to the flight number, dividing the flight state into a check-in start state, a check-out stop state, a local boarding state, a check-in prompting state, a check-in stop state and a take-off state, and then comparing the planned take-off time with the current time according to the flight, wherein the planned take-off time is subtracted from the current time; finally, the flight status information is converted into normal, delayed, cancelled and urgent flight status information, and the converted flight status information is used as an input parameter for dynamically adjusting the federal learning weight in the federal learning layer; the dynamic weight adjustment contract adjusts federal learning weights according to the flight status classification contract, including dynamically adjusting federal learning weights of airlines and airports according to the flight status output by the flight status classification contract; The method comprises the steps that a certification contract performs certification on a passenger risk level, a channel allocation result and a federal learning model version number, wherein certification comprises the steps of setting a certification function and storing data; The timeout callback contract is used for performing cross-link timeout processing, fault tolerance and performance optimization, and comprises the steps of initiating a cross-link request, setting a timeout time limit, and enabling a degradation strategy and using only local airport data if the request fails.
6. The airport hierarchical security method based on blockchain and federal learning of claim 4, wherein the airline chain generates passenger encryption features, and the airline nodes are utilized to extract flight feature vectors and homomorphic encrypt the flight feature vectors through Raft consensus mechanism to generate passenger encryption features; The flight feature vector is extracted through Raft a consensus mechanism, the flight feature vector is extracted through the cooperative work of a plurality of aviation nodes through Raft a consensus mechanism, and Raft is used for guaranteeing the consistency of feature extraction in a distributed environment.
7. The blockchain and federal learning-based airport hierarchical security method of claim 4, wherein generating zk-SNARK proofs comprises: The risk level contract uses zero knowledge to prove that the zk-SNARK circuit carries out privacy and security design, is integrated to a node, and provides judicial trusted verification for airport grading security check, and the method comprises the steps of firstly inputting a secret value by a system, then generating zk-SNARK evidence, checking by a verifier, finally FISCO BCOS alliance on-chain verification results, and returning validity and risk level of the verification results.
8. The blockchain and federal learning based airport hierarchical security check method of claim 2, wherein in step S2, passenger risk levels are obtained in the federal learning layer by adjusting federal learning weights, comprising: the federal learning layer interacts with the intelligent contract layer to extract local characteristic values of the edge nodes, integrate cross-chain data of the federal learning server and calculate risk levels of edge decision terminals; Inputting original data, including flight numbers, destinations, high risk areas and number of change labels, extracting a navigation node to convert the data into feature vectors, and generating zk-SNARK proving information by the node; Federal learning server cross-chain data integration includes: Federal learning dynamic weight adjustment algorithm: ; In the formula, Is the first The data sources are at the current moment Is a fusion weight of (2); To dynamically adjust the coefficients, the balance of the new confidence level and the historical weight is controlled, Is the first The data sources are at the current moment Is used to determine the confidence score of the (c) for the (c), Is the first The last time of the data sources Is used for the weight of the (c), The sum of confidence scores for all data sources at the current time, Is the total number of data sources; Aggregation algorithm: ; In the formula, For the parameters updated for the global model, Is the first The local model parameters trained by the individual edge nodes, Is the first The amount of data samples used by the edge nodes for this training, The sum of the total amount of data samples used for all edge nodes, ; For the flight status influencing factor, Model parameter increment calculated based on real-time flight status information; The final risk assessment algorithm includes: Risk algorithm: ; In the formula, For the final risk score of the passenger, The function is activated for Sigmoid, Respectively fusing weights of flight and behavior data, For the flight risk feature value extracted from the flight data, For the recorded risk value derived from the data, Is a behavioral anomaly characteristic value extracted from airport IoT data.
9. The airport hierarchical security method based on blockchain and federal learning of claim 2, wherein in step S3, the edge decision terminal dynamically adjusts the security channels according to the real-time avionic information, comprising: federal learning adds differential privacy noise at gradient update: ; In the formula, Is the first The local model parameters of the individual edge nodes, For the global sensitivity to be a function of the global sensitivity, To control the intensity of the added noise; Wherein the standard deviation of the noise Calculated from the following formula: ; In the formula, In order to be a privacy loss parameter, To allow a probability of failure of the privacy preserving mechanism; noise is dynamically adjusted with flight status-when delayed, In the normal course of this, the control signal, ; Risk class value Dynamically distributing to a standard security inspection channel.
10. An airport hierarchical security system based on blockchain and federal learning, the system implementing the blockchain and federal learning based airport hierarchical security method of any of claims 1-9, the system comprising: The data access layer is used for accessing airport flight real-time data in the data access layer, configuring air traffic, airports and user information, communicating with the data service layer through a TCP/IP network protocol and initiating a cross-link request; The system comprises a data service layer, an airport ACDM system, an interface, a federal learning layer, a data link layer, a link layer and a link layer, wherein the data service layer is used for receiving a request result, generating a passenger encryption characteristic in the data service layer, generating zk-SNARK evidence, generating a passenger encryption behavior abstract, integrating the passenger encryption behavior abstract into an airport chain, acquiring the traveling behavior of a passenger in the airport, encrypting and then using the passenger encryption behavior abstract as the input of a dynamic weight contract, pushing the flight state information in real time through the airport ACDM system, establishing an interface based on the acdm system of each airport, acquiring the airport flight dynamic information in real time, adjusting federal learning parameters based on the flight state information and other chain data, and acquiring the risk level of the passenger by adjusting the federal learning weight in the federal learning layer based on the generated and pushed information; The application layer is used for dynamically adjusting the security inspection channel by the edge decision terminal according to the passenger risk level and the real-time avionic security information, storing the adjustment result, and transmitting the stored data information to the application layer for display.

Description

Airport grading security inspection method and system based on blockchain and federal learning Technical Field The invention belongs to the technical field of airport grading security inspection, and particularly relates to an airport grading security inspection method and system based on blockchain and federal learning. Background In recent years, the throughput of airport passengers is gradually increased, the requirements of the passengers on the security inspection efficiency of the airports are higher, and the airports in each place are successively provided with classified security inspection, but the current classified security inspection of the airport passengers has the following problems: and the problem of data island is that passenger information is scattered in different institutions such as airports, avionics, side inspection, customs and the like, and is difficult to share in real time. And privacy disclosure risk that sensitive data (such as identity information and travel records) are stored in a centralized way and are easy to attack or misuse. Dynamic assessment is inadequate-traditional static rules cannot be combined with real-time data (e.g., temporary flight changes) to adjust security policies. With the development of the blockchain technology, the blockchain technology is gradually applied to various industries. The following characteristics of the blockchain can pertinently solve the above-described problems: Decentralizing trust, namely multi-party data chaining, breaking information islands and ensuring data consistency. And the security check record and the passenger behavior data are completely traceable, so that the manual tampering is avoided. Privacy protection-the "minimizing information disclosure" is achieved by encryption algorithms. Through the analysis, the problems and defects of the prior art are that island problems exist in the prior art data, privacy leakage risks are high, and dynamic evaluation accuracy is poor. Disclosure of Invention In order to overcome the problems in the related art, the disclosed embodiments of the invention provide an airport grading security inspection method and system based on blockchain and federal learning. The technical scheme is that the airport grading security inspection method based on blockchain and federal learning comprises the steps of obtaining multiparty data through a cross-chain, calculating the risk level of passengers by using a federal learning model, and dynamically adjusting security inspection channel allocation strategies according to real-time flight information, wherein the airport grading security inspection method specifically comprises the following steps: s1, accessing airport flight real-time data in a data access layer, configuring air traffic, airports and user information, communicating with a data service layer through a TCP/IP network protocol, and initiating a cross-link request; S2, receiving a request result, generating a passenger encryption feature in a data service layer, generating zk-SNARK evidence, generating a passenger encryption behavior abstract, integrating the passenger encryption behavior abstract into an airport chain, acquiring the travel behavior of a passenger at the airport, encrypting and then using the passenger encryption behavior abstract as the input of a dynamic weight contract, pushing flight state information in real time through an airport ACDM system, establishing an interface based on a acdm system of each airport, acquiring the flight dynamic information of the airport in real time, adjusting federal learning parameters based on the flight state information and other chain data, and acquiring the risk level of the passenger in the federal learning layer by adjusting the weight of federal learning based on the generated and pushed information; and S3, dynamically adjusting a security inspection channel by the edge decision terminal according to the passenger risk level and real-time avionic security information, storing the adjustment result, and transmitting the stored data information to an application layer for display. In the step S2, the data service layer comprises a blockchain physical layer, a federal learning layer, a blockchain layer, a cross-chain gateway layer and an intelligent contract layer; The block chain layer is composed of airport chains, airline department chains and integrated with organic airport nodes and airline department nodes respectively, the airline department nodes are integrated with different airline department information to provide flight numbers, flight dates, destinations and luggage information, the airport nodes are composed of airports and adopt HYPERLEDGERFARIC alliance chains to provide passenger behaviors, equipment states and flight dynamic information. In the inter-chain gateway layer, the inter-chain data acquisition process adopts a relay bridge mode, firstly, an airport chain transmits a data request to a relay, the relay t