Turnout fault diagnosis system and method based on vibration spectrum analysis

Abstract

The invention discloses a turnout fault diagnosis system and method based on vibration spectrum analysis, and relates to the technical field of railway signal equipment fault diagnosis, and the system comprises a multi-source vibration sensing module, a mechanical transmission chain signal decoupling module, a dynamic baseline modeling module, a fault decision engine module, and a cross-correlation verification module. Vibration signals of different channels can be collected at a turnout through the multi-source vibration sensing module, and the mechanical transmission chain signal decoupling module decomposes the complex vibration signals into three types of independent components by adopting constrained variational mode decomposition signals, so that the running state of each mechanical part can be analyzed more carefully, and the reliability of the system is improved. The dynamic baseline modeling module dynamically updates a threshold value according to factors of the turnout action times and the environment temperature, the fault decision engine module combines judgment rules and compares feature values of all components with the threshold value, and the fault types of gear tooth breakage and locking oil shortage of the switch machine can be accurately distinguished.

Description

Technical Field

[0001] The present invention relates to the technical field of railway signal equipment fault diagnosis, and specifically to a turnout fault diagnosis system and method based on vibration spectrum analysis. Background Art

[0002] Railway transportation, as an infrastructure and a popular means of transportation, occupies an important position in transportation. A turnout, as a line connection device on a railway line that enables a locomotive or vehicle to switch from one track to another, the reliability of its operating state is directly related to the safety and efficiency of railway transportation. With the development of railway transportation towards high-speed and heavy-haul directions, the usage frequency and load of turnouts are continuously increasing, and the probability of faults occurring is also correspondingly increasing. Timely and accurately diagnosing turnout faults is of great significance for ensuring the safe and smooth railway transportation and reducing operating costs.

[0003] Currently, in the field of turnout fault diagnosis, traditional vibration spectrum analysis methods mainly rely on simple frequency-domain or time-domain analysis of the overall vibration signal. This method cannot effectively decompose the vibration signal during the turnout operation process according to different mechanical transmission chains, and it is difficult to accurately distinguish signals generated by different vibration sources such as motor drive, gear transmission, and locking mechanism impact. When facing faults such as broken teeth of the gear of the switch machine and lack of oil in the locking mechanism, due to the certain similarity in the vibration spectrum performance of these faults, traditional methods cannot establish a time-frequency domain energy ratio baseline for each independent component, resulting in a low accuracy rate of fault diagnosis, prone to misjudgment and missed judgment, and unable to meet the demand for precise diagnosis of turnout faults in railway transportation.

[0004] In summary, the existing turnout fault diagnosis technologies have obvious deficiencies in dealing with complex vibration signals and accurately distinguishing fault types. Therefore, there is an urgent practical need to develop a system and method that can effectively decompose vibration signals, establish accurate fault feature baselines, and achieve precise fault diagnosis. Summary of the Invention

[0005] The purpose of the present invention is to make up for the deficiencies of the prior art and provide a turnout fault diagnosis system and method based on vibration spectrum analysis. It can accurately collect vibration signals of different channels at key parts of the turnout through a multi-source vibration sensing module. The mechanical transmission chain signal decoupling module uses constrained variational mode decomposition signals to decompose complex vibration signals into three independent components: motor drive, gear drive, and locking mechanism impact, enabling a more detailed analysis of the operating state of each mechanical component. The dynamic baseline modeling module dynamically updates the threshold according to factors such as the number of turnout operations and environmental temperature to improve the accuracy of fault judgment. The fault decision engine module constructs a three-dimensional feature space joint decision rule, compares the eigenvalue of each component with the threshold, and can accurately distinguish fault types such as broken teeth of the turnout gear and lack of oil in the locking mechanism, reducing misjudgment and missed judgment. This can not only timely detect potential faults of the turnout and avoid further deterioration of the faults, but also provide accurate fault information for railway maintenance personnel, thus effectively ensuring the safe and smooth railway transportation.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: On the one hand, a turnout fault diagnosis system based on vibration spectrum analysis, the system consists of: a multi-source vibration sensing module, a mechanical transmission chain signal decoupling module, a dynamic baseline modeling module, a fault decision engine module, and a cross-correlation verification module;

[0007] The multi-source vibration sensing module is composed of a triaxial acceleration sensor array arranged axially on the turnout housing, on the top of the gearbox, and at the root of the locking rod, synchronously collecting three-channel vibration signals of motor drive, gear drive, and locking mechanism;

[0008] The mechanical transmission chain signal decoupling module physically decomposes the vibration signal into motor drive component, gear drive component, and locking impact component, and calculates three types of time-frequency domain features: harmonic distortion rate H, modulation sideband ratio M, and impact period variance S of each component;

[0009] The dynamic baseline modeling module uses a sliding time window to dynamically update the energy ratio baseline thresholds H th 、M th 、S th ;

[0010] The fault decision engine module generates a fault type diagnosis conclusion according to the over-standard combination of the three types of eigenvalues, and the fault type diagnosis conclusion includes broken gear fault, lack of oil in locking fault, and compound fault;

[0011] The cross-correlation verification module judges the effectiveness of the decomposition by comparing the cross-correlation function of the motor drive component decomposed in real time with the preset no-load experiment template.

[0012] Further, the mechanical transmission chain signal decoupling module synchronously collects vibration signals of three channels, and the physical correspondence of the three-channel vibration signals is as follows:

[0013] Motor drive component: used to reflect the low-frequency vibration caused by motor winding imbalance and bearing defects, with a frequency range of 0 - 500 Hz, to judge the potential faults of the motor;

[0014] Gear transmission component: the medium-frequency vibration representing the gear meshing impact and wear condition, with a frequency range of 500 Hz - 5 kHz, and this component is used to monitor the health status of the gear;

[0015] Locking impact component: used to capture the high-frequency vibration generated at the moment of the locking mechanism collision, with a frequency greater than 5 kHz, to judge the working condition of the locking mechanism.

[0016] Furthermore, during the process of decomposing the mixed vibration signal collected by the multi-source vibration sensing module by the mechanical transmission chain signal decoupling module, by minimizing the function value, the mathematical decomposition of the mixed vibration signal is realized, and it is decomposed into K mutually independent modal components {u k}, and K = 3, that is, K corresponds to the three components of motor drive, gear transmission, and locking impact, and the center frequency {ω k} corresponding to each modal component is determined. The minimized function value satisfies: Among them, {u k} represents each modal component obtained by decomposition, u k (t) represents the real-time decomposition result, {ω k} represents the center frequency corresponding to each modal component, represents the derivative with respect to time t, δ(t) is a time function, j is the imaginary unit, * represents the convolution operation, α is a balance parameter used to adjust the bandwidth of each modal component, P k is a constraint operator, which is set according to the characteristics of mechanical components. Specifically:

[0017] Motor component constraint: P1(u1) = |FFT(u1)| f>500Hz = 0, which means that the motor component u1 is processed by the fast Fourier transform FFT, and the high-frequency energy with a frequency greater than 500 Hz is forced to zero, for detecting the low-frequency vibration of the motor drive component;

[0018] Gear component constraint: represents the integral calculation of the energy of the gear component u2 in the frequency range of 500 Hz to 5 kHz, so that the proportion of this energy in the total energy E total is greater than or equal to 80%, so as to ensure the proportion of medium-frequency energy in the gear component;

[0019] Locking component constraint: Among them, max(|u3(t)|) represents the maximum amplitude of the locking component u3 in the time domain, and A th is the amplitude threshold and A th = 0.5g, where g is the acceleration due to gravity, freq(u3) represents the main frequency of the locking component u3, and [f min , f max is the characteristic frequency range of the vibration signal generated by the impact of the locking mechanism. This constraint requires that the maximum amplitude of the locking component exceeds the set threshold, and its main frequency falls within the frequency range, so as to strengthen the detection of the impact event of the locking mechanism.

[0020] Furthermore, the mechanical transmission chain signal decoupling module calculates three types of time-frequency domain characteristics of each component, namely the harmonic distortion rate H, the modulation sideband ratio M, and the impact period variance S, where:

[0021] The harmonic distortion rate H is used to measure the content of harmonic components in the signal and reflects the degree of deviation of the signal from the fundamental wave. For a periodic signal x(t), its harmonic distortion rate X n represents the amplitude of the nth harmonic component of the signal x(t), and X1 represents the amplitude of the fundamental wave component of the signal x(t), that is, the normal operating frequency of the device;

[0022] The modulation sideband ratio M is used to describe the strength of the modulation phenomenon in the signal. The modulation sidebands are the frequency components generated on both sides of the fundamental wave frequency due to the modulation of the signal. That is, for the modulated signal x(t), its modulation sideband ratio A s-a represents the amplitude of the a-th modulation sideband component, and A c-j represents the amplitude of the j-th carrier component corresponding to the meshing frequency of the gear. N is the number of modulation sidebands, and M is the number of carrier components;

[0023] The impact period variance S is used to describe the periodic stability of the impact events in the signal. For the signal x(t) containing impact events, by detecting the occurrence times t1, t2,..., t n of the impact events in the signal, calculate the time interval T i = t i+1 - t i , i = 1, 2,..., n - 1, to obtain the impact period variance T i is the time interval between the i-th adjacent impact events, is the average value of the time intervals between all adjacent impact events, that is n is the number of detected impact events.

[0024] Furthermore, the dynamic baseline modeling module dynamically updates the threshold based on a sliding time window T W = 30 ± ΔT, and the specific threshold is set as follows:

[0025] The threshold of the motor harmonic distortion rate: H th = 5% + 0.5%·ln(N), where N ∈ Z + represents the cumulative number of switch operations;

[0026] The threshold of the gear modulation sideband ratio: M th = 0.25 + 0.02(T env - 20), where T env ∈ [-10, 50] °C is the ambient temperature;

[0027] The threshold of the variance of the locking impact period: S th = 0.1 ms 2 is a fixed value.

[0028] Furthermore, the fault diagnosis conclusion logic in the fault decision engine module is as follows:

[0029]

[0030] Furthermore, the cross - correlation verification module establishes a pure vibration template u r (t) of the motor through no - load operation, and calculates the normalized cross - correlation coefficient ρ of the real - time decomposition result u k (t), that is where T represents the length of the time series, is the mean value of the pure vibration template u r (t) of the motor, is the mean value of the real - time decomposition result u k (t);

[0031] When ρ ≥ 0.9, the correlation between the real - time decomposition result and the pure vibration template of the motor is high, the decomposition is effective, and fault determination is allowed;

[0032] When 0.8 ≤ ρ < 0.9, it indicates low correlation and signal interference, and at this time, a signal resampling operation is triggered;

[0033] When ρ < 0.8, it means poor correlation, and at this time, a sensor calibration instruction is output.

[0034] On the other hand, a switch fault diagnosis method based on vibration spectrum analysis, and the specific steps of this method are as follows:

[0035] S100. Use the multi - source vibration sensing module to synchronously collect the vibration signals of the three channels of the motor, gear, and locking mechanism through the tri - axial acceleration sensor array;

[0036] S200. Adopt a mechanical transmission chain signal decoupling module to decompose the signal and calculate the harmonic distortion rate, modulation sideband ratio, and impact period variance;

[0037] S300. With the help of a dynamic baseline modeling module, dynamically update the energy ratio baseline thresholds of the three types of components using a sliding time window;

[0038] S400. Through the fault decision engine module, generate a fault type diagnosis result based on the comparison of the exceeded combination of eigenvalue and the threshold;

[0039] S500. Use the cross - correlation verification module to perform cross - correlation verification on the real - time decomposition result and the pure drive component template.

[0040] Compared with the prior art, the turnout fault diagnosis system and method based on vibration spectrum analysis have the following beneficial effects:

[0041] First, through the multi - source vibration sensing module, the present invention can accurately collect vibration signals of different channels at key parts of the turnout. The mechanical transmission chain signal decoupling module uses constrained variational mode decomposition to decompose complex vibration signals into three independent components: motor drive, gear transmission, and locking mechanism impact, enabling a more detailed analysis of the operating state of each mechanical component. The dynamic baseline modeling module dynamically updates the threshold according to factors such as the turnout action times and environmental temperature, improving the accuracy of fault judgment. The fault decision engine module constructs a three - dimensional feature space joint decision rule, compares the eigenvalue of each component with the threshold, and can accurately distinguish fault types such as broken teeth of the turnout gear and lack of oil in the locking mechanism, reducing misjudgment and missed judgment. This can not only timely detect potential faults of the turnout, avoid further deterioration of the faults, but also provide accurate fault information for railway maintenance personnel, thus effectively ensuring the safe and smooth railway transportation.

[0042] Second, the fault diagnosis system and method of the present invention can quickly and accurately locate the fault location and type, enabling maintenance personnel to directly repair and replace the faulty components, avoiding unnecessary comprehensive inspections and maintenance work. At the same time, the cross - correlation verification module can timely detect abnormal conditions of the sensor and trigger calibration instructions, ensuring the stability and reliability of the system, reducing misdiagnosis and unnecessary maintenance caused by sensor faults. In addition, the dynamic baseline modeling module dynamically updates the threshold according to the actual operating conditions, enabling the system to better adapt to different working environments and working condition changes, further improving the efficiency and accuracy of fault diagnosis.

[0043] Other advantages, objects and features of the present invention will be set forth in part in the following description, and in part will be obvious to those skilled in the art based on the examination of the following, or can be learned from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or in the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0045] Figure 1 It is an operation flowchart of a turnout fault diagnosis system based on vibration spectrum analysis;

[0046] Figure 2 It is a step diagram of a turnout fault diagnosis method based on vibration spectrum analysis. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following will, in conjunction with the accompanying drawings and preferred embodiments, describe in detail the specific embodiments, structures, features and their effects of the present invention as follows.

[0048] Embodiment 1

[0049] This embodiment focuses on a turnout fault diagnosis system based on vibration spectrum analysis, and elaborates on its specific operation process in detail. The vibration signals of different parts of the turnout are collected by the multi-source vibration sensing module, the mechanical transmission chain signal decoupling module decomposes the signals and calculates the characteristic values, the dynamic baseline modeling module updates the thresholds, the fault decision engine module combines the characteristic values and the thresholds to diagnose faults, and the cross-correlation verification module ensures the effectiveness of the decomposition, realizing the accurate diagnosis of turnout faults and providing strong support for railway transportation safety.

[0050] Precise monitoring of the operating state of railway switches on-site. In this embodiment, a multi-source vibration sensing module is installed on the switch, which is composed of an array of three-axis acceleration sensors and is respectively arranged on the axial direction of the actuator housing, the top of the gearbox, and the root of the locking rod. Since the axial direction of the actuator housing can effectively collect the vibration generated by the motor drive, the top of the gearbox can obtain the vibration signal during gear transmission, and the root of the locking rod can capture the vibration condition during the operation of the locking mechanism. After the sensors are installed, three-channel vibration signals of the motor drive, gear transmission, and locking mechanism are synchronously collected. During the collection process, the sensors convert mechanical vibrations into electrical signals, and these electrical signals contain rich information about the operating state of the switch. Since the switch will generate vibrations with different frequencies and amplitudes at different working stages (conversion, locking), it is ensured that various vibration signals can be completely collected.

[0051] The vibration signals collected from the multi-source vibration sensing module are mixed signals, which contain vibration information from different mechanical components and cannot be directly used to accurately judge the operating state of each component of the switch. Therefore, the mechanical transmission chain signal decoupling module processes it. Using the constrained variational mode decomposition technology, the mixed vibration signal is decomposed into mutually independent modal components, corresponding to the three key parts of the motor drive, gear transmission, and locking impact of the switch. This goal is achieved by minimizing the function value, that is {u k} represents each modal component obtained by decomposition, that is, the vibration components corresponding to the motor drive, gear transmission, and locking impact, and {ω k} represents the center frequency corresponding to each modal component, which reflects the main vibration frequency characteristics of the modal component. represents the derivative with respect to time t, which is used to highlight the rate-of-change information of the signal. δ(t) is a time function, j is the imaginary unit, and * represents the convolution operation. Through convolution, the time-frequency domain transformation and feature extraction of the signal can be performed. α is a balance parameter, and its role is to adjust the bandwidth of each modal component and control the fineness of the decomposition result. If the value of α is too large, the bandwidth of each modal component will become narrow, resulting in the loss of signal features. If the value is too small, the decomposition effect will be poor and different modal components cannot be effectively separated. P k is a constraint operator, which is set according to the characteristics of mechanical components and is used to further constrain the decomposition result. Its decomposition logic is:

[0052] Motor component constraint: P1(u1) = |FFT(u1)| f>500Hz = 0. The motor component u1 is processed by the fast Fourier transform (FFT) to force the high-frequency energy with a frequency greater than 500 Hz to zero, because the motor drive component is mainly low-frequency vibration. Through this constraint, the high-frequency interference components in the motor vibration signal can be effectively removed, highlighting the low-frequency vibration characteristics, which is convenient for subsequent fault analysis of the motor drive part.

[0053] Gear component constraint: It represents the integration calculation of the energy of the gear component u2 in the frequency range of 500 Hz to 5 kHz, and it is required that this energy accounts for the total energy E total by a ratio greater than or equal to 80%. This is because the vibration generated by gear transmission is mainly concentrated in this intermediate frequency range. By ensuring the proportion of intermediate frequency energy in the gear component, the working state of the gear can be analyzed more accurately, and whether there are faults such as meshing impact and wear in the gear can be detected.

[0054] Locking component constraint: Among them, max(|u3(t)|) represents the maximum amplitude of the locking component u3 in the time domain, and A th is the amplitude threshold, freq(u3) represents the main frequency of the locking component u3, and [f min , f max is the characteristic frequency range of the vibration signal generated by the impact of the locking mechanism. When the maximum amplitude of the locking component exceeds the set threshold and its main frequency falls within the specific frequency range, P3(u3) takes the value of 1, indicating that an impact event of the locking mechanism is detected; otherwise, it takes the value of 0. This constraint condition can strengthen the detection of the impact event of the locking mechanism and timely discover the faults existing in the locking mechanism.

[0055] After decomposition, the independent motor drive component, gear transmission component, and locking impact component are obtained. Calculate three types of time-frequency domain characteristics of each component: harmonic distortion rate H, modulation sideband ratio M, and impact period variance S. Among them, the harmonic distortion rate That is, for a periodic signal x(t), X n represents the amplitude of the nth harmonic component of the signal x(t), and X1 represents the amplitude of the fundamental wave component of the signal x(t). The harmonic distortion rate H is used to measure the content of harmonic components in the signal and reflects the degree of deviation of the signal from the fundamental wave. In the fault diagnosis of the switch, the harmonic distortion rate of the motor drive component can reflect problems such as whether the motor winding is balanced and whether there are defects in the bearings. If there is a fault in the motor, its harmonic distortion rate will change. By monitoring the change of the harmonic distortion rate, it can be judged whether the operating state of the motor is normal; the modulation sideband ratio That is, for the modulated signal x(t), A s-a represents the amplitude of the a-th modulation sideband component, and A c-j represents the amplitude of the j-th carrier component. The modulation sideband ratio M is used to describe the strength of the modulation phenomenon in the signal. In gear transmission, when there are faults such as gear wear and tooth surface damage, a modulation phenomenon will occur, resulting in a change in the modulation sideband ratio. By monitoring the modulation sideband ratio, the health state of the gear can be judged. For the impact period variance S, by detecting the occurrence times t1, t2,..., t of the impact events in the signaln , calculate the time interval T between adjacent impact events i = t i+1 - t i (i = 1, 2, …, n - 1), and obtain the variance of the impact period Among them, is the average value of the time intervals of all adjacent impact events, n is the number of detected impact events, and the impact period variance S is used to describe the periodic stability of the impact events in the signal. In the locking mechanism, under normal circumstances, the impact period is relatively stable. When the locking mechanism fails (such as lack of oil, component wear, etc.), the impact period will change, and the impact period variance will also change accordingly. By monitoring the impact period variance, the working condition of the locking mechanism can be judged.

[0056] The dynamic baseline modeling module is used to judge whether the operating states of the components of the switch are normal, and establish a dynamic update threshold based on the sliding time window T W = 30 ± ΔT. The operating state of the switch is affected by various factors, such as environmental temperature, the number of switch operations, etc. A fixed threshold cannot adapt to these changes. Through the sliding time window, the changes of these factors are considered in real time to update the threshold and improve the accuracy of fault judgment. The specific threshold settings are as follows: the threshold H of the motor harmonic distortion rate th = 5% + 0.5%·ln(N), where N ∈ Z + represents the cumulative number of switch operations. As the number of switch operations increases, the mechanical components of the motor will gradually wear, resulting in changes in the harmonic distortion rate. Through this formula, the threshold of the motor harmonic distortion rate can be dynamically adjusted according to the actual number of switch operations to more accurately judge whether there is a fault in the motor; the threshold M of the gear modulation sideband ratio th = 0.25 + 0.02(T env - 20), where T env ∈ [-10, 50] °C is the environmental temperature. The environmental temperature affects the lubrication performance of the gear, and thus affects the meshing state of the gear. When the temperature changes, the gear modulation sideband ratio will also change. Through this formula, the threshold of the gear modulation sideband ratio can be dynamically adjusted according to the environmental temperature to improve the accuracy of gear fault judgment; the threshold S of the locking impact period variance th = 0.1 ms 2 is a fixed value, reflecting the influence of mechanical clearance tolerance on the locking impact period.

[0057] The fault decision engine module is based on the eigenvalues (harmonic distortion rate H, modulation sideband ratio M, impact period variance S) of each component calculated by the mechanical transmission chain signal decoupling module, and the thresholds (H th , M th , S th), generate a diagnosis conclusion for the fault type. The logic of the fault diagnosis conclusion is as follows: When the modulation sideband ratio M of the gear transmission component is greater than 1.2M th , and the harmonic distortion rate H of the motor drive component is less than 0.8H th , and at the same time the impact period variance S of the locking impact component is less than 0.08 ms 2 , it is determined as a gear tooth breakage fault. This is because when a gear tooth breaks, it will cause an increase in the gear meshing impact and an increase in the modulation sideband ratio; while the motor drive part is normal, the harmonic distortion rate should be within the range, and when the locking mechanism works normally, the impact period variance is small; when the harmonic distortion rate H of the motor drive component is greater than 1.2H th , the impact period variance S of the locking impact component is greater than 0.15 ms 2 , and the modulation sideband ratio M of the gear transmission component is less than 0.8M th , it is determined as a locking oil shortage fault. This is because the locking oil shortage will cause an increase in the friction force of the locking mechanism and an increase in the motor load, resulting in an increase in the motor harmonic distortion rate and an increase in the locking impact period variance; while if the gear transmission part is normal, the modulation sideband ratio should be within the normal range; when (M > 1.2M th ∪ H > 1.2H th ) and S > 0.15 ms 2 , it is determined as a compound fault, that is, there are problems with the gear, motor, and locking mechanism at the same time.

[0058] The cross-correlation verification module, in order to ensure the effectiveness of the decomposition result of the mechanical transmission chain signal decoupling module, establishes a pure vibration template u of the motor through no-load experiments r (t). This template represents the vibration characteristics of the motor in the normal no-load operation state, and calculates the normalized cross-correlation coefficient ρ of the real-time decomposition result u k (t). The formula is: where T represents the length of the time series, is the mean value of the pure vibration template u of the motor r (t), is the real-time decomposition result u kThe mean value of (t). When ρ ≥ 0.9, it indicates a high correlation between the real-time decomposition result and the pure vibration template of the motor, and the decomposition is effective, allowing for fault determination. This means that the mechanical transmission chain signal decoupling module can accurately separate the motor drive component, and the subsequent fault diagnosis results have high reliability. When 0.8 ≤ ρ < 0.9, it shows a low correlation and signal interference. At this time, the signal resampling operation is triggered to re-collect the vibration signal to obtain more accurate data and ensure the reliability of the decomposition result. When ρ < 0.8, it indicates a poor correlation, which may be due to problems with the sensor. At this time, the sensor calibration instruction is output to calibrate the sensor to ensure the accuracy of the data collected by the sensor, and thus ensure the normal operation of the entire fault diagnosis system.

[0059] In summary, this embodiment details the operation process of the turnout fault diagnosis system based on vibration spectrum analysis. The multi-source vibration sensing module collects the vibration signals of key parts of the turnout. The mechanical transmission chain signal decoupling module decomposes the signals and calculates the characteristic values. The dynamic baseline modeling module dynamically updates the threshold according to factors such as the number of turnout operations and environmental temperature. The fault decision engine module combines the characteristic values and the threshold for fault diagnosis. The cross-correlation verification module ensures the effectiveness of the decomposition result. Each module cooperates closely to achieve accurate diagnosis of turnout faults. This kind of fault diagnosis system can timely detect potential faults of the turnout, reduce misjudgment and missed judgment situations, provide accurate fault information for railway maintenance personnel, and effectively ensure the safe and smooth railway transportation. At the same time, the system has good adaptability and reliability, can operate stably under different working environments and working conditions, and provides strong technical support for the development of the railway industry.

[0060] Embodiment 2

[0061] As Figure 1 shown, based on Embodiment 1, this embodiment details the specific process of a turnout fault diagnosis system based on vibration spectrum analysis when performing turnout fault diagnosis. Through the orderly operation of each link of the system, the turnout fault condition can be comprehensively and accurately analyzed. The specific process is as follows:

[0062] (1) Data acquisition stage

[0063] Sensor installation: The multi-source vibration sensing module installs the triaxial acceleration sensor array on the axial direction of the switch machine housing, the top of the gearbox, and the root of the locking rod respectively. These positions can effectively capture the vibrations generated during motor drive, gear transmission, and locking mechanism operation.

[0064] Signal acquisition: Start the sensor to synchronously collect the three-channel vibration signals of the motor drive, gear transmission, and locking mechanism. During different working stages of the turnout (such as conversion, locking, etc.), the sensor continuously records the vibration signals.

[0065] (2) Signal decoupling stage

[0066] Initial processing: Transmit the collected mixed vibration signal to the mechanical transmission chain signal decoupling module, and decompose the mixed signal into three independent modal components: motor drive, gear transmission, and locking impact.

[0067] Motor component constraint: Process the motor drive component, remove high-frequency interference components, and highlight low-frequency vibration characteristics.

[0068] Gear component constraint: Ensure the energy ratio of the gear transmission component within the frequency range of 500 Hz to 5 kHz to accurately analyze the working state of the gear.

[0069] Locking component constraint: Detect the maximum amplitude and main frequency of the locking impact component. When the amplitude exceeds the set threshold and the main frequency falls within the range, it is determined that an impact event of the locking mechanism is detected.

[0070] Feature calculation: Calculate the time-frequency domain features of the harmonic distortion rate, modulation sideband ratio, and impact period variance of each component.

[0071] (3) Threshold setting stage

[0072] Dynamic update consideration: The dynamic baseline modeling module dynamically updates the threshold based on a sliding time window, considering factors such as the cumulative number of switch operations and environmental temperature that affect the operating state of the switch.

[0073] Motor threshold setting: Set the threshold for the harmonic distortion rate of the motor according to the cumulative number of switch operations.

[0074] Gear threshold setting: Set the corresponding threshold for the gear modulation sideband ratio based on the environmental temperature.

[0075] Locking threshold setting: Determine the fixed threshold for the impact period variance of the locking impact.

[0076] (4) Fault diagnosis stage

[0077] Rule judgment: The fault decision engine module makes a judgment according to the characteristic values of each component and the dynamically updated threshold, in accordance with the set fault diagnosis rules.

[0078] Fault type determination: As shown in Embodiment 1, if the modulation sideband ratio of the gear transmission component, the harmonic distortion rate of the motor drive component, and the impact period variance of the locking impact component meet the constraint conditions, it is determined as a gear tooth breakage fault; when the harmonic distortion rate of the motor drive component, the impact period variance of the locking impact component, and the modulation sideband ratio of the gear transmission component meet the corresponding conditions, it is determined as a locking oil shortage fault; when the modulation sideband ratio of the gear transmission component or the harmonic distortion rate of the motor drive component, and the impact period variance of the locking impact component meet the conditions, it is determined as a compound fault.

[0079] (5) Verification and calibration phase

[0080] Template establishment: Establish a pure vibration template of the motor through the cross - correlation verification module. This template represents the vibration characteristics of the motor during normal no - load operation.

[0081] Correlation calculation: Calculate the normalized cross - correlation coefficient between the real - time decomposition result and the pure vibration template of the motor to evaluate the effectiveness of the decomposition result.

[0082] Handling of different situations: Judge the correlation coefficient, perform cross - correlation verification between the real - time decomposition result and the pure drive component template, and output feedback to the multi - source vibration sensing module.

[0083] The above are only the preferred embodiments of the present invention and do not impose any form of limitation on the present invention. Although the present invention has been disclosed above with preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments by using the above - disclosed technical content within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A turnout fault diagnosis system based on vibration spectrum analysis, characterized in that: The system consists of: multi-source vibration sensing module, mechanical transmission chain signal decoupling module, dynamic baseline modeling module, fault decision engine module, and cross-correlation verification module; The multi-source vibration sensing module is composed of a triaxial acceleration sensor array arranged in the axial direction of the switch housing, the top of the gear box, and the root of the locking rod, and synchronously collects three-channel vibration signals of the motor drive, gear transmission, and locking mechanism; The mechanical transmission chain signal decoupling module physically decomposes the vibration signal into motor drive component, gear transmission component, and locking impact component, and calculates three types of time-frequency domain features of each component: harmonic distortion rate H, modulation sideband ratio M, and impact period variance S; The dynamic baseline modeling module uses a sliding time window to dynamically update the energy proportion baseline threshold H of the three types of components. th 、M th , S th ; The fault decision engine module generates a fault type diagnosis conclusion according to the excessive combination of the three types of characteristic values, and the fault type diagnosis conclusion includes a gear tooth breakage fault, a locking oil shortage fault, and a compound fault; The cross-correlation verification module determines the validity of the decomposition by comparing the cross-correlation function of the motor drive component decomposed in real time with the preset no-load test template.

2. A turnout fault diagnosis system based on vibration spectrum analysis according to claim 1, characterized in that: The mechanical transmission chain signal decoupling module synchronously collects three-channel vibration signals, and the physical correspondence of the three-channel vibration signals is: Motor drive component: used to reflect the low-frequency vibration caused by motor winding imbalance and bearing defects, with a frequency range of 0-500Hz, to determine the potential faults of the motor; Gear transmission component: medium-frequency vibration indicating gear meshing impact and wear, with a frequency range of 500Hz-5kHz. This component is used to monitor the health of the gear; Locking impact component: used to capture the high-frequency vibration generated by the locking mechanism at the moment of collision, with a frequency greater than 5kHz, to determine the working condition of the locking mechanism.

3. A turnout fault diagnosis system based on vibration spectrum analysis according to claim 2, characterized in that: The mechanical transmission chain signal decoupling module decomposes the mixed vibration signal collected by the multi-source vibration sensing module by minimizing the function value to achieve mathematical decomposition of the mixed vibration signal into K independent modal components {u k } and K = 3, that is, K corresponds to the three components of motor drive, gear transmission, and locking impact, and the center frequency {ω k }, the minimization function value satisfies: Among them, {u k } represents the modal components obtained by decomposition, u k (t) represents the real-time decomposition result, {ω k } represents the center frequency corresponding to each modal component, represents the derivative with respect to time t, δ(t) is the time function, j is the imaginary unit, * represents the convolution operation, α is the balance parameter used to adjust the bandwidth of each modal component, P k It is a constraint operator, which is set according to the characteristics of the mechanical parts. Specifically: Motor component constraint: P1(u1) = |FFT(u1)| f>500Hz =0, indicating that the motor component u1 is processed by fast Fourier transform FFT, forcing the high-frequency energy with a frequency greater than 500Hz to zero, which is used to detect the low-frequency vibration of the motor drive component; Gear component constraints: It means that the energy of the gear component u2 in the frequency range of 500Hz to 5kHz is integrated so that this energy accounts for the total energy E total The proportion of is greater than or equal to 80%, so as to ensure the proportion of medium frequency energy in the gear component; Locking component constraints: Among them, max(|u3(t)|) represents the maximum amplitude of the locking component u3 in the time domain, A th is the amplitude threshold and A th =0.5g, g is the acceleration due to gravity, freq(u3) represents the main frequency of the locking component u3, [f min ,f max ] is the characteristic frequency range of the vibration signal generated by the impact of the locking mechanism, which is used to enhance the detection of impact events of the locking mechanism.

4. A turnout fault diagnosis system based on vibration spectrum analysis according to claim 1, characterized in that: The mechanical transmission chain signal decoupling module calculates three types of time-frequency domain features of each component, namely, harmonic distortion rate H, modulation sideband ratio M, and impulse cycle variance S, where: The harmonic distortion rate H is used to measure the content of harmonic components in the signal, reflecting the degree to which the signal deviates from the fundamental wave. For a periodic signal x(t), its harmonic distortion rate X n represents the amplitude of the nth harmonic component of the signal x(t), and X1 represents the amplitude of the fundamental component of the signal x(t), that is, the normal operating frequency of the equipment; The modulation sideband ratio M is used to describe the strength of the modulation phenomenon in the signal. The modulation sideband is the frequency component generated on both sides of the fundamental frequency due to the modulation of the signal. That is, for the modulated signal x(t), its modulation sideband ratio A s-a represents the amplitude of the ath modulation sideband component, A c-j It indicates that the amplitude of the jth carrier component corresponds to the meshing frequency of the gear, N is the number of modulation sidebands, and m is the number of carrier components; The shock period variance S is used to describe the periodic stability of shock events in the signal. For a signal x(t) containing shock events, the occurrence time t1, t2, …, t n , calculate the time interval T between adjacent impact events i =t i+1 -t i , i=1,2,…,n-1, and the shock period variance is obtained T i is the time interval between the ith adjacent shock events, is the average time interval between all adjacent impact events, that is, n is the number of impact events detected.

5. A turnout fault diagnosis system based on vibration spectrum analysis according to claim 4, characterized in that: The dynamic baseline modeling module is based on a sliding time window T W =30±ΔT Dynamic update threshold, the specific threshold setting is: The motor harmonic distortion rate threshold: H th =5%+0.5%·ln(N), where N∈Z + Represents the cumulative number of turnout actions; The gear modulation sideband ratio threshold: M th =0.25+0.02(T env -20), where T env ∈[-10,50]℃ is the ambient temperature; Locking shock cycle variance threshold: S th =0.1ms 2 Is a fixed value.

6. A turnout fault diagnosis system based on vibration spectrum analysis according to claim 1, characterized in that: The fault diagnosis conclusion logic in the fault decision engine module is:

7. A turnout fault diagnosis system based on vibration spectrum analysis according to claim 1, characterized in that: The cross-correlation verification module establishes a pure vibration template u of the motor by no-load r (t), calculate the real-time decomposition result u k The normalized correlation coefficient ρ of (t) is Where T represents the length of the time series, It is the pure vibration template of the motor r The mean of (t), is the real-time decomposition result u k The mean of (t); When ρ≥0.9, the real-time decomposition result has a high correlation with the pure vibration template of the motor, the decomposition is effective, and fault judgment is allowed; When 0.8≤ρ<0.9, it indicates that the correlation is low and there is signal interference, which triggers the signal resampling operation; When ρ<0.8, it indicates poor correlation, and a sensor calibration instruction is output.

8. A turnout fault diagnosis method based on vibration spectrum analysis, applicable to a turnout fault diagnosis system based on vibration spectrum analysis as claimed in any one of claims 1 to 7, characterized in that: The specific steps of this method are: S100, using a multi-source vibration sensing module to synchronously collect three-channel vibration signals of a motor, a gear, and a locking mechanism through a three-axis acceleration sensor array; S200, using the mechanical transmission chain signal decoupling module to decompose the signal and calculate the harmonic distortion rate, modulation sideband ratio, and impact cycle variance; S300, using a dynamic baseline modeling module, dynamically updating the energy percentage baseline thresholds of the three types of components using a sliding time window; S400, generating a fault type diagnosis result through a fault decision engine module according to a comparison between a combination of characteristic value exceeding the standard and a threshold value; S500, using a cross-correlation verification module, cross-correlation verification is performed between the real-time decomposition result and the pure driving component template.

Images