A levitation control fault detection and recovery method, device and medium for a maglev train

Abstract

This invention discloses a method, device, and medium for levitation control fault detection and recovery in maglev trains, relating to the field of maglev trains. The method includes the following steps: S1, constructing a linearized levitation system model; S2, reconstructing the state vector and generating an estimated levitation gap; S3, constructing a set of healthy probes and executing a signal selection strategy; S4, detecting slowly changing faults based on the statistical cumulative characteristics of residuals and outputting the slowly changing fault detection results; S5, monitoring the residual change trend, and triggering a reset mechanism to clear historical accumulated data when the reset condition is met; S6, generating a final fault flag, and if the reset mechanism is triggered, forcibly updating the fault flag to a normal state. Based on the hierarchical collaborative decision-making architecture of the maglev train's overlapping structure dynamic model, the algorithm of this invention will not experience performance degradation or failure in real systems, thereby improving the engineering applicability and reliability of fault diagnosis.

Description

Technical Field

[0001] This invention relates to the field of maglev train technology, and more particularly to electrical digital processing, specifically a method, equipment, and medium for detecting and recovering levitation control faults in maglev trains. Background Technology

[0002] High-speed maglev trains, as a new type of rail transit that breaks through the speed limitations of traditional wheel-rail transportation, have become a core support for building a new generation of high-speed passenger transport systems due to their advantages of contactless operation, low friction, and high operating efficiency. The suspension control system plays a crucial role in maintaining stable levitation of the train, ensuring operational safety, and improving passenger comfort. In this system, the gap sensor is a key feedback element used to accurately measure the levitation gap between the electromagnet and the track in real time; the accuracy of its signal directly determines the effectiveness of closed-loop control. However, in actual operation, the sensor probe is exposed to extreme and complex conditions such as strong electromagnetic fields, high-frequency mechanical vibrations, and drastic temperature changes for extended periods. This makes it highly susceptible to faults such as output signal drift, distortion, and even complete failure. If these faults are not detected and handled in a timely manner, the control system will adjust based on erroneous information. This can range from causing levitation fluctuations that affect comfort to potentially leading to system instability and serious safety accidents.

[0003] To ensure system reliability, hardware redundancy strategies are commonly adopted in engineering, which involves deploying multiple independent probes at each measurement point. For sensor fault diagnosis, existing technologies mainly include the following approaches: first, the instantaneous residual comparison method based on a fixed threshold, which responds quickly to sudden step faults; second, the sequential probability ratio test algorithm based on statistical hypothesis testing, which has good early detection capabilities for slowly changing drift faults; and third, the intelligent diagnostic methods based on deep learning that have emerged in recent years. Furthermore, most research is based on a simplified single-iron suspension model.

[0004] However, existing technical solutions still have a series of inherent defects. First, the simplified dynamic models used in many studies do not fully consider the overlapping structure formed by the mechanical coupling of adjacent electromagnets through the suspension frame in actual maglev trains, resulting in insufficient robustness of the algorithms in real systems. Second, the fault diagnosis granularity is coarse, often judging the entire sensor unit as a whole, failing to accurately locate specific fault probes, and wasting hardware redundancy resources. Third, different detection algorithms have fragmented capabilities; the threshold method has a high false negative rate for slowly changing faults, the sequential probability ratio test algorithm will have a recovery delay after the physical fault ends due to the influence of historical accumulation, while complex intelligent algorithms suffer from problems such as large computational load, poor interpretability, and difficulty in meeting real-time control requirements. Finally, there is a lack of a hierarchical collaborative fault-tolerant architecture that can adaptively switch strategies according to fault type and severity.

[0005] Therefore, it is necessary to improve upon the shortcomings of existing technologies in order to solve the above problems. Summary of the Invention

[0006] This invention overcomes the shortcomings of existing technologies and provides a method, device, and medium for fault detection and recovery in levitation control of maglev trains. Based on the dynamic model of the overlapping structure of a high-speed maglev train, a three-layer hybrid intelligent decision-making framework is constructed, integrating sequential probability ratio testing of signal selection strategies and residual chi-square detection. An intelligent reset mechanism driven by residual chi-square statistics is introduced, enabling probe-level refined fault location, rapid detection of both sudden and gradually changing faults, and effectively solving the algorithm delay problem after the fault ends, thereby improving the system's safety, reliability, and availability.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: Firstly, the present invention provides a method for detecting and recovering levitation control faults in maglev trains, comprising the following steps:

[0008] S1. Based on the dynamic characteristics of the maglev train's connecting structure, a linear suspension system model is constructed;

[0009] S2. Reconstruct the complete system state vector and generate the estimated value of the suspension gap based on the full-state feedback control algorithm;

[0010] S3. Calculate the residual between each probe signal and the estimated value. Based on the comparison between the absolute value of the residual and the preset threshold, dynamically construct a set of health probes and execute the signal selection strategy.

[0011] S4. Run the sequential probability ratio test algorithm independently for each probe, detect slowly changing faults based on the statistical cumulative characteristics of the residuals, and output the slowly changing fault detection results.

[0012] S5. Run the residual chi-square detection algorithm independently for each probe. Based on the residual square sum and sudden fault detection results within the sliding window, monitor the residual change trend. When the preset reset conditions are met, trigger the reset mechanism to clear historical accumulated data.

[0013] S6. Combine the results of gradual fault detection with the results of sudden fault detection to generate the final fault flag. If the reset mechanism is triggered, the fault flag is forcibly updated to the normal state to achieve rapid fault-tolerant recovery of the system.

[0014] In a preferred embodiment of the present invention, the construction of the linearized suspension system model in step S1 specifically includes the following steps:

[0015] S11. Based on the electromagnetic force model, electrical model and mechanical coupling relationship of the electromagnet in the maglev train's connecting structure, establish a system mechanism model;

[0016] S12. The mechanism model is linearized near the steady-state operating point of the system to obtain a linearized suspension system model in state-space form.

[0017] In a preferred embodiment of the present invention, in step S2, the reconstructed state vector estimates the state variables in the system that cannot be directly measured by a reduced-order observer, and based on all directly measurable and estimable state variables, a full-state feedback controller is designed using a linear quadratic optimal control algorithm, and the estimated value of the suspension gap is generated.

[0018] The performance index of the quadratic form is: ;

[0019] in, x Represents the system state vector. x T Represents the transpose of the system state vector; u The voltage across the electromagnet is... u T Indicates its transpose; Q , R Let be the performance index parameter matrix, and satisfy: Q T = Q ≥0, R T = R ≥0. Quadratic performance index J for: .

[0020] in, v The optimal feedback gain matrix; B The input matrix describes the effect of the input on the state. B T Indicates its transpose; for a symmetric positive definite matrix P , is an equation Riccati Solution:

[0021] ;

[0022] in, A It is a state matrix, used to describe the internal dynamic characteristics of the system. A T This indicates its transpose.

[0023] In a preferred embodiment of the present invention, step S3, which involves constructing a set of health probes and executing a signal selection strategy, specifically includes the following steps:

[0024] S31. Compare the absolute value of the residual of each probe with a preset threshold, and add probes with residuals not greater than the threshold to the health probe set;

[0025] S32. Based on the different number of probes in the set of health probes, a corresponding signal selection strategy is adopted to determine the suspension gap feedback value used for control.

[0026] In a preferred embodiment of the present invention, the signal selection strategy adopts the following different processing methods according to the number of probes in the health probe set:

[0027] Case A: If there are at least two health probes, the average value of all health probe signals is output as the floating gap feedback value.

[0028] Scenario B: If there is only one health probe, the signal from that health probe will be output as the feedback value of the suspension gap, and a sensor warning will be triggered.

[0029] Scenario C: If there is no health probe, switch to the redundant control mode based on the estimated suspension gap value and trigger a sensor failure alarm.

[0030] In a preferred embodiment of the present invention, in step S4, the sequential probability ratio test algorithm specifically includes the following steps:

[0031] S41. For each probe, establish a null hypothesis that the system is fault-free and an alternative hypothesis that it is faulty;

[0032] S42. Based on the historical residual sequence of the probe, sequentially calculate the cumulative log-likelihood ratio; where, the... k The likelihood ratio of the steps for:

[0033] ;

[0034] in, It is the variance of the residuals; M s For residuals r ( k Mean; H 1 As an alternative hypothesis, i.e., the system has faults, the residuals r ( k The mean is M s ;

[0035] No. k The log-likelihood of the step is greater than the cumulative sum. S ( k )for:

[0036] ;

[0037] S43. Compare the cumulative sum of the log-likelihood ratios with preset upper and lower thresholds to determine whether the probe has a slow-change fault.

[0038] In a preferred embodiment of the present invention, step S5, which involves the residual chi-square detection algorithm and triggering a reset mechanism, specifically includes the following steps:

[0039] S51. Calculate the chi-square statistic of the residuals for each probe at the current time: ;in, It is the covariance of the residuals;

[0040] S52. When the sequential probability ratio test algorithm has determined a fault and the residual chi-square statistic has decreased from its historical peak by more than the preset amplitude threshold, a reset signal is triggered.

[0041] S53. The reset signal is used to clear the historical accumulated data of the sequential probability ratio test algorithm of the corresponding probe.

[0042] In a preferred embodiment of the present invention, in step S6, generating the final fault flag specifically involves:

[0043] The slowly varying fault detection result output by the sequential probability ratio test algorithm is used as the initial value of the final fault flag of the corresponding probe.

[0044] If the reset mechanism is triggered, the final fault flag of the corresponding probe will be forcibly updated to a value indicating a normal state, regardless of the initial value.

[0045] In a second aspect, the present invention provides an electronic device, comprising: at least one processor; and a memory communicatively connected to at least one of the processors;

[0046] The memory stores a computer program that is executed by at least one of the processors, such that the at least one processor is able to execute any of the above-described methods for detecting and recovering levitation control faults in maglev trains.

[0047] Thirdly, the present invention provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the suspension control fault detection and recovery method for maglev trains described above.

[0048] This invention addresses the shortcomings of the prior art and has the following beneficial effects:

[0049] (1) This invention provides a method, device, and medium for fault detection and recovery in levitation control of maglev trains. Based on a hierarchical collaborative decision-making architecture of the maglev train's overlapping structure dynamic model, it can accurately characterize the multibody dynamics formed by the mechanical coupling of adjacent electromagnets through a flexible suspension frame. It can reflect the dynamic coupling effect between the left and right electromagnets, enabling targeted processing of different fault types. This ensures that the fault diagnosis algorithm is highly consistent with the physical characteristics of the controlled object, avoiding deviations caused by model simplification. Compared with existing methods that rely on simplified models of single-iron suspension systems, the algorithm of this invention will not experience performance degradation or failure in real systems, thereby improving the engineering applicability and reliability of fault diagnosis.

[0050] (2) In this invention, a three-layer hybrid intelligent decision-making framework is constructed that integrates signal selection strategy, sequential probability ratio test and residual chi-square detection. The signal selection strategy responds quickly to sudden step faults, the sequential probability ratio test identifies slowly changing faults in the early stage, and the residual chi-square detection assists in detection and triggers reset. This framework can cover all types of faults such as step and slowly changing faults, thereby achieving accurate detection under different fault modes. Compared with the isolated operation of various detection methods in the prior art, which cannot meet the detection needs of sudden and slowly changing faults, this invention solves the problem through a collaborative mechanism, further improving the comprehensiveness and accuracy of fault detection, thereby reducing missed detections and misjudgments.

[0051] (3) In this invention, by introducing an intelligent reset mechanism driven by residual chi-square statistics, residual chi-square detection can sense the end time of the fault in real time and trigger the historical data clearing operation of the sequential probability ratio test algorithm. This can effectively solve the memory effect problem inherent in the sequential probability ratio test algorithm and avoid the situation where the fault flag continues to be output after the fault ends. Compared with the prior art, the sequential probability ratio test algorithm has a recovery delay after the fault is eliminated, which leads to the misadjustment of the control system. This invention can achieve rapid recovery after the fault ends through the reset mechanism, thereby improving the transient performance and stability of the system. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is the overall system architecture diagram of the suspension control fault detection and recovery method of Embodiment 1 of the present invention;

[0054] Figure 2This is a force analysis diagram (b) of the end vehicle structure of the high-speed maglev train in Embodiment 1 of the present invention and the lap structure suspension system.

[0055] Figure 3 This is a schematic diagram of the current loop design of the full-state feedback controller in Embodiment 1 of the present invention;

[0056] Figure 4 This is a schematic diagram illustrating the working principle of the reset decision logic in Embodiment 1 of the present invention;

[0057] Figure 5 The diagram shows the probe health status statistics (a) and the simulation diagram of the fault tolerance effect of the suspension gap when a single sensor probe experiences a step fault after adopting the signal selection strategy of the present invention (b).

[0058] Figure 6 The diagram shows the probe health status statistics (a) and the simulation diagram of the fault tolerance effect of the suspension gap when the dual sensor probes simultaneously experience a step fault after adopting the signal selection strategy of the present invention (b).

[0059] Figure 7 The diagram shows the health status statistics of the sensor probe when a slow-change fault occurs after adopting the signal selection strategy of the present invention (a), and the simulation effect comparison (b) when the fault detection capability is insufficient.

[0060] Figure 8 The simulation results of using the sequential probability ratio test algorithm (SPRT) of this invention to detect slowly changing faults in single probe (a) and dual probe (b) are shown in the figure.

[0061] Figure 9 The simulation results show the residual chi-square detection algorithm introduced in this invention providing a reset signal for SPRT and solving the recovery delay problem after single probe (a) and dual probe (b) failures.

[0062] Figure 10 This is a simulation diagram showing the auxiliary detection effect of the residual chi-square detection algorithm introduced in this invention on single-probe (a) and dual-probe (b) step faults;

[0063] Figure 11 The diagram shows the probe health status statistics (a) and the simulation diagram of the fault tolerance effect of the suspension gap when a single sensor probe fails under the actual working conditions of crossing the rail gap after adopting the intelligent decision-making algorithm of this invention.

[0064] Figure 12 The diagram shows the probe health status statistics (a) and the simulation diagram of the fault tolerance effect of the suspension gap when both sensor probes fail under the actual working conditions of crossing the rail gap after adopting the intelligent decision-making algorithm of this invention.

[0065] Figure 13These are simulation results of single probe (a) and double probe (b) of the residual chi-square detection algorithm introduced in this invention for auxiliary detection of actual working conditions of rail gaps;

[0066] Figure 14 A schematic diagram of an electronic device structure that can be used to implement Embodiment 1 of the present invention is shown. Detailed Implementation

[0067] 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.

[0068] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein. Therefore, the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0069] Example 1: As Figure 1 As shown, a method for detecting and recovering levitation control faults in maglev trains includes the following steps:

[0070] S1. Based on the dynamic characteristics of the maglev train's connecting structure, a linear suspension system model is constructed;

[0071] S2. Estimate the suspension gap and other unmeasured state variables using a reduced-order observer to reconstruct the complete system state vector; based on the reconstructed state vector, use a full-state feedback control algorithm to generate the suspension gap estimate.

[0072] S3. Calculate the residual between each probe signal and the estimated value. Based on the comparison between the absolute value of the residual and the preset threshold, dynamically construct a set of health probes and execute the signal selection strategy.

[0073] S4. Run the sequential probability ratio test algorithm independently for each probe, detect slowly changing faults based on the statistical cumulative characteristics of the residuals, and output the slowly changing fault detection results.

[0074] S5. Run the residual chi-square detection algorithm independently for each probe. Based on the residual square sum and sudden fault detection results within the sliding window, monitor the residual change trend. When the preset reset conditions are met, trigger the reset mechanism to clear historical accumulated data.

[0075] S6. Combine the results of gradual fault detection with the results of sudden fault detection to generate the final fault flag. If the reset mechanism is triggered, the fault flag is forcibly updated to the normal state to achieve rapid fault-tolerant recovery of the system.

[0076] In some specific implementations, high-speed maglev trains use 3-5 cars per train, with each car mainly composed of subsystems such as the car body, levitation, guidance, and traction. For example... Figure 2 The diagram (a) shows the end vehicle structure of a high-speed maglev train. In a high-speed maglev train, levitation electromagnets are connected end-to-end by suspension frames to form a flexible suspension structure. The structure in which adjacent electromagnets are supported together by suspension frames is usually called an overlapping structure. The entire carriage achieves load distribution through 16 suspension frames, which are physically decoupled from each other through mechanical structures. Therefore, this invention selects a suspension frame unit as the research object to analyze the suspension control problem of the maglev train.

[0077] Among them, the levitation overlap structure is the basic unit of the levitation system of high-speed maglev trains. The term "levitation overlap structure system" is usually used to refer to a levitation system consisting of a levitation frame and corresponding controllers, sensors and tracks. Figure 2 Figure (b) shows the force analysis diagram of the overlapping structure suspension system. The left and right suspension electromagnets are supported by a metal-rubber spring, which can be approximated as a spring-damped system, supporting the suspension frame. The suspension frame, in turn, is supported by air springs on the vehicle body. The suspension frame consists of arms and beams, used to connect the two electromagnets and the vehicle body.

[0078] In this embodiment, in step S1, for the levitation electromagnet, let... N e This represents the number of turns in the electromagnet coil. R e For winding resistance; A e The area of ​​the electromagnet's poles; μ 0 air permeability; subscript Let represent the variables on the left and right sides, respectively. The system's relevant equations can be obtained as follows:

[0079] ;

[0080] ;

[0081] ;

[0082] ;

[0083] in, z The displacement of the upper surface of the electromagnet core relative to the lower surface of the track is called the levitation gap of the electromagnet. It is the first derivative of the levitation gap of the electromagnet, representing the rate of change of the levitation gap, that is, the relative velocity between the iron core and the track; It is the second derivative of the levitation gap of the electromagnet, representing the acceleration of the change in the levitation gap, that is, the relative acceleration between the iron core and the track; h The height of the electromagnet, M The mass of the electromagnet; g It is the acceleration due to gravity; l The natural lengths of the disc springs on both the left and right sides. K For its stiffness, η Its equivalent damping; u The voltage across the electromagnet is... i For its current, F e The electromagnetic force it generates F s The restoring force of the disc spring on the support arm is upward; M h , M c The equivalent mass of the suspension frame and the equivalent mass of the vehicle body are given. M c g Applying interference forces to the suspension frame from the vehicle body; H This represents the displacement of the lower edge of the support arm relative to the lower surface of the track. The absolute vertical acceleration of the lower edge of the support arm; The inertial force required to generate absolute acceleration in the suspended frame; It is the gravity acting on the suspended frame itself. N e This represents the number of turns in the electromagnet coil. R e For winding resistance; A e The area of ​​the electromagnet's poles; μ 0 air permeability; d / dt It is the symbol for the derivative with respect to time, and is usually used to describe the rate at which a physical quantity changes with time.

[0084] Furthermore, based on the principle of Taylor series expansion, ignoring higher-order infinitesimal terms, a Taylor series expansion is performed near the steady-state levitation point of the suspension system to obtain a linearized model of the high-speed maglev train's lap-structure suspension system:

[0085] ;

[0086] in, x ( t ) represents the system state vector, The derivative of the state vector is represented by the rate of change of the state. A It is a state matrix used to describe the internal dynamic characteristics of the system;B The input matrix is ​​used to describe the effect of the input on the state; a Indicates control variables; y ( t ) is the output variable; C It is the output matrix, used to describe the effect of the state on the output; D It is a direct transfer matrix, used to describe the direct effect of the input on the output, and is usually negligible as a zero matrix.

[0087] In this embodiment, in step S2, in order to improve the current response speed of the electromagnet and to complete the order reduction simplification of the system model, a strategy of introducing current feedback to form a new current loop is usually adopted to process the electromagnet component, such as... Figure 3 As shown. By selecting the appropriate k c1 , k c2 The coefficient can reduce the time constant of the electromagnet current with respect to voltage. Therefore, the reduced-order system state equation is:

[0088] ;

[0089] Among them, variables ; z j0 The set suspension gap for the electromagnets on both sides; i j0 The electromagnet current is the current of the electromagnets that are stably suspended at a set gap on the left and right sides. M h The equivalent mass of the floating frame; K The stiffness of the disc springs on the left and right sides is given by the subscript. These represent the variables on the left and right sides, respectively. K 1 This indicates the stiffness of the spring on the left disc. Similarly, u 1 The voltage across the left electromagnet is... y 1 This is the output variable on the left.

[0090] Preferably, relevant property analysis is performed on the established state-space equations: system matrix A The eigenvalues ​​contain solutions in the right half-plane of the complex plane, therefore the suspension system is unstable, and a controller needs to be designed to achieve closed-loop stability. Secondly, the controllability matrix... Its rank is 6, which is full rank; observability matrix The rank is 6, which is also a full rank. Among them, A , B , C The matrix is ​​a parametric matrix. AIt is a state matrix used to describe the internal dynamic characteristics of the system; B The input matrix is ​​used to describe the effect of the input on the state; C It is the output matrix, used to describe the influence of the state on the output; therefore, it shows that the suspended system is controllable and the system variables can be observed by designing an observer.

[0091] Meanwhile, considering that there are two unknown states in the suspended overlapping structure model, namely the vertical displacement of the support arm... H Vertical displacement velocity of the support arm To design a full-state feedback controller, a reduced-order observer is used to observe the unknowns. The system is then decomposed according to its measurable states as follows:

[0092] ;

[0093] in, For measurable state variables, For the state variables that need to be observed, A 11 , A 12 , A 21 , A 22 , B 1. B 2 are matrices in the system state equations. A , B Corresponding to x 1 , x 2 The block matrix.

[0094] Furthermore, according to Luenberger By observing the observer and eliminating the differential signal, the reduced-order observer model can be obtained as follows:

[0095] ;

[0096] in, ; L The vector can be freely chosen. According to observer theory, its convergence rate is related to the parameters. L This is relevant, therefore, it is necessary to choose the appropriate one. L Make the matrix ( A 22 – LA 12 )become Hurwitz The matrix is ​​used to ensure the stability of the observer.

[0097] Furthermore, after designing the reduced-order observer, all state variables of the lap structure can be obtained through direct measurement or estimation by the observer. Therefore, a linear quadratic optimal controller design method is adopted for the design of the full-state feedback controller. For a steady-state system, the quadratic performance index is:

[0098] ;

[0099] in, x Represents the system state vector. x T Represents the transpose of the system state vector; u The voltage across the electromagnet is... u T Indicates its transpose; Q , R Let be the performance index parameter matrix, and satisfy: Q T = Q ≥0, R T = R ≥0. Quadratic performance index J for:

[0100] .

[0101] in, v The optimal feedback gain matrix; B The input matrix describes the effect of the input on the state. B T Indicates its transpose; for a symmetric positive definite matrix P , is an equation Riccati Solution:

[0102] ;

[0103] in, A It is a state matrix, used to describe the internal dynamic characteristics of the system. A T This indicates its transpose.

[0104] In this embodiment, in step S3, as the primary execution layer, the signal selection strategy can identify faulty probes in real time and exclude them from the signal fusion process, while automatically selecting the signals from the remaining healthy probes as the feedback value for the electromagnet levitation gap. In this way, the control system can still obtain accurate and effective gap information in most fault scenarios, thus remaining largely unaffected by individual probe failures, achieving effective positioning of faulty probes and seamless fault-tolerant operation of the system.

[0105] This embodiment uses the left electromagnet front-end sensor 1 as an example to describe the algorithm execution flow:

[0106] a) Calculate the independent residuals for each probe:

[0107] ;

[0108] in, This indicates the first sensor at the front end of the left electromagnet, sensor 1. i Each probe outputs a signal; This indicates the first sensor at the front end of the left electromagnet, sensor 1. i The independent residual values ​​of each probe; This represents the estimated value of the suspension gap of the front sensor of the left electromagnet based on the full-state observer.

[0109] b) Fault diagnosis: Compare the calculated residuals with preset thresholds. By comparing and determining the set of health probes for the system:

[0110] .

[0111] c) Signal selection: based on the set of health probes V The number of elements in the system can be divided into three cases for the levitation system:

[0112] Scenario A: If If at least two probes are healthy, then the output is the average of the outputs from all healthy probes. The system is functioning perfectly normally. .

[0113] Scenario B: If If only one probe is healthy, the value of the healthy probe will be output, and a sensor warning will be triggered, indicating that two probes have failed and require immediate maintenance. .

[0114] Case C: If If no probe is found to be healthy, all probes are deemed faulty. The system switches to redundant control mode based on observer estimates and triggers the highest level sensor failure alarm. .

[0115] In this embodiment, in step S4, to effectively detect slowly varying faults in the sensor probe and overcome the insufficient sensitivity of traditional threshold methods in early fault identification, the Sequential Probability Ratio Test (SPRT) algorithm is introduced. SPRT is a sequence analysis method based on statistical hypothesis testing. Its core idea is to progressively determine whether a system fault has occurred by continuously accumulating statistical evidence from historical residual sequences. It no longer relies solely on abrupt changes in a single measurement value, but rather on the sequential calculation of the likelihood ratio and comparison with a threshold, making it more sensitive to early fault detection and diagnosis. The specific algorithm is described below:

[0116] a) Definition of assumptions:

[0117] H 0 (Null hypothesis): The system is fault-free, and the residuals are zero. r ( k The mean is 0.

[0118] H 1 (Alternative Assumption): The system has faults, and the residuals... r ( k The mean is M s .

[0119] b) Calculate the likelihood ratio: Assuming the residuals follow a Gaussian distribution, then the first... k The likelihood ratio of the steps :

[0120] ;

[0121] in, It is the variance of the residuals; M s For residuals r ( k (mean)

[0122] c) Calculate and accumulate the log-likelihood ratio: To simplify calculations, the logarithm is usually used. Define the... k The log-likelihood of the step is greater than the cumulative sum. S ( k )for:

[0123] .

[0124] d) Setting thresholds: The upper and lower thresholds are determined by the desired false alarm rate. and false alarm rate pass Wald Formula determined:

[0125] ;

[0126] in, T 1 The upper threshold is used to determine when a fault occurs; T 0 This is the lower threshold used to determine if there is no fault. When ,accept H 1 The system is faulty. ,accept H 0 No fault was found; Further testing is needed, but no decision can be made.

[0127] In this embodiment, such as Figure 4 As shown, in steps S5 and S6, to overcome the delay in response after a fault ends and to enhance the ability to detect sudden faults, this implementation introduces a residual chi-square detection algorithm. This algorithm can effectively detect step faults, supplementing the shortcomings of the signal selection strategy; on the other hand, it can provide an intelligent reset signal for SPRT, significantly shortening the system recovery time by clearing its historical accumulated values, thus achieving a complementary advantage of the two methods. The specific algorithm is described below:

[0128] a) Calculate the chi-square statistic:

[0129] ;

[0130] in, It is the covariance of the residuals.

[0131] b) Set the detection threshold:

[0132] ;

[0133] in, False alarm rate; T d This is the detection threshold.

[0134] c) Fault decision-making logic:

[0135] .

[0136] d) Reset decision logic: Combine the sequential probability ratio test with the residual chi-square algorithm to generate the system's final fault flag. F ( k ):

[0137] ;

[0138] in, It is the decision output of SPRT; It is a reset signal, and its triggering condition is:

[0139] ;

[0140] in, This indicates the extent to which the residual chi-square statistic has decreased from its historical peak. This is the threshold for the rate of decrease.

[0141] Furthermore, when the reset signal R ( k When )=1 is triggered, the history data of the SPRT sliding window is forcibly cleared: .

[0142] To verify the effectiveness of the proposed method in fault location and system fault tolerance, simulation verification was conducted. Considering the low probability of all probes failing simultaneously, the simulation analysis mainly focused on single-probe and dual-probe fault scenarios. Taking the left electromagnet front-end sensor as the research object, step fault signals with an amplitude of 2 mm were injected into probe 1 individually and simultaneously into probes 1 and 2 within a time interval of 3-8 seconds to simulate common sudden fault conditions in actual systems.

[0143] Simulation results are as follows Figure 5 Statistics on the health status of a single sensor probe experiencing a step failure (a) and simulation diagram of the fault tolerance effect of the suspension gap (b), and Figure 6 The probe health status statistics (a) and simulation diagram (b) of the fault tolerance effect of the suspension gap when the dual sensor probes simultaneously experience a step failure are shown.

[0144] It should be noted that the simulation results fully verify the excellent performance of the signal selection full-state feedback control strategy proposed in this invention in fault diagnosis and fault-tolerant control. For example... Figure 5 As shown, when the system encounters a step fault in probe 1 at 3 s, the strategy can achieve rapid and accurate fault location in 3.02 s, with a false negative rate as low as 3.38%, indicating that the system has high fault detection sensitivity and reliability. More importantly, during the fault occurrence, the system successfully maintained the stability of the suspension gap by automatically switching to the healthy probe signal. The fluctuation amplitude of the gap of the left electromagnet was controlled within 0.14 mm, and the fluctuation of the right electromagnet was even smaller, only 0.06 mm, which is about 4.5 times less than the fluctuation value (0.62 mm / 0.28 mm) under the fault-tolerant control strategy, fully meeting the requirements of high-speed maglev train operation smoothness and passenger comfort.

[0145] Even in more extreme dual-probe failure scenarios (probe 1 and probe 2 fail simultaneously), this invention still demonstrates strong fault tolerance. For example... Figure 6 As shown, the system completed fault identification and isolation within the same time frame (3.02 s), with the false negative rate of probe 1 remaining at 3.38% and the false negative rate of probe 2 at 6.98%. Despite the increased severity of the fault, the system stabilized the fluctuation of the suspension gap within the range of 0.34 mm on the left and 0.16 mm on the right through an effective signal reconstruction mechanism, further demonstrating its good robustness.

[0146] The above results demonstrate that this invention not only achieves rapid and accurate fault location, but more importantly, it maintains stable system operation after a fault occurs through an effective redundancy management mechanism, avoiding control performance degradation or system instability caused by sensor failure, thus exhibiting good engineering application value and reliability. However, the signal selection strategy only performs binary judgment based on instantaneous residuals, and has low sensitivity to slowly changing faults (such as drift faults). Figure 7 As shown in the figure, it is impossible to effectively identify faults in their early stages, resulting in a high false negative rate and delayed fault-tolerant response. To address this, this invention introduces the Sequential Probability Ratio Test (SPRT) as an advanced diagnostic algorithm for fault result simulation.

[0147] Furthermore, to verify the performance of the SPRT algorithm in detecting slowly changing faults, this invention conducts a simulation study. The simulation focuses on the front-end sensor of the left electromagnet, injecting slowly changing fault signals into probe 1 alone and simultaneously into probes 1 and 2 within a time frame of 3-8 seconds. The simulation results are as follows: Figure 8 The simulation results of detecting slowly changing faults using a single probe (a) and a dual probe (b) are shown in the figure.

[0148] like Figure 8 As shown, the Sequential Probability Ratio Test (SPRT) algorithm can effectively detect slowly changing faults, and its response speed and false negative rate are significantly better than those of the signal selection strategy for detecting slowly changing faults. Simulation results show that SPRT can trigger an alarm within about 3.21 s after a fault occurs, which is about 0.39 s earlier than the signal selection strategy; at the same time, the false negative rate is significantly reduced from 13.98% for the signal selection strategy to 4.6% (single fault) and 3.8% (double fault).

[0149] However, due to its inherent cumulative judgment mechanism, the SPRT algorithm retains a fault state misjudgment for approximately 0.6 seconds after the fault ends. This delay may cause the suspension control system to continuously adjust based on erroneous signals, leading to unnecessary control fluctuations or overshoot, and affecting the system's transient recovery performance. Therefore, this invention further introduces a residual chi-square detection algorithm to construct a hybrid decision-making and reset mechanism.

[0150] Furthermore, to verify the dual function of the residual chi-square algorithm—namely, providing a reset signal for SPRT and assisting in the detection of step faults—this invention conducts simulations for both slowly varying and step fault scenarios. Under slowly varying fault conditions (3–8 s), the effectiveness of its reset mechanism is evaluated; under step fault conditions (3–8 s), its auxiliary detection performance for sudden faults is examined. Simulation results are as follows: Figure 9 Simulation results showing the recovery delay after single-probe (a) and dual-probe (b) failures, and... Figure 10 The simulation results of auxiliary detection of step faults using a single probe (a) and a dual probe (b) are shown in the figure.

[0151] In scenarios with slowly varying fault injection, the residual chi-square algorithm demonstrates excellent fault termination identification capabilities. For example... Figure 9 As shown, when the fault ends at 8 seconds, the SPRT algorithm, relying on historical accumulated values, still misclassifies the system as being in a fault state, with a recovery delay of approximately 0.6 seconds. Meanwhile, the residual chi-square algorithm, by monitoring residual changes in real time, captures a significant decrease in residual amplitude at the moment the fault ends (8 seconds) and triggers a reset signal at 8.01 seconds. This signal forcibly clears the SPRT historical window, allowing it to quickly return to its initial state, reducing the system recovery delay from the fault state to 0.01 seconds.

[0152] In step fault detection, the residual chi-square algorithm exhibits excellent response performance. For the condition where both probe 1 and probe 2 fail simultaneously, the algorithm can complete fault identification within 3.01 seconds with a false negative rate of only 2%, which is significantly better than the false negative rate of 3.38–6.98% of the signal selection strategy under similar fault conditions, demonstrating its rapid perception and high reliability detection capability for sudden faults.

[0153] Notably, even in the extreme scenario of simultaneous failure of all three probes, the residual chi-square algorithm still exhibits excellent detection characteristics, with an initial detection time approaching 0 seconds, indicating near-transient fault response capability. Although a false negative rate of -0.2% (actually an extremely short delay of approximately 0.01 seconds) occurs, this result remains within an acceptable range for engineering applications and does not affect its effectiveness as an auxiliary detection module in practical systems. The slight delay likely stems from adaptive adjustments to the system's transient processes during the algorithm's initialization phase and does not substantially impact overall control performance.

[0154] The above results demonstrate that the constructed reset mechanism effectively overcomes the response lag problem of the SPRT algorithm after the fault ends, significantly improves the system's adaptability to transient processes and recovery speed, and achieves complementary advantages of different detection methods. Furthermore, the algorithm exhibits fast response and low false negative rate characteristics in step fault detection, effectively supplementing the detection capability of signal selection strategies in sudden fault scenarios.

[0155] Based on previous simulation results, the proposed hybrid decision-making and reset logic framework was applied to the gap-crossing condition in actual operation of a high-speed maglev train to verify its fault detection and fault-tolerant control performance in a complex real-world environment. The gap-crossing condition is a common scenario in maglev train operation. When the gap sensor passes through a track joint, due to the lack of metal material, the probe output will momentarily jump to its maximum value (20 mm). If this transient change is not handled promptly, it may lead to control misjudgment and system instability. Treating the gap-crossing condition as a special transient fault mode, this study examines whether the proposed framework can quickly identify faulty probes, take timely fault-tolerant measures, and reduce suspension gap fluctuations by adjusting the output of the full-state feedback control algorithm, thereby enhancing train running stability and passenger comfort.

[0156] For scenario one, where a single probe passes over the track joint, the output of probe 1, the sensor at the front end of the left electromagnet, is set to 20 mm during the 3.0–3.02 s period. For scenario two, where two probes pass over the track joint, both probe 1 and probe 2 simultaneously output 20 mm during the 3.5–3.52 s period, simulating a more extreme scenario of reduced redundancy. The simulation results are as follows: Figure 11 Statistics on probe health status when a single sensor probe fails (a) and simulation diagram of the fault tolerance effect of the suspension gap (b). Figure 12 Statistics on probe health status when both sensor probes fail (a) and simulation diagram of the fault tolerance effect of the suspension gap (b), and Figure 13 The simulation results of single probe (a) and dual probe (b) are shown for auxiliary detection of the actual working conditions of the rail gap.

[0157] Under operating conditions, the signal selection strategy exhibits excellent fault detection and fault tolerance performance. For example... Figure 11 As shown, the system identified the abnormal state of probe 1 the instant the fault occurred (3.0 s), with an initial detection time of only 3.0 s and a false negative rate of 0%. This indicates that the signal selection strategy is highly sensitive to transient faults and can achieve near real-time fault location. Simultaneously, the system effectively maintained the stability of the suspension gap by automatically switching to the healthy probe signal. After fault-tolerant control, the gap fluctuation value of the left electromagnet was only 0.08 mm, and the fluctuation value of the right electromagnet was only 0.04 mm, which is about 89-90% lower than the fluctuation values ​​without the fault-tolerant control strategy (0.9 mm on the left and 0.4 mm on the right), thus significantly improving the system stability.

[0158] Even under more extreme dual-probe gap-crossing conditions, the proposed framework still exhibits strong fault-tolerant control capabilities. For example... Figure 12As shown, the signal selection strategy simultaneously identified faults in both probe 1 and probe 2 within 3.0 s, with an initial detection time of 3.0 s and a false negative rate of 0%. Despite the increase in the number of faulty probes, the system quickly switched the control signal to the remaining healthy probes through a signal reconstruction mechanism, ensuring the continuity of feedback information. After fault-tolerant control, the gap fluctuation value of the left electromagnet was 0.15 mm, and that of the right electromagnet was 0.07 mm, which is a reduction of approximately 91-91.2% compared to the fluctuation values ​​without fault-tolerant control (1.7 mm on the left and 0.8 mm on the right). The fluctuation amplitude was strictly controlled within a safe range.

[0159] Furthermore, the residual chi-square algorithm also played an important auxiliary detection role in this practical working condition. For example... Figure 13 As shown, the algorithm synchronously triggers an alarm when a fault occurs, with an initial detection time of 3.0 s and a false negative rate of 0%. This further ensures the reliability of fault identification and avoids the risk of false negatives caused by transient noise or other factors in the signal selection strategy. The introduction of the residual chi-square algorithm enhances the redundant detection capability for sudden faults, thus providing dual protection for the system.

[0160] Example 2: Figure 14 A schematic diagram of an electronic device structure that can be used to implement Embodiment 1 of the present invention is shown. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0161] like Figure 14 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor 11, and the computer program is executed by the at least one processor 11 to enable the at least one processor 11 to perform the method provided by the present invention.

[0162] Furthermore, the processor 11 can perform various appropriate actions and processes based on a computer program stored in the read-only memory (ROM) 12 or a computer program loaded from the storage unit 18 into the random access memory (RAM) 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0163] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0164] Furthermore, processor 11 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as methods for resource management of a database.

[0165] In some specific embodiments, the method for managing database resources can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for managing database resources described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the method for managing database resources by any other suitable means (e.g., by means of firmware).

[0166] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard parts (ASSPs), systems-on-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0167] Computer programs used to implement the methods of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs can be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0168] In the context of this invention, a computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method for resource management of a database provided by this invention. The computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. The computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0169] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device for displaying information to the user (e.g., a cathode ray tube (CRT) or a liquid crystal display (LCD monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0170] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0171] Optionally, the computing system may include clients and servers. Clients and servers are generally geographically separated and typically interact via a communication network. The client-server relationship is established by computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system. This addresses the shortcomings of traditional physical hosts and Virtual Private Server (VPS) services, such as high management difficulty and weak business scalability.

[0172] The above description is based on the preferred embodiments of the present invention. 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 implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered exemplary and non-limiting in all respects. The scope of the invention is defined by the appended claims rather than the foregoing description, and all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0173] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for detecting and recovering levitation control faults in maglev trains, characterized in that, Includes the following steps: S1. Based on the dynamic characteristics of the maglev train's connecting structure, a linear suspension system model is constructed; S2. Reconstruct the complete system state vector and generate the estimated value of the suspension gap based on the full-state feedback control algorithm; S3. Calculate the residual between each probe signal and the estimated value. Based on the comparison between the absolute value of the residual and the preset threshold, dynamically construct a set of health probes and execute the signal selection strategy. S4. Run the sequential probability ratio test algorithm independently for each probe, detect slowly changing faults based on the statistical cumulative characteristics of the residuals, and output the slowly changing fault detection results. S5. Run the residual chi-square detection algorithm independently for each probe. The residual chi-square detection algorithm includes calculating the residual chi-square statistics at the current moment, monitoring the residual change trend based on the residual square sum and sudden fault detection results within the sliding window, and triggering a reset mechanism to clear historical accumulated data when the preset reset conditions are met. S6. Combine the results of gradual fault detection and sudden fault detection to generate the final fault flag. If the reset mechanism is triggered, the fault flag will be forcibly updated to the normal state to achieve rapid fault-tolerant recovery of the system. In step S1, the construction of the linearized suspension system model specifically includes the following steps: S11. Based on the electromagnetic force model, electrical model and mechanical coupling relationship of the electromagnet in the maglev train's connecting structure, establish a system mechanism model; S12. Linearize the mechanism model near the steady-state operating point of the system to obtain a linearized suspension system model in state-space form; In step S2, the reconstructed complete system state vector estimates the state variables in the system that cannot be directly measured by a reduced-order observer, and based on all directly measurable and estimated state variables, a full-state feedback controller is designed using a linear quadratic optimal control algorithm, and the estimated value of the suspension gap is generated. The performance index of the quadratic optimal control algorithm is: ; in, x Represents the system state vector. x T Represents the transpose of the system state vector; u The voltage across the electromagnet is... u T Indicates its transpose; Q , R Let be the performance index parameter matrix, and satisfy: Q T = Q ≥0, R T = R ≥0; Quadratic performance index J for: ; in, v The optimal feedback gain matrix; B The input matrix describes the effect of the input on the state. B T Indicates its transpose; for a symmetric positive definite matrix P , is an equation Riccati Solution: ； in, A It is a state matrix, used to describe the internal dynamic characteristics of the system. A T Indicates its transpose; In step S5, the residual chi-square detection algorithm triggers a reset mechanism, specifically including the following steps: S51. Calculate the chi-square statistic of the residuals for each probe at the current time: ;in, It is the covariance of the residuals; S52. When the sequential probability ratio test algorithm has determined a fault and the residual chi-square statistic has decreased from its historical peak by more than the preset amplitude threshold, a reset signal is triggered. S53. The reset signal is used to clear the historical accumulated data of the sequential probability ratio test algorithm of the corresponding probe.

2. The method for detecting and recovering levitation control faults in maglev trains according to claim 1, characterized in that: In step S3, constructing the health probe set and implementing the signal selection strategy specifically includes the following steps: S31. Compare the absolute value of the residual of each probe with a preset threshold, and add probes with residuals not greater than the threshold to the health probe set; S32. Based on the different number of probes in the set of health probes, a corresponding signal selection strategy is adopted to determine the suspension gap feedback value used for control.

3. The method for detecting and recovering levitation control faults in maglev trains according to claim 2, characterized in that: The signal selection strategy adopts the following different processing methods based on the number of probes in the health probe set: Case A: If there are at least two health probes, the average value of all health probe signals is output as the floating gap feedback value. Scenario B: If there is only one health probe, the signal from that health probe will be output as the feedback value of the suspension gap, and a sensor warning will be triggered. Scenario C: If there is no health probe, switch to the redundant control mode based on the estimated suspension gap value and trigger a sensor failure alarm.

4. The method for detecting and recovering levitation control faults in maglev trains according to claim 1, characterized in that: In step S4, the sequential probability ratio test algorithm specifically includes the following steps: S41. For each probe, establish a null hypothesis that the system is fault-free and an alternative hypothesis that it is faulty; S42. Based on the historical residual sequence of the probe, sequentially calculate the cumulative log-likelihood ratio; where, the... k The likelihood ratio of the steps for: ； in, It is the variance of the residuals; M s For residuals r ( k Mean; H 1 As an alternative hypothesis, i.e., the system has faults, the residuals r ( k The mean is M s ; No. k The log-likelihood of the step is greater than the cumulative sum. S ( k )for: ； S43. Compare the cumulative sum of the log-likelihood ratios with preset upper and lower thresholds to determine whether the probe has a slow-change fault.

5. The method for detecting and recovering levitation control faults in maglev trains according to claim 1, characterized in that: In step S6, generating the final fault flag specifically involves: The slowly varying fault detection result output by the sequential probability ratio test algorithm is used as the initial value of the final fault flag of the corresponding probe. If the reset mechanism is triggered, the final fault flag of the corresponding probe will be forcibly updated to a value indicating a normal state, regardless of the initial value.

6. An electronic device, characterized in that, include: At least one processor; and a memory communicatively connected to at least one of the processors; The memory stores a computer program executed by at least one of the processors, which enables the at least one processor to perform the suspension control fault detection and recovery method for a maglev train as described in any one of claims 1-5.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the method for detecting and recovering levitation control faults for maglev trains as described in any one of claims 1-5.

Images