Millimeter wave security inspection instrument health management method and system

Abstract

The invention discloses a millimeter wave security check instrument health management method and system, and the method comprises the steps: obtaining the monitoring data of the working state and working performance of each subsystem of a millimeter wave security check instrument, and forming a real-time operation state message; according to the real-time operation state message, based on the attention mechanism, fusing the working state and working performance information characteristics, constructing a millimeter wave security check instrument digital model, and based on a reinforcement learning algorithm, updating the millimeter wave security check instrument digital model parameters in real time; and completing the fault detection of the millimeter wave security check instrument according to the residual error of the predicted value and the measured value output by the millimeter wave security check instrument digital model. Based on the working state of the security check instrument, the working performance of the security check instrument is monitored, the working state and working information are subjected to feature fusion through an attention mechanism and mapped into a digital twin model in real time, model parameters are updated in real time based on improved reinforcement learning, and upgrading from passive maintenance to active health management is achieved. And a high-reliability and low-cost operation and maintenance guarantee is provided for a security check task.

Description

Technical Field

[0001] The present invention relates to the technical field of millimeter wave security inspection instruments, and in particular to a health management method and system for millimeter wave security inspection instruments. Background Art

[0002] To improve the efficiency of security inspection channels, millimeter-wave security scanners are evolving from linear arrays to area arrays. As the number of millimeter-wave security scanner channels increases, system complexity and integration levels significantly increase. Existing millimeter-wave security scanner health management methods have significant shortcomings in terms of data collection during security scanner operation, specifically:

[0003] Conventional millimeter-wave security scanners only monitor the overall operating status of the device, including temperature and humidity, and lack dynamic monitoring of performance indicators. Traditional health management models, based on single threshold criteria and scheduled maintenance, only monitor the operating status of individual subsystems, lacking dynamic monitoring of performance indicators. Suboptimal health in multiple subsystems can lead to degradation of system indicators, resulting in increased missed detection and false positive rates, making it difficult to ensure reliable security inspections.

[0004] 2) Conventional millimeter-wave security inspection instrument digital twin model parameters rely on manual experience calibration and expert experience, lack of adaptive online update model parameter mechanism, and have disadvantages such as subjectivity, poor real-time performance, and high limitations, which affect the dynamic adaptability and accuracy of the digital twin model and make it difficult to meet the high reliability requirements of security inspection tasks.

[0005] In the prior art, the invention patent application with publication number CN118828145A discloses a video summary generation method and related apparatus. After obtaining a multimodal video representation of each video clip in the target video, the process of extracting the optimal video clip combination from multiple video clips in the target video is modeled as a Markov decision process, taking into account the local importance and global coherence of the video clips. A reinforcement learning algorithm is used to solve the optimal strategy of the Markov decision process, extracting a more concise and coherent optimal video clip combination composed of important video clips, thereby generating an accurate video summary based on the optimal video clip combination and improving the quality of the video summary. In addition, a multi-head attention mechanism is applied to perform multimodal splicing and fusion on each video clip. However, in this patent, the reinforcement learning algorithm is used to solve the optimal strategy of the Markov decision process in order to obtain the optimal video clip combination for the multimodal video. In addition, the multi-head attention mechanism in this patent only fuses multi-source data with the same attributes but different modal representations. Summary of the Invention

[0006] The technical problem to be solved by the present invention is that conventional millimeter wave security inspection devices only monitor the working status of the entire device, and lack dynamic monitoring of working performance indicators and coordinated monitoring of working status and working performance indicators.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] A millimeter wave security inspection device health management method, comprising:

[0009] Obtain monitoring data on the working status and performance of each subsystem of the millimeter wave security inspection instrument and generate real-time operating status messages;

[0010] Based on real-time operating status messages, the system integrates working status and performance information features using an attention mechanism to build a digital model of the millimeter-wave security scanner. The model parameters are then updated in real time using a reinforcement learning algorithm.

[0011] The fault detection of the millimeter wave security scanner is completed based on the residual of the predicted value and the measured value output by the digital model of the millimeter wave security scanner.

[0012] Technical Effect: Based on the application scenarios of millimeter-wave security inspection devices in airports, customs, and other places with long-term and high-reliability requirements, this invention eliminates the limitations of single sensor data through dynamic fusion detection of working status and working performance, improves fault detection accuracy, and can accurately locate complex faults. Secondly, by utilizing dynamic fusion of multi-dimensional data, the nonlinear dynamic correlation between working status and working performance can be obtained in real time. This invention can significantly improve fault detection accuracy and greatly reduce the misjudgment problem caused by data fragmentation in traditional methods.

[0013] In this embodiment, forming a real-time running status message includes:

[0014] Working status monitoring dataX i,s (t) includes at least temperature information X i,s1 , humidity information X i,s2 、Salt spray informationX i,s3 , vibration information X i,s4 ; Where t is time, i is the i-th millimeter wave security inspection instrument subsystem;

[0015] Work performance monitoring dataX i,p (t) at least includes channel amplitude and phase consistency parameter X i,p1 , inter-channel amplitude and phase consistency parameter X i,p2 , noise figure parameter X i,p3 , local oscillator phase noise parameter X i,p4 , Simulated target self-test parameter X i,p5 ;

[0016] The real-time running status message is [X i,s1 ,X i,s2 ,X i,s3 ,X i,s3 ,X i,s4;X i,p1 ,X i,p2 ,X i,p3 ,X i,p4 ,X i,p5 ] T , where T is the transpose.

[0017] In this embodiment, a digital model of a millimeter wave security inspection instrument is constructed, including:

[0018] Extract working state features based on fully connected neural network and performance characteristics

[0019] Based on the attention mechanism, the working status characteristics of each subsystem of the millimeter wave security inspection instrument are and performance characteristics Perform feature fusion and obtain the fused feature vector h f ;

[0020] The fused feature vector h f Map to the output space to obtain the digital model of the millimeter-wave security inspection instrument.

[0021] In this embodiment, the fusion feature vector h is obtained. f ,include:

[0022]

[0023]

[0024]

[0025] Where, is the attention weight of working state characteristics and working performance characteristics, n is the subsystem of millimeter wave security inspection instrument, is the attention score of work status features and work performance features, v is the attention vector, W a is the attention weight matrix, b a is the attention bias vector, tanh(·) is the hyperbolic tangent function, and exp is the exponential function.

[0026] Technical effect: By integrating the working status and working performance characteristics, the digital model of the millimeter-wave security inspection instrument can not only reflect the health status of each device from the real-time performance indicators, but also estimate the performance indicator degradation from the health data of each device, effectively ensuring the reliable execution of security inspection tasks.

[0027] Real-time updating of the weight coefficients for the fusion of operating status and performance data enhances the millimeter-wave security scanner's digital model's core capabilities, including dynamic adaptability, robust fault tolerance, and high sensitivity. This mechanism not only improves the accuracy of the fusion of the operating status and performance data of each device in the scanner, but also enhances its practicality and reliability in complex security inspection scenarios.

[0028] In this embodiment, the digital model of the millimeter wave security inspection instrument is expressed as:

[0029]

[0030] Where y(t) is the measured value of the digital model of the millimeter wave security inspection instrument, is the predicted value of the millimeter wave security inspection instrument digital model, Activation Function (·) is the activation function, W m 、b m are the weight matrix and bias vector corresponding to the digital model of millimeter wave security inspection instrument.

[0031] In this embodiment, the parameters W of the millimeter wave security inspection instrument digital model are dynamically updated based on reinforcement learning. m 、b m ;

[0032] Let the parameters of the millimeter wave security inspection instrument digital model be in the form of θ, so that θ=[W m ; b m ] T ;

[0033] Define state s t , represents the comprehensive state of the millimeter wave security inspection instrument at time t;

[0034] Define the action as a t , represents the update amount Δθ of the millimeter wave security inspection instrument digital model t ;

[0035] Define reward r t , represents the error between the measured and predicted values of the digital model of the millimeter wave security detector;

[0036] Define strategy π θ (a t |s t ), indicating that in state s t Next, select action a t The probability distribution of

[0037] Construct the objective function J(θ) to represent the strategy π θ (a t |s t ) under the expected cumulative discounted reward;

[0038] Combined reward r t , discount factor Υ t and state s t , construct the time difference error δ t ;

[0039] According to the timing difference error δ t , calculate the policy gradient Δ of the objective function θ J(θ);

[0040] For the policy gradient Δ θ J(θ), use the gradient ascent method to update θ until the strategy converges, and obtain the updated parameters of the millimeter wave security inspection instrument digital model.

[0041] In this embodiment, the timing differential error δ t , expressed by the following formula:

[0042] δ t =r t +Υ t V π (s t+1 )-V π (s t );

[0043] Where V π (·) represents the value function.

[0044] In this embodiment, the policy gradient Δ of the objective function θ J(θ) is obtained by the following formula:

[0045]

[0046] Where, is the expectation, Δ θ logπ θ (a t |s t ) is the logarithmic gradient of the policy, and T is the total time.

[0047] In this embodiment, the policy convergence criteria are:

[0048] ‖θ k+1 -θ k ‖2≤∈;

[0049] Where θ k+1 represents the parameters of the millimeter-wave security inspection instrument digital model at the k+1th iteration; ∈ represents the convergence threshold.

[0050] Technical effect: The digital model of the millimeter-wave security inspection instrument realizes dynamic correction and prediction of model parameters through an improved reinforcement learning algorithm. Compared with the traditional reinforcement learning algorithm, the optimization strategy accurately adopts the 2-norm convergence criterion of model parameters. The 2-norm convergence criterion can significantly improve the stability, convergence speed, generalization ability and reliability of the model, and significantly improve the optimization stability and adaptive adjustment ability of model parameters.

[0051] The present invention also provides a millimeter wave security inspection instrument health management system, which applies the above-mentioned millimeter wave security inspection instrument health management method, including:

[0052] Dynamic monitoring module, used to obtain monitoring data on the working status and performance of each subsystem of the millimeter wave security inspection instrument and generate real-time operating status messages;

[0053] The digital twin module is used to build a digital model of the millimeter-wave security scanner based on real-time operating status messages and the integration of working status and working performance information features using an attention mechanism. It also updates the parameters of the millimeter-wave security scanner digital model in real time based on a reinforcement learning algorithm.

[0054] The health management module completes the fault detection of the millimeter wave security scanner based on the residual of the predicted value and the measured value output by the digital model of the millimeter wave security scanner.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] The present invention is not only based on the working status of the security scanner, but also monitors the working performance of the security scanner. It performs feature fusion of the working status and working information through the attention mechanism, maps them into the digital twin model in real time, and updates the model parameters in real time based on improved reinforcement learning, realizing the upgrade from passive maintenance to active health management, and providing highly reliable and low-cost operation and maintenance guarantees for security inspection tasks.

[0057] This invention is applied to the field of millimeter-wave security scanner management technology. Based on an attention mechanism, this invention fuses working status and performance data, enabling the millimeter-wave security scanner digital model to capture the nonlinear dynamic relationship between working status and performance in real time, overcoming the pain point of traditional methods that often lead to misjudgments due to data fragmentation.

[0058] The reinforcement learning algorithm of the present invention is applied to the real-time update of the digital model parameters of the millimeter-wave security inspection instrument. It clearly defines important modules in reinforcement learning, such as state, reward, and convergence strategy. Based on the reinforcement learning model, the convergence criterion is optimized. The 2-norm convergence criterion of the model parameters can significantly improve the stability and reliability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a flow chart of a millimeter wave security inspection device health management method according to an embodiment of the present invention.

[0060] Figure 2 This is a block diagram of a millimeter wave security inspection device health management system according to an embodiment of the present invention.

[0061] Figure 3 This is a block diagram of a dynamic monitoring module according to an embodiment of the present invention.

[0062] Figure 4 This is a block diagram of the digital twin module of an embodiment of the present invention.

[0063] Figure 5 This is a block diagram of the health management module of an embodiment of the present invention. DETAILED DESCRIPTION

[0064] To facilitate those skilled in the art to understand the technical solution of the present invention, the technical solution of the present invention is further described with reference to the accompanying drawings.

[0065] The terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, "plurality" means two or more, unless otherwise specifically defined.

[0066] See also Figure 1 and Figure 2 As shown, this embodiment provides a millimeter wave security inspection instrument health management method, including:

[0067] S10, obtaining monitoring data of the working status and working performance of each subsystem of the millimeter wave security inspection device, and generating a real-time operating status message.

[0068] See also Figures 1 to 3 As shown, in this embodiment, the millimeter-wave security inspection instrument's subsystems include a millimeter-wave transceiver subarray, a frequency synthesis and synchronization module, a millimeter-wave imaging module, and a hazardous materials detection module. The millimeter-wave transceiver subarray primarily utilizes a millimeter-wave radar array, the frequency synthesis and synchronization module primarily synchronizes and synchronizes the millimeter-wave signals from the millimeter-wave radar array, the millimeter-wave imaging module constructs object contours based on the frequency synthesis and synchronization of the millimeter-wave signals by analyzing the reflected signal's delay, intensity, and Doppler effect, and the hazardous materials detection module primarily performs hazardous materials detection.

[0069] In this embodiment, the working status monitoring data X i,s (t) includes at least temperature information X i,s1 , humidity information X i,s2 、Salt spray informationX i,s3 , vibration information X i,s4 ;

[0070] Working status monitoring dataX i,s (t) is: [X i,s1 ,X i,s2 ,X i,s3 ,X i,s3 ,X i,s4 ,...] T ;

[0071] Where t is time, i is the i-th millimeter wave security inspection instrument subsystem, and T is the transpose.

[0072] In this embodiment, the work performance monitoring data X i,p (t) at least includes channel amplitude and phase consistency parameter X i,p1 , inter-channel amplitude and phase consistency parameter X i,p2 , noise figure parameter X i,p3 , local oscillator phase noise parameter X i,p4 , Simulated target self-test parameter X i,p5 .

[0073] Then the work performance monitoring data X i,p (t) is: [,X i,p2 ,X i,p3 ,X i,p4 ,X i,p5 ,...] T ;

[0074] The real-time running status message is [X i,s1 ,X i,s2 ,X i,s3 ,X i,s3 ,X i,s4 ;X i,p1 ,X i,p2 ,X i,p3 ,X i,p4 ,X i,p5 ] T .

[0075] S20, based on the real-time operation status message, the working status and working performance information features are integrated based on the attention mechanism to build a digital model of the millimeter wave security scanner, and the parameters of the digital model of the millimeter wave security scanner are updated in real time based on the reinforcement learning algorithm.

[0076] See also Figures 1 to 4 As shown in this embodiment, before constructing the digital model of the millimeter wave security inspection instrument, data preprocessing is first performed. The real-time operation status message is denoised and normalized:

[0077]

[0078] Where, is X′ s (t) represents the working status monitoring data after preprocessing; X′p (t) represents the working performance monitoring data after preprocessing; μ s 、μ p : represents the working state monitoring data and the mean of the working state monitoring data respectively; σ s , σ p Represent the working status monitoring data and the standard deviation of the working status monitoring data respectively.

[0079] In this embodiment, after the preprocessing is completed, multi-dimensional data feature fusion is performed. Specifically, the working status and working performance information features are fused based on the attention mechanism, and the working status and working performance monitoring data of each subsystem of the millimeter wave security scanner are effectively combined through dynamic weight distribution to generate a unified feature representation, and construct a digital model of the millimeter wave security scanner.

[0080] See also Figures 1 to 4 As shown, in this embodiment, a digital model of a millimeter wave security inspection instrument is constructed, including:

[0081] S21, extract working status features based on fully connected neural network and performance characteristics

[0082] In this embodiment,

[0083] Where, Represents the working status data of the i-th millimeter wave security inspection instrument subsystem; The weight matrix representing the working status monitoring data of the i-th millimeter-wave security inspection instrument subsystem; represents the bias vector of the monitoring data of the working status of the i-th millimeter-wave security inspection instrument subsystem; ReLU(·) represents the activation function. There is no doubt that, is the preprocessed data.

[0084]

[0085] Where, Represents the working performance monitoring data of the i-th millimeter wave security inspection instrument subsystem; The weight matrix representing the performance monitoring data of the i-th millimeter-wave security inspection instrument subsystem; Represents the bias vector of the performance monitoring data of the i-th millimeter-wave security inspection instrument subsystem.

[0086] S22, based on the attention mechanism, the working status characteristics of each subsystem of the millimeter wave security inspection instrument are and performance characteristics Perform feature fusion and obtain the fused feature vector h f .

[0087] In this embodiment,

[0088] Where, is the attention weight of working state characteristics and working performance characteristics, and n is the subsystem of the millimeter wave security inspection instrument.

[0089]

[0090]

[0091] is the attention score of work status feature and work performance feature, v is the attention vector, which is used to map the features of attention space to scalar attention score, W a is the attention weight matrix, which is used to map the feature vector to the attention space, b a is the attention bias vector, tanh(·) is the hyperbolic tangent function, and exp is the exponential function.

[0092] S23, the fusion feature vector h f Map to the output space to obtain the digital model of the millimeter-wave security inspection instrument.

[0093] In this embodiment, the output parameter y(t) of the millimeter wave security inspection instrument digital model is combined with the input fusion feature vector h f The equation is:

[0094]

[0095] Where y(t) is the measured value of the digital model of the millimeter wave security inspection instrument, is the predicted value of the millimeter wave security inspection instrument digital model, Activation Function (·) is the activation function, W m 、b m are the weight matrix and bias vector corresponding to the digital model of millimeter wave security inspection instrument.

[0096] In this embodiment, the parameters of the millimeter wave security detector digital model are updated in real time based on the reinforcement learning algorithm. Specifically, the parameters W of the millimeter wave security detector digital model are dynamically updated based on the reinforcement learning algorithm. m 、b m ,include:

[0097] S241, let the parameters of the millimeter wave security inspection instrument digital model be in the form of θ, so that θ=[W m ; b m ] T .

[0098] S242, define state s t , which represents the comprehensive state of the millimeter wave security inspection instrument at time t.

[0099] In this embodiment, s t =h f (t).

[0100] S243, define the action as a t , represents the update amount Δθ of the millimeter wave security inspection instrument digital model t , that is, the direction and amplitude of parameter adjustment.

[0101] In this embodiment, a t =Δθ t .

[0102] S244, define reward r t , which represents the error between the measured and predicted values of the digital model of the millimeter-wave security scanner.

[0103] In this embodiment,

[0104] S245, define strategy π θ (a t |s t ), indicating that in state s t Next, select action a t The probability distribution of .

[0105] S246, construct the objective function J(θ), representing the strategy π θ (a t |s t ) under the expected cumulative discounted reward.

[0106] In this embodiment, the objective function J(θ) is expressed as:

[0107]

[0108] Where, is the expectation, T is the total time, Υ t is a discount factor, representing the importance of current rewards and future rewards. The objective function J(θ) is the measure of the strategy π θ long-term performance.

[0109] S247, combined with reward r t , discount factor Υ t and state s t , construct the time difference error δ t .

[0110] In this embodiment, δ t =r t +Υ t V π (s t+1 )-V π (st );where V π (·) is the value function, which represents the expected accumulated discounted reward.

[0111] S248, according to the timing difference error δ t , calculate the policy gradient Δ of the objective function θ J(θ).

[0112] In this embodiment,

[0113] Where, Δ θ logπ θ (a t |s t ) is the logarithmic gradient of the strategy, representing the parameter θ for the selected action a t the impact of;

[0114] S249, for the policy gradient Δ θ J(θ), use the gradient ascent method to update θ until the strategy converges, and obtain the updated parameters of the millimeter wave security inspection instrument digital model.

[0115] In this embodiment, the gradient ascent method is used to update θ as follows:

[0116]

[0117] Where, is the updated parameter θ, and α is the learning rate, which is used to control the step size of the updated parameter θ.

[0118] In this embodiment, the 2-norm convergence criterion is used to perform strategy convergence. Specifically, the strategy convergence criterion is:

[0119] ‖θ k+1 -θ k ‖2≤∈;

[0120] Where θ k+1 represents the parameters of the millimeter-wave security inspection instrument digital model at the k+1th iteration; ∈ represents the convergence threshold.

[0121] S30, completing fault detection of the millimeter wave security inspection instrument based on the residual between the predicted value and the measured value output by the digital model of the millimeter wave security inspection instrument.

[0122] See also Figures 1 to 5 As shown in this embodiment, the predicted values of the output parameters of the millimeter wave security inspection instrument digital model are The residual error from the measured value y(t) completes fault detection, or health assessment. If the residual exceeds a threshold, a fault is determined. After a fault is determined, a deep learning model is used to identify the health status of the security scanner, including lifespan predictions for each subsystem component and risk predictions for each performance indicator. In this embodiment, the deep learning approach to fault diagnosis is not limited. Furthermore, based on the results of fault diagnosis and fault pattern recognition, a maintenance strategy can be generated according to the corresponding plan.

[0123] See also Figures 1 to 5 As shown, the present invention also provides a millimeter wave security inspection instrument health management system, which applies the above-mentioned millimeter wave security inspection instrument health management method, including:

[0124] The dynamic monitoring module is used to obtain monitoring data on the working status and performance of each subsystem of the millimeter wave security inspection instrument and generate real-time operating status messages.

[0125] The digital twin module is used to build a digital model of the millimeter-wave security scanner based on real-time operating status messages and the integration of working status and working performance information features based on the attention mechanism, and to update the parameters of the digital model of the millimeter-wave security scanner in real time based on the reinforcement learning algorithm.

[0126] The health management module completes the fault detection of the millimeter wave security scanner based on the residual of the predicted value and the measured value output by the digital model of the millimeter wave security scanner.

[0127] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims rather than the foregoing description. It is intended that all variations within the meaning and range of equivalents of the claims be embraced herein, and any reference signs in the claims should not be construed as limiting the claims to which they relate.

[0128] The above-mentioned embodiments merely represent the implementation methods of the invention. The protection scope of the present invention is not limited to the above-mentioned embodiments. For those skilled in the art, several variations and improvements can be made without departing from the concept of the present invention, which all fall within the protection scope of the present invention.

Claims

1. A millimeter wave security inspection instrument health management method, characterized in that: include: Obtain monitoring data on the working status and performance of each subsystem of the millimeter wave security inspection instrument and generate real-time operating status messages; Based on real-time operating status messages, the system integrates working status and performance information features using an attention mechanism to build a digital model of the millimeter-wave security scanner. The model parameters are then updated in real time using a reinforcement learning algorithm. The fault detection of the millimeter wave security scanner is completed based on the residual of the predicted value and the measured value output by the digital model of the millimeter wave security scanner.

2. The millimeter wave security inspection instrument health management method according to claim 1 is characterized in that: Generate real-time operation status message, including: Working status monitoring dataX i,s (t) includes at least temperature information X i,s1 , humidity information X i,s2 、Salt spray informationX i,s3 , vibration information X i,s4 ; Where t is time, i is the i-th millimeter wave security inspection instrument subsystem; Work performance monitoring dataX i,p (t) at least includes channel amplitude and phase consistency parameter X i,p1 , inter-channel amplitude and phase consistency parameter X i,p2 , noise figure parameter X i,p3 , local oscillator phase noise parameter X i,p4 , Simulated target self-test parameter X i,p5 ; The real-time running status message is [X i,s1 ,X i,s2 ,X i,s3 ,X i,s3 ,X i,s4 ;X i,p1 ,X i,p2 ,X i,p3 ,X i,p4 ,X i,p5 ] T , where T is the transpose.

3. The millimeter wave security inspection instrument health management method according to claim 1, characterized in that: Build a digital model of the millimeter-wave security scanner, including: Extract working state features based on fully connected neural network and performance characteristics Based on the attention mechanism, the working status characteristics of each subsystem of the millimeter wave security inspection instrument are and performance characteristics Perform feature fusion and obtain the fused feature vector h f ; The fused feature vector h f Map to the output space to obtain the digital model of the millimeter-wave security inspection instrument.

4. The millimeter wave security inspection instrument health management method according to claim 3, characterized in that: Get the fused feature vector h f ,include: Where, is the attention weight of working state characteristics and working performance characteristics, n is the subsystem of millimeter wave security inspection instrument, is the attention score of work status features and work performance features, v is the attention vector, W a is the attention weight matrix, b a is the attention bias vector, tanh(·) is the hyperbolic tangent function, and exp is the exponential function.

5. The millimeter wave security inspection instrument health management method according to claim 3, characterized in that: The expression of the digital model of millimeter wave security inspection instrument is: Where y(t) is the measured value of the digital model of the millimeter wave security inspection instrument, is the predicted value of the millimeter wave security inspection instrument digital model, ActivationFunctino(·) is the activation function, W m 、b m are the weight matrix and bias vector corresponding to the digital model of millimeter wave security inspection instrument.

6. The millimeter wave security inspection instrument health management method according to claim 5, characterized in that: Dynamically update the parameters W of the millimeter wave security inspection instrument digital model based on reinforcement learning m 、b m ; Let the parameters of the millimeter wave security inspection instrument digital model be in the form of θ, so that θ=[W m ; b m ] T ; Define state s t , represents the comprehensive state of the millimeter wave security inspection instrument at time t; Define the action as a t , represents the update amount Δθ of the digital model of the millimeter wave security inspection instrument t ; Define reward r t , represents the error between the measured and predicted values of the digital model of the millimeter wave security detector; Define strategy π θ (a t |s t ), indicating that in state s t Next, select action a t The probability distribution of Construct the objective function J(θ) to represent the strategy π θ (a t |s t ) under the expected cumulative discounted reward; Combined reward r t , discount factor Υ t and state s t , construct the time difference error δ t ; According to the timing difference error δ t , calculate the policy gradient Δ of the objective function θ J(θ); For the policy gradient Δ θ J(θ), use the gradient ascent method to update θ until the strategy converges, and obtain the updated parameters of the millimeter wave security inspection instrument digital model.

7. The millimeter wave security inspection instrument health management method according to claim 6, characterized in that: Timing difference error δ t , expressed by the following formula: δ t =r t +Υ t V π (s t+1 )-V π (s t ); Where V π (·) represents the value function.

8. The millimeter wave security inspection instrument health management method according to claim 6, characterized in that: Policy gradient Δ of the objective function θ J(θ) is obtained by the following formula: Where, is the expectation, Δ θ logπ θ (a t |s t ) is the logarithmic gradient of the policy, and T is the total time.

9. The millimeter wave security inspection instrument health management method according to claim 6, characterized in that: The criterion for strategy convergence is: ‖θ k+1 -θ k ‖2≤∈; Where θ k+1 represents the parameters of the millimeter-wave security inspection instrument digital model at the k+1th iteration; ∈ represents the convergence threshold.

10. A millimeter wave security inspection instrument health management system, characterized in that: The millimeter wave security inspection device health management method according to any one of claims 1 to 9 comprises: Dynamic monitoring module, used to obtain monitoring data on the working status and performance of each subsystem of the millimeter wave security inspection instrument and generate real-time operating status messages; The digital twin module is used to build a digital model of the millimeter-wave security scanner based on real-time operating status messages and the integration of working status and working performance information features using an attention mechanism. It also updates the parameters of the millimeter-wave security scanner digital model in real time based on a reinforcement learning algorithm. The health management module completes the fault detection of the millimeter wave security scanner based on the residual of the predicted value and the measured value output by the digital model of the millimeter wave security scanner.

Images