A method and system for positioning a metro train

Abstract

The application relates to a subway train positioning method and system, the system comprising a processor configured to: process multi-source data of a train collected by an inertial navigation unit based on an error Kalman filtering algorithm to preliminarily correct acceleration and angular velocity; construct a time-mile curve of the train and correct the time-mile curve based on a first correction ratio; identify dynamic change characteristics of a turn based on the angular velocity collected by the inertial navigation unit, calculate total mileage in the turn section, and correct the total mileage in the turn section based on a second correction ratio of the turn section; and identify abnormal vibration characteristics based on high-order change rates of the acceleration and angular velocity of the train and mark the abnormal vibration characteristics in a visual manner. Without relying on GPS, the application collects acceleration and angular velocity data of the train in the running and calibrates the data to construct a mileage calibration model based on track characteristics, so that accurate train mileage and position information can be obtained in a GPS-free environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of urban rail detection, and in particular to a subway train positioning method and system. BACKGROUND

[0002] Currently, the positioning system for smoothness detection of rail transit mainly relies on GPS or Beidou. However, in the underground environment such as subway trains, these satellite positioning systems often cannot work normally due to signal shielding, resulting in the inability to obtain the accurate position of the train. In order to solve this problem, transponders and positioning tags are also used as auxiliary positioning devices in existing systems. The transponders are installed along the track, and the train transmits position information through short-range communication when passing through, while the positioning tags provide auxiliary positioning by sensing the movement of the train. However, the installation and maintenance of these external devices are costly, and the subway environment has strict restrictions on device arrangement, making it difficult to meet the comprehensive and long-term application requirements.

[0003] Therefore, the current smoothness detection instrument usually uses a pure inertial navigation system, that is, an inertial navigation unit (INS) is carried on the train to perform inertial navigation calculation, so as to estimate the motion mileage in the absence of GPS. However, pure inertial navigation also has the problem of error accumulation, making it difficult to maintain the accuracy of the position after a long time of running.

[0004] Moreover, in the subway rail transit system, long-time running of the train can cause rail wear and track geometry changes, thereby causing abnormal vibration of the train. Such abnormal vibration not only affects the comfort of passengers, but also may pose a potential threat to the safe operation of the train. Therefore, how to accurately detect and locate the abnormal vibration area of the train in the track running has become an important technical requirement for subway safety maintenance. However, the existing positioning system faces the problems of signal limitation and error accumulation in the subway environment, making it difficult to achieve accurate monitoring and precise positioning of the smoothness of the train running.

[0005] The current technology for train positioning mainly includes GPS and inertial navigation unit (INS) fusion positioning, UWB positioning, or GPS and differential GPS system (DGPS) and inertial navigation unit (INS).

[0006] Currently, most smoothness detection instruments use the fusion positioning method of GPS or Beidou and inertial navigation unit (INS) system, but in the underground environment such as subway trains, GPS and Beidou signals are limited, resulting in the inability to position and the difficulty in accurately identifying the position of abnormal vibration occurrence. Therefore, the effect of this system in the subway is limited.

[0007] Another positioning method for underground trains such as subway trains is real-time correction through a UWB or differential GPS system, but such a system requires the installation of a base station, resulting in high installation and maintenance costs, and certain subway line environments do not allow the addition of external equipment, so it is not suitable for positioning needs during subway train operation.

[0008] Pure inertial navigation is not dependent on external equipment and is suitable for signal shielding environments such as subway trains, but the main problem with this approach is error accumulation. As time goes on, the error accumulation effect becomes more and more pronounced, and generally within 10 minutes, the mileage error can accumulate to about 15%. In the case of high-precision track detection requirements, the pure inertial navigation solution is difficult to meet the accuracy requirements for long-term operation.

[0009] For example, CN107976697A discloses a train safety positioning method and system based on a combination of Beidou / GPS, which includes: determining the current track of the train by reading the transponders configured on the track, checking the accuracy of satellite positioning, and calibrating the GNSS module and the inertial navigation combination module; obtaining the travel direction of the current train by reading the two groups of transponders through which the current train continuously passes, or by traveling a certain opening speed for a distance, obtaining satellite positioning information and train inertial navigation information through the GNSS module and the inertial navigation combination module, and thus analyzing the travel direction of the train; after determining the current track and travel direction of the train, continuously obtaining satellite positioning information and train inertial navigation information of the current train through the GNSS module and the inertial navigation combination module, and determining whether there is false or incorrect satellite signal through a predetermined method; if so, combining the ground differential station system to select the correct positioning method of the GNSS module and the inertial navigation combination module; finally, setting the tail safety boundary of the current train as the last triggered electronic fence, extending the tail safety boundary by the length of the train to the position of the next electronic fence to obtain the head safety boundary of the train; the head safety boundary and the tail safety boundary together form the safety footprint of the entire train, thereby forming train positioning. This technical solution belongs to a typical GPS and inertial navigation unit (INS) fusion positioning scheme, so in underground environments, the positioning signal is easy to be unclear, not continuous, and leads to positioning failure, making it difficult to accurately identify the location of abnormal vibration.

[0010] Therefore, the present application hopes to provide a subway train positioning method and system that can use only an inertial navigation system to achieve accurate positioning of a subway train in underground environments where GPS signals or Beidou satellite signals are inaccurate.

[0011] In addition, on the one hand, due to the difference in understanding of those skilled in the art; on the other hand, due to the fact that the applicant has studied a large number of literatures and patents when making the present application, but due to the limitation of space, all the details and contents are not listed in detail, but this does not mean that the present application does not have these prior art characteristics. On the contrary, the present application already has all the characteristics of the prior art, and the applicant reserves the right to add relevant prior art in the background art. SUMMARY

[0012] In the subway rail transit system, long-time operation of the train will cause rail wear and track geometry changes, thereby causing abnormal vibration of the train. Such abnormal vibration not only affects the comfort of passengers, but also may pose a potential threat to the safe operation of the train. Therefore, accurately detecting and locating the abnormal vibration area of the train in the track operation has become an important technical requirement for the safety maintenance of the subway. However, the existing positioning system faces the problems of signal limitation and error accumulation in the subway environment, which makes it difficult to realize accurate monitoring and precise positioning of the train operation stability.

[0013] Therefore, the present application proposes a subway train positioning system, which is expected to solve the following problems:

[0014] First, the positioning problem in the GPS-free environment. Since the subway train is mostly operated underground, the signals of satellite positioning systems such as GPS and Beidou cannot be covered, resulting in the loss of positioning function of traditional stability detection equipment in underground environment. How to accurately position the train operation without external signal support has become one of the technical difficulties.

[0015] Second, the error accumulation problem in the pure inertial navigation system. The inertial navigation unit (INS) can measure the train motion information, but the error of the inertial navigation system will gradually accumulate during long-time operation, which is difficult to meet the positioning accuracy requirement.

[0016] Third, the identification and mileage calibration problem of the rail feature points. The present application uses the turning points, stopping points and acceleration mutation points at the turnout of the subway track as features for mileage reference. Therefore, how to use the natural feature points of the track to effectively calibrate the inertial navigation system is also a problem to be solved by the present application, which involves two difficulties: (1) identifying the track feature points from the data of the inertial navigation unit (INS); (2) using the mileage data of the feature points to obtain the calibrated mileage data after data processing.

[0017] In view of the deficiencies of the prior art, the present application provides, from a first aspect, a subway train positioning system, comprising a processor configured to: process multi-source data of a train collected by an inertial navigation unit based on an error Kalman filtering algorithm to preliminarily correct acceleration and angular velocity; construct a time-mileage curve of the train and correct the time-mileage curve based on a first correction ratio; identify dynamic change characteristics of a turn based on angular velocity collected by the inertial navigation unit, wherein a total mileage in a turn section is calculated, and the total mileage in the turn section is corrected based on a second correction ratio of the turn section; and identify abnormal vibration characteristics based on high-order change rates of acceleration and angular velocity of the train and mark the abnormal vibration characteristics in a visual manner.

[0018] Without relying on GPS, the present application constructs a mileage calibration model based on track characteristics by collecting acceleration and angular velocity data of the train in the running and calibrating the processed data, thereby ensuring accurate train mileage and position information in a GPS-free environment.

[0019] According to a preferred embodiment, the processor is further configured to: before constructing the time-mileage curve of the train, perform data filtering and smoothing processing on the preliminarily corrected acceleration and angular velocity to remove noise in the data. In the data measurement process of the inertial measurement unit, due to environmental interference, vibration and sensor self-noise, the original acceleration, angular velocity, speed and position information usually contain high-frequency noise. These noises will affect subsequent turn detection, mileage calculation and position estimation. Therefore, the present application adopts smoothing processing and filtering technology to suppress noise, thereby ensuring the stability and accuracy of the data.

[0020] According to a preferred embodiment, the processor is configured to preliminarily correct the acceleration and angular velocity in the following manner: in the initial stage of running of the processor, predict the next time state of the train based on a preset motion model and acceleration and angular velocity of the train collected by the inertial navigation unit; correct the predicted state information based on Kalman gain and update an error covariance model related to the prediction, thereby correcting the predicted state of the train. The error Kalman filtering performs state estimation on the original acceleration and angular velocity data collected by the inertial measurement unit, dynamically corrects measurement noise and system error by combining the motion model of the system and the actual measurement value of the inertial measurement unit. Through real-time estimation of the noise of the inertial measurement unit, the error Kalman filtering algorithm can effectively reduce the drift and random error in the data of the inertial measurement unit, thereby providing more accurate basic data for subsequent data processing.

[0021] According to a preferred embodiment, the processor is configured to correct the time-mileage curve in the following manner: convert the acceleration and angular velocity into position and velocity to construct the time-mileage curve; take the ratio of the mileage to the first target mileage as a first correction ratio, and scale the mileage data at each time point proportionally based on the first correction ratio to obtain corrected mileage, and correct the time-mileage curve. The time-mileage curve is used to visually display the mileage change of the train.

[0022] According to a preferred embodiment, the processor is configured to identify the dynamic change feature of the turning in the following manner: perform nonlinear fitting on the angular velocity curve, calculate the rate of change of the angular velocity and the angular acceleration, set a dynamic threshold based on the current speed of the train and the curve radius of the track, and determine the start time in the turning section of the train when the angular velocity is greater than the dynamic threshold; determine the end time in the turning section of the train based on the rate of change of the angular velocity and the negative change of the angular acceleration. High-order analysis is performed on the angular velocity, the rate of change of the angular velocity and the second derivative are used to capture the slight changes of the start time and the end time in the turning section, and the dynamic threshold is dynamically adjusted to adapt to different turning radii and turning speeds.

[0023] According to a preferred embodiment, the processor is configured to correct the total mileage in the turning section in the following manner: calculate the total mileage in the turning section based on the speed of the train; take the ratio of the total mileage in the turning section to the second target distance in the turning section as a second correction ratio; and scale the total mileage in the turning section proportionally based on the second correction ratio. The corrected mileage in the turning section of the present application is consistent with the input mileage marker, and the scaling ratio is dynamically adjusted to match the real-time changing speed.

[0024] According to a preferred embodiment, the processor is configured to correct the total mileage in the turning section in the following manner: adopt a transition function to smooth the corrected total mileage in the turning section, so that the corrected time-mileage curve is continuous and smoothly connected at the calibration point. In this way, the discontinuity of the corrected time-mileage curve at the calibration point can be avoided, and the time-mileage curve is smoothly connected before and after the calibration point, avoiding sudden changes.

[0025] According to a preferred embodiment, the processor is configured to identify the abnormal vibration feature in the following manner: calculate the third derivative functions of the acceleration and angular velocity of the train respectively, and the absolute value of the third derivative function is the vibration intensity; dynamically adjust the identification threshold of the vibration point based on the speed of the train; and the position corresponding to the vibration intensity greater than the identification threshold is the abnormal vibration point.

[0026] The present application determines the size and color of the visualized vibration point based on the vibration intensity, thereby representing the severity of the vibration, so that the user can see the specific information of each vibration point at a glance.

[0027] The method provided by the present application has the advantages of improving positioning accuracy, reducing error accumulation, combining an inertial navigation system with metro track feature points, and correcting data of an inertial navigation unit (INS) by using an error Kalman filtering algorithm, so that high-precision positioning is achieved in a GPS-free environment, and errors are controlled within 5%. In addition, the present application is free of dependence on external signals, and has low cost. Compared with existing GPS and UWB schemes, the present application does not need to depend on external signals or install additional base station equipment, eliminates signal limitation problems of GPS and Beidou in underground environments, avoids high installation and maintenance costs of base stations, and has good economy and operability. The present application also improves the convenience and effectiveness of metro track detection. The positioning and smoothness detector used in the present application has high integration, and can generate a time-mileage curve of a train closer to the real one, so as to facilitate accurate detection and maintenance of metro track abnormal points in a GPS-free environment, and greatly improve the accuracy and effectiveness of train smoothness detection.

[0028] The method provided by the present application has the advantages of improving positioning accuracy, reducing error accumulation, combining an inertial navigation system with metro track feature points, and correcting data of an inertial navigation unit (INS) by using an error Kalman filtering algorithm, so that high-precision positioning is achieved in a GPS-free environment, and errors are controlled within 5%. In addition, the present application is free of dependence on external signals, and has low cost. Compared with existing GPS and UWB schemes, the present application does not need to depend on external signals or install additional base station equipment, eliminates signal limitation problems of GPS and Beidou in underground environments, avoids high installation and maintenance costs of base stations, and has good economy and operability. The present application also improves the convenience and effectiveness of metro track detection. The positioning and smoothness detector used in the present application has high integration, and can generate a time-mileage curve of a train closer to the real one, so as to facilitate accurate detection and maintenance of metro track abnormal points in a GPS-free environment, and greatly improve the accuracy and effectiveness of train smoothness detection.

[0029] According to a preferred embodiment, the method further comprises: before constructing the time-mileage curve of the train, performing data filtering and smoothing processing on the preliminarily corrected acceleration and angular velocity, so as to remove noise in the data.

[0030] In the data measurement process of the inertial measurement unit, due to environmental interference, vibration and sensor self-noise, high-frequency noise is usually contained in the original acceleration, angular velocity, speed and position information. These noises will affect subsequent turn detection, mileage calculation and position estimation. Therefore, the present application uses smoothing processing and filtering technology to suppress noise and ensure data stability and accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 is a schematic diagram of a module connection relationship of a metro train positioning system provided by the present application;

[0032] Figure 2 is a flowchart of a metro train positioning method provided by the present application;

[0033] Figure 3 is a subway track line map provided by the present application;

[0034] Figure 4 is a direction diagram of IMU (Inertial Measurement Unit) planed on a train provided by the present application;

[0035] Figure 5 is a diagram of angular velocity variation provided by the present application;

[0036] Figure 6 is a display diagram of operation test data of a subway line provided by the present application;

[0037] Figure 7 is a comparison diagram of mileage-time curves before and after correction provided by the present application;

[0038] Figure 8 is a curve diagram of absolute speed variation with time provided by the present application;

[0039] Figure 9 is a test diagram of total mileage data of preliminary correction in example data provided by the present application;

[0040] Figure 10 is a test diagram of calculated total mileage data in example data provided by the present application;

[0041] Figure 11 is a table diagram of example data obtained by test of a subway train positioning system provided by the present application.

[0042] List of reference signs

[0043] 100: processor; 110: initial correction module; 120: curve generation module; 130: turn identification module; 140: abnormal vibration identification module; 200: smoothness detector; 300: memory. DETAILED DESCRIPTION

[0044] The present application will be described in detail below with reference to the accompanying drawings.

[0045] Some noun terms of the present application are explained.

[0046] Stability detection system: A data acquisition hardware system that integrates an inertial measurement unit (IMU), a voltage reduction module, a single-chip microcomputer, and a power supply. It is designed to measure and analyze the running stability of vehicles, especially trains. The stability detection system captures the acceleration and attitude of the train through the inertial measurement unit (IMU), ensures voltage stability through the voltage reduction module, and processes signals from sensors and performs control tasks through the single-chip microcomputer. In addition, the stability detection system also communicates with the host computer, which is used to process the data collected by the stability detection system and conduct in-depth analysis, including data trend analysis, anomaly detection, and report generation.

[0047] The core of the stability detection system also includes a mileage calibration algorithm based on absolute mileage and speed. This algorithm calculates the arithmetic mean of the running speed of multiple sensor positions to determine the running speed of the train and eliminates abnormal speed values to improve measurement accuracy. This algorithm helps to reduce calculation errors and ensures the measurement accuracy of the train's running speed, providing important data support and analysis tools for the stability and safety of the train.

[0048] IMU: Inertial Measurement Unit. IMU is an electronic device that measures and reports the three basic linear motions (acceleration) and three basic angular motions (angular velocity) of an object.

[0049] 6-axis inertial measurement unit (IMU): A sensor system that integrates three acceleration axes (X, Y, Z) and three angular velocity axes (X, Y, Z), capable of measuring the motion state of an object in three-dimensional space. The acceleration axis is responsible for capturing the linear acceleration of the object in three mutually perpendicular directions, while the angular velocity axis measures the rotation rate of the object around these three axes. This design allows the 6-axis inertial measurement unit (IMU) to provide accurate data on the speed, position, and attitude of the object.

[0050] Multi-source data: Includes three-axis acceleration and three-axis angular velocity (angle) collected by the inertial measurement unit (IMU), real mileage between two stations, and turn point mileage information.

[0051] Error Kalman filtering algorithm: Used to process data collected by the inertial measurement unit (IMU). The error Kalman filtering algorithm is an application of the Kalman filtering algorithm in processing measurement data with errors. The Kalman filtering algorithm is an optimal estimation algorithm for estimating the state of a dynamic system, especially suitable for situations with noise and uncertainty.

[0052] In practical applications, observation data is often affected by various noises, which can be random or systematic. The error Kalman filter algorithm describes the statistical properties of these errors, such as their mean and covariance, through a mathematical model, and then uses this information to optimize state estimation and reduce the impact of errors.

[0053] Embodiment 1

[0054] Current technologies for train positioning mainly include GPS and inertial navigation unit (INS) fusion positioning, UWB positioning, or GPS and differential GPS system (DGPS) and inertial navigation unit (INS).

[0055] Currently, most of the stationarity detectors 200 use a fusion positioning method of GPS or Beidou and inertial navigation unit (INS), but in underground environments such as subway trains, GPS and Beidou signals are limited, resulting in positioning failure and difficulty in accurately identifying the location of abnormal vibration. Therefore, the effect of this system in the subway is limited.

[0056] Another positioning method for subway trains is real-time correction by UWB or differential GPS system, but such systems require the installation of base stations, resulting in high installation and maintenance costs, and certain subway line environments do not allow the addition of external equipment, so they are not suitable for positioning needs during subway train operation.

[0057] Pure inertial navigation does not rely on external equipment and is suitable for signal shielding environments such as subway trains, but the main problem with this method is error accumulation. As time goes on, the error accumulation effect becomes more and more obvious, and generally, the mileage error can accumulate to about 15% within 10 minutes. In the demand for high-precision track detection, the pure inertial navigation scheme is difficult to meet the accuracy requirements for long-time operation.

[0058] The present application provides a subway train positioning method and system, especially a subway train positioning method and system based on error Kalman filter algorithm and multi-source data mileage correction. The present application can also provide an electronic device for subway train positioning. The present application can also provide a storage medium storing an encoding program of the subway train positioning method. The present application can also provide a processor 100 of an encoding program of the subway train positioning method. The present application can also provide a terminal that can be connected to the processor 100. The present application can also provide a subway control system, which contains the subway train positioning system of the present application or executes the subway train positioning method of the present application. Preferably, the present application can also provide a subway, which is installed with the subway train positioning system of the present application or the electronic device executing the subway train positioning method of the present application.

[0059] The present application hopes to solve the following problems:

[0060] First, the positioning problem in the no-GPS environment. Since the subway train is mostly operated underground, the signals of satellite positioning systems such as GPS and Beidou cannot be covered, resulting in that the traditional smoothness detection equipment loses the positioning function in the underground environment. How to perform high-precision positioning on the train operation without external signal support becomes one of the technical difficulties.

[0061] Second, the error accumulation problem in the pure inertial navigation system. The inertial navigation unit (INS) can measure and calculate the train motion information, but the error of the inertial navigation system will gradually accumulate in the long-time operation process, which is difficult to meet the positioning accuracy requirement.

[0062] Third, the identification and mileage calibration problem of the rail feature points. The turning points, parking points and acceleration mutation points in the subway track and other features are used as mileage references. Therefore, how to use the natural feature points of the track to effectively calibrate the inertial navigation system is also a problem to be solved by the present application, which involves two difficulties: (1) identifying the track feature points from the data of the inertial navigation unit (INS); (2) using the mileage data of the feature points to obtain the calibrated mileage data after data processing.

[0063] In view of the deficiencies of the prior art, the present application provides a subway train positioning system, as shown in the accompanying drawings, comprising a processor 100. The processor 100 is configured with multiple modules, including an initial correction module 110, a curve generation module 120, a turning identification module 130 and an abnormal vibration identification module 140. Figure 1

[0064] Preferably, when the processor 100 is an independent processor, the processor 100 can be a CPU, a GPU and a combination thereof to form hardware, and can also be a special integrated chip capable of executing the subway train positioning method of the present application. In this case, the initial correction module 110, the curve generation module 120, the turning identification module 130 and the abnormal vibration identification module 140 are different execution task modules inside the processor 100.

[0065] Preferably, when the processor 100 is a non-independent processor, it can be composed of multiple special integrated chips and / or processing hardware (CPU and / or GPU, storage module) in communication connection. The initial correction module 110, the curve generation module 120, the turning identification module 130 and the abnormal vibration identification module 140 are respectively independent processors or at least two module sets in the same processor 100. Preferably, different independent processors are connected through communication lines.

[0066] Preferably, the processor 100 can be connected with a terminal for receiving user input data and displaying the processing process and processing result data of the processor 100. ​

[0067] Preferably, as shown in Figure 1 The processor 100 of the present application is in communication connection with an inertial measurement unit (IMU) of the train for receiving multi-source data of the train collected by the inertial measurement unit (IMU).

[0068] Preferably, the subway train positioning system of the present application can comprise the processor 100 and the inertial measurement unit (IMU) or inertial navigation unit (INS). Further preferably, the subway train positioning system of the present application can comprise the processor 100, the inertial measurement unit and the memory 300.

[0069] The subway train of the present application is referred to as the train hereinafter.

[0070] Preferably, the smoothness detector 200 on the train is equipped with an inertial measurement unit (IMU), a voltage reduction module, a power supply and an embedded chip. Alternatively, the smoothness detector 200 on the train is equipped with an inertial navigation unit (INS), a voltage reduction module, a power supply and an embedded chip.

[0071] The inertial measurement unit (IMU) or inertial navigation unit (INS) collects high-frequency signals, and the embedded chip uses an error Kalman filtering algorithm to process the collected data, thereby reducing the random errors of the inertial measurement unit (IMU) or inertial navigation unit (INS) and improving the accuracy of the data. The characteristic points of the subway track, such as the track curve change when the train turns, the parking time and position of the station, can all be used as key reference points for mileage calibration: the position of the turn and the start or end time of the turn section have obvious characteristics and are easy to collect, and the parking time and position provide a clear mileage calibration reference.

[0072] As shown in Figure 2 The processor 100 of the present application is configured to perform the following steps.

[0073] S100: The initial correction module 110 processes the multi-source data of the train collected by the inertial navigation unit based on the error Kalman filtering algorithm to preliminarily correct the acceleration and angular velocity.

[0074] The error Kalman filtering performs state estimation on the original acceleration and angular velocity data collected by the inertial measurement unit (IMU), and dynamically corrects the measurement noise and system error by combining the motion model of the system and the actual measurement value of the inertial measurement unit (IMU). Through real-time estimation of the noise of the inertial measurement unit (IMU), the error Kalman filtering algorithm can effectively reduce the drift and random error in the data of the inertial measurement unit (IMU), thereby providing more accurate basic data for subsequent data processing.

[0075] S200: The curve generation module 120 constructs a time-mileage curve of the train and corrects the time-mileage curve based on a first correction ratio.

[0076] S300: The turn identification module 130 identifies the dynamic change characteristics of the turn based on the angular velocity collected by the inertial navigation unit. Preferably, the turn identification module 130 calculates the total mileage within the turn section and corrects the total mileage within the turn section based on a second correction ratio of the turn section.

[0077] S400: The abnormal vibration identification module 140 identifies the abnormal vibration characteristics based on the high-order change rate of the acceleration and angular velocity of the train and marks in a visualized manner.

[0078] The present application constructs a mileage calibration model based on track characteristics without relying on GPS, by collecting the acceleration and angular velocity data of the train in the running of these characteristics and calibrating the processed data, to ensure that the accurate train mileage and position information is obtained in the GPS-free environment.

[0079] The present application details the steps of step S100.

[0080] S110: State initialization.

[0081] In the initial stage of the system operation of the present application, the state vector [x0] is set according to the initial acceleration, angular velocity and position data of the inertial measurement unit (IMU). The state error covariance matrix [P0] is initialized according to the characteristics of the inertial measurement unit (IMU). The state error covariance matrix includes several initial state vectors. The initial state vector can include the initial position, speed and attitude of the train.

[0082] S120: Predict state (time update).

[0083] The next time state of the train is predicted based on the preset motion model and the acceleration and angular velocity of the train collected by the inertial navigation unit.

[0084] The time update formula is:

[0085] wherein, represents the predicted state, x k-1 represents the last predicted state, A represents the state transition matrix, u k represents the input data (such as acceleration and angular velocity) measured by the inertial measurement unit (IMU), and B represents the input matrix.

[0086] S130: Predict error covariance.

[0087] The error covariance matrix of the predicted state is:

[0088] wherein A T denotes the transpose of the state transition matrix; Q denotes the process noise covariance matrix, used to describe the uncertainty in the system model.

[0089] S140: Correct the predicted state.

[0090] Based on the Kalman gain, the predicted state information is corrected, and the error covariance model related to prediction is updated.

[0091] The predicted state is corrected according to the acceleration and angular velocity data collected by the inertial measurement unit (IMU).

[0092] First, the Kalman gain K k :

[0093] wherein H T denotes the measurement matrix, R denotes the measurement noise covariance matrix, and H denotes the state observation matrix.

[0094] The predicted state and the error covariance are corrected using the Kalman gain:

[0095]

[0096] wherein z k denotes the actual measurement value of the inertial measurement unit (IMU), and x k denotes the corrected state estimate.

[0097] The error covariance is updated:

[0098] S150: Repeat the prediction and update steps.

[0099] The Kalman filter performs prediction and correction (update) at each time, continuously reducing the impact of noise and drift on system state estimation through a recursive process. This filtering process continues until the data processing is completed.

[0100] After the error Kalman filter processing of the inertial measurement unit (IMU) data, the system obtains the acceleration and angular velocity data that have been preliminarily error corrected. Subsequently, these filtered inertial measurement unit (IMU) data are further smoothed and integrated to generate the train's time-mileage curve, and combined with the track feature points for mileage calibration, finally realizing high-precision train running trajectory and abnormal vibration detection.

[0101] Preferably, the execution steps of the processor 100 can further include S160. S160 is executed between steps S100 and S200.

[0102] S160: Before constructing the time-mileage curve of the train, the data of the preliminary corrected acceleration and angular velocity are filtered and smoothed to remove noise in the data.

[0103] During the data measurement process of the inertial measurement unit (IMU), due to environmental interference, vibration and sensor self-noise, high-frequency noise is usually contained in the original acceleration, angular velocity, speed and position information. These noises will affect the subsequent turn detection, mileage calculation and position estimation. Therefore, the present application adopts smoothing and filtering technology to suppress noise and ensure the stability and accuracy of the data.

[0104] The present application will now describe step S160 as follows.

[0105] S161: Low-pass filtering is performed.

[0106] The data of acceleration and angular velocity are often affected by high-frequency noise.

[0107] The acceleration is represented as:

[0108] The angular velocity is represented as:

[0109] Therefore, low-pass filtering is performed on the acceleration and angular velocity signals to eliminate high-frequency noise.

[0110] Preferably, a Butterworth filter is used for low-pass filtering. The characteristic of low-pass filtering is that the frequency response is smooth, which can effectively filter out high-frequency noise without introducing phase shift. Assuming that the filter is 4th order and the filter cutoff frequency is ω c , the transfer function of the filter is:

[0111]

[0112] This transfer function describes how the filter responds to signals of different frequencies.

[0113] where s represents a complex frequency variable, ζ represents a damping coefficient, and ω c represents the cutoff frequency of the filter. For example, the ζ value of the Butterworth filter is fixed, and ζ=0.707 is usually selected.

[0114] Low-frequency vibration: vibration with low frequency caused by track irregularities, wheel asymmetry, etc. Usually less than 20Hz.

[0115] The above high-frequency noise may come from sensor noise, electrical noise, etc., usually greater than 50Hz.

[0116] Preferably, when using a Butterworth low-pass filter, the cutoff frequency ωc For example, the cutoff frequency ω c is 30 Hz, then how to filter out the high frequency noise is determined by calculating the frequency response of the Butterworth filter.

[0117] ω c = 2π×30 rad / s ≈ 188.4 rad / s. This is the value of the cutoff frequency ω c in the state of the low-pass filter with a frequency of 30 Hz.

[0118] For example, assuming that the cutoff frequency ω c is 30 Hz, the cutoff frequency determines the frequency at which the filter begins to attenuate.

[0119] Substitute each cutoff frequency into the transfer function of the filter to calculate the corresponding gain. This represents the response of the filter at that frequency.

[0120] For signals below the cutoff frequency, the filter hardly attenuates.

[0121] For signals above the cutoff frequency, the filter gradually attenuates the signal, reducing high frequency noise.

[0122] Specifically, when the cutoff frequency ω c is 30 Hz, the frequency components 1 Hz-30 Hz in the vibration signal of the train acceleration are basically not affected, and high frequency noise above 30 Hz is filtered out.

[0123] The transfer function after filtering out high frequency noise becomes:

[0124]

[0125] Set s = jω, where ω is the actual frequency of the signal.

[0126] Substitute the complex frequency variable s = jω into the transfer function to calculate the processing effect of the filter on the signal.

[0127] In the present application, after the acceleration and angular velocity data are low-pass filtered by the filter, the filtered signal is:

[0128]

[0129]

[0130] where * represents convolution operation, and represent the acceleration and angular velocity signals obtained after filtering, respectively.

[0131] S162: Smooth the data.

[0132] The filtered acceleration and angular velocity signals may still have slight noise fluctuations, so a smoothing process is used to further suppress residual noise and enhance data stability. The smoothing process uses a moving average method and double smoothing technique.

[0133] S1621: Moving average method.

[0134] The moving average method is a common smoothing technique that eliminates short-term fluctuations by taking the average of the data within a window. For position, velocity, and angular velocity data, the smoothing formula for the moving average is:

[0135]

[0136] where N represents the window size, and x(t+k) represents the value at the Kth point after the original signal x(t). k is an index that represents the value at t+k within N time points before and after time t. Since the data measured by the invention is angular velocity and acceleration data, only angular velocity and acceleration data are smoothed.

[0137] S1622: Double smoothing.

[0138] To further improve the smoothing effect, the system performs double smoothing on the data. First, the preliminary smoothed data is subjected to a second moving average:

[0139]

[0140] Through this double smoothing process, high-frequency noise can be further removed while maintaining signal stability. In particular, in turn detection and mileage calibration, using smoothed data can reduce jitter and ensure the accuracy of subsequent calculations.

[0141] S163: Overall formula explanation of data processing process.

[0142] As shown above, the system processes the original inertial measurement unit (IMU) data in two steps.

[0143] Low-pass filtering: remove high-frequency noise through a Butterworth low-pass filter:

[0144]

[0145]

[0146] Double smoothing: further smooth the filtered signal through two moving averages:

[0147]

[0148]

[0149] x can represent acceleration a or angular velocity ω. When x represents acceleration a, x smooth (t) represents filtered acceleration a, x smooth2 (t) represents twice-smoothed acceleration a.

[0150] When x represents angular velocity ω, x smooth (t) represents filtered angular velocity ω, x smooth2 (t) represents twice-smoothed angular velocity ω.

[0151] Through the above low-pass filtering and twice-smoothing processing, the system can effectively remove high-frequency noise in the inertial measurement unit (IMU) data, ensuring that in the calculation of key data such as acceleration and angular velocity, noise interference is reduced, and a smoother signal is provided. This processing method is particularly suitable for turn detection and mileage calibration in inertial navigation, improving the stability and accuracy of the system.

[0152] Figure 3 A subway track line map is shown. Figure 3 XK0+139 in the above represents the 0th km and the 139th m. XK1 represents 1 kilometer (1 km). Figure 3 XK1+355.944 in the above represents the start of a turn at 1355.944 m. That is, there is a turn location between two stations: XK1+355.944.

[0153] As Figure 4 shown, in the case of an inertial measurement unit (IMU) placed on a train, the X direction or Y direction of the inertial measurement unit (IMU) is the direction of train operation. The curved arrow represents the angular velocity around the Z axis.

[0154] When the train passes through the turn location, the inertial measurement unit (IMU) will detect and record the angular velocity change, as Figure 5 shown. In Figure 5 , the horizontal axis represents time, and the vertical axis represents the magnitude of angular velocity around the z axis. The angular velocity changes significantly within the turn section.

[0155] The present application explains step S200 as follows.

[0156] S210: Convert acceleration and angular velocity into position and velocity to construct a time-mileage curve.

[0157] Numerical integration of acceleration and angular velocity data, the acceleration and angular velocity data of the train at different time points are converted into position and velocity information through integral operation, generating a time-mileage curve of the train.

[0158] S211: The acceleration data is calculated by integration to obtain the speed.

[0159] The integral formula is: V(t) = V0 + ∫a(t)dt.

[0160] Wherein, V(t) represents the speed at time t, a(t) represents the acceleration data, and V0 represents the initial speed.

[0161] S212: The speed data is twice integrated to obtain the mileage of the train.

[0162] S(t) = S0 + ∫V(t)dt = S0 + ∫(V0 + ∫a(t)dt)dt.

[0163] Wherein, S(t) represents the mileage of the train at time t, and S0 represents the initial mileage.

[0164] S220: The ratio of the mileage to the first target mileage is taken as the first correction ratio.

[0165] As shown in Figure 6 , the running test data of XX subway line 9 is shown. Figure 6 The information includes total mileage setting, query time point, query result, running mileage, running time, correction file option, algorithm selection box and other information.

[0166] S221: The first target mileage is input.

[0167] In the "total mileage setting", the true mileage S target is input. target The total mileage calculated by the inertial measurement unit (IMU) is corrected.

[0168] S222: Total mileage correction.

[0169] The ratio of the mileage S(t) calculated by the inertial measurement unit (IMU) integration to the first target mileage S target input by the user is calculated to obtain the first correction ratio k.

[0170]

[0171] When k>1, it means that the calculated mileage is less than the first target mileage, and stretching is needed; when k<1, it means that the calculated mileage is greater than the first target mileage, and compression is needed.

[0172] S230: Based on the first correction ratio, the mileage data at each time point is scaled in proportion, so as to obtain the corrected mileage, and the time-mileage curve is corrected. The time-mileage curve is used to intuitively display the mileage change of the train.

[0173] S231: Scale the mileage data at each time point by the first correction ratio k.

[0174] S corrected (t) = S(t) x k.

[0175] wherein S corrected (t) represents the corrected mileage.

[0176] S232: Draw the time-mileage curve.

[0177] As Figure 7 shown, the corrected mileage-time curve is generated and displayed to visually show the mileage change of the train.

[0178] Figure 7 The right table in the figure shows the time on the horizontal axis and the mileage on the vertical axis. The curve formed by the original data and the curve formed by the corrected data are shown in the table.

[0179] Compare the mileage-time curves before and after correction. The corrected mileage-time curve can more accurately show the mileage change of the train.

[0180] The present application explains step S300 as follows.

[0181] In order to more accurately identify the time when the train enters and leaves the turn, the Z-axis angular velocity ω z (t) is analyzed in higher order, the rate of change of angular velocity and the second derivative are used to capture the small changes of the start and end points of the turn, and dynamic threshold adjustment is used to adapt to different turn radii and turn speeds.

[0182] S310: Nonlinear fitting of angular velocity curve, calculation of rate of change of angular velocity and angular acceleration.

[0183] The change of Z-axis angular velocity with time can be modeled by a third-order Bezier curve:

[0184] ω z (t) = a3t 3 +a2t 2 +a1t+a0.

[0185] wherein a0, a1, a2, a3 represent the coefficients obtained by least squares fitting. Here, the nonlinear fitting of the angular velocity curve is used to capture the dynamic change characteristics of the turn.

[0186] S320: Set a dynamic threshold based on the current speed of the train and the curve radius of the track, and determine the start time of the train in the turning section when the angular velocity is greater than the dynamic threshold; determine the end time of the train in the turning section based on the rate of change of angular velocity and the negative change of angular acceleration.

[0187] S321: Determination of the start time of the curve section.

[0188] The start time of the curve section is determined by the first derivative and the second derivative of the angular velocity.

[0189] The rate of change of the Z-axis angular velocity is calculated and the acceleration which is defined as follows:

[0190]

[0191]

[0192] The dynamic threshold Δω is set threshold (t), which is based on the current speed of the train and the curve radius of the track, and is calculated as:

[0193]

[0194] where α represents the proportional coefficient, V(t) represents the current speed, and R(t) represents the curvature radius of the track.

[0195] Preferably, α is used to adjust the sensitivity of the curve dynamic detection. It controls the size of the threshold, thereby determining how to detect the change in angular velocity of the train during the curve. When the speed of the train and the curvature radius of the track change, the dynamic threshold adjusts accordingly. The value of α ranges from 0.8 to 1.5.

[0196] When the rate of change of the angular velocity reaches the threshold, the train starts to turn:

[0197]

[0198] The start time of the curve section of the train not only depends on the change in angular velocity, but also depends on the positive change in angular acceleration, to ensure the dynamic characteristics of the curve.

[0199] S322: Determination of the end time of the curve section.

[0200] The end time of the curve section is determined by the rate of change of the angular velocity and the negative change in acceleration.

[0201] The specific formula is as follows:

[0202]

[0203] At this time, the angular velocity decreases and the angular acceleration is negative, indicating that the train is leaving the curve section.

[0204] High-order analysis of angular velocity is performed, and the rate of change and the second derivative of angular velocity are used to capture the slight changes in the start and end times within the turning section, and dynamic threshold is dynamically adjusted to adapt to different turning radii and turning speeds.

[0205] S330: Correct the total mileage within the turning section.

[0206] S331: Calculate the total mileage within the turning section based on the speed of the train.

[0207] To calculate the total mileage of the train within each turning section, the speed of the train is integrated. The speed V(t) is obtained by twice integrating the acceleration data of the inertial measurement unit (IMU):

[0208]

[0209] where a(t) represents acceleration and V0 represents the initial speed.

[0210] Further integration of the speed gives the total mileage within the turning section:

[0211]

[0212] The calculation of the total mileage within the turning section not only considers the speed, but also corrects the nonlinear change of acceleration by integrating the acceleration.

[0213] S332: Take the ratio of the total mileage within the turning section to the second target distance within the turning section as the second correction ratio.

[0214] The total mileage within the turning section calculated by the inertial measurement unit (IMU) may have a significant error from the actual total mileage within the turning section. By calibrating the total mileage with a nonlinear correction model, it is assumed that the terminal input second target mileage is S input,bi , input, ei,…(representing the input mileage begin of the first turning start, the input mileage end of the first turning end, respectively), the second correction ratio k i (t) is a dynamic function that changes over time to adapt to changes in the train under different speeds and turning conditions:

[0215]

[0216] where the denominator is the total mileage actually integrated within the turning section.

[0217] S333: Scale the total mileage within the turning section by the second correction ratio.

[0218] Specifically, using the second correction ratio, the total mileage within each turning section is nonlinearly scaled:

[0219]

[0220] S(t) represents the total mileage in the original turn section, S corrected (t) represents the total mileage in the corrected turn section. represents the start time in the first turn section, represents the end time in the first turn section.

[0221] The formula ensures that, in each turn section, the corrected mileage in the turn section is consistent with the input mileage mark, and the scaling ratio is dynamically adjusted to match the real-time changing speed.

[0222] S334: Smoothly calibrating the total mileage in the corrected turn section by using a transition function, so that the corrected time-mileage curve is continuous and smoothly connected at the calibration point.

[0223] The transition function is:

[0224] where β represents a parameter for controlling the transition. The corrected total mileage can be represented as:

[0225]

[0226] S final (t) represents the total mileage in the turn section after being smoothed and calibrated. Discontinuity of the corrected time-mileage curve at the calibration point can be avoided, so that the time-mileage curve is smoothly connected before and after the calibration point, avoiding abrupt changes.

[0227] The present application explains step S400 as follows.

[0228] The identification of abnormal vibration is based on the analysis of high-order change rates of acceleration and angular velocity. By analyzing the third derivative of acceleration and the third derivative of angular velocity , the violent vibration change can be captured.

[0229] S410: Calculate the third derivative functions of the acceleration and angular velocity of the train respectively, and the absolute value of the third derivative function is the vibration intensity.

[0230] In order to capture the vibration more accurately, the third derivative of acceleration and the third derivative of angular velocity

[0231]

[0232]

[0233] These third-order derivatives represent the rate of change of acceleration and angular velocity, respectively. In other words, they are used to capture dramatic changes in acceleration and angular velocity. Based on the magnitude of these changes, vibration points with significant changes can be quickly identified. Specifically, the locations and durations of vibrations with significant changes in angular velocity and acceleration can be quickly determined. This effectively captures vibration points that experience dramatic changes within a short period of time.

[0234] S420: Dynamically adjust the vibration point identification threshold based on the train speed. The position corresponding to the vibration intensity greater than the identification threshold is an abnormal vibration point.

[0235] According to the different speeds and environmental conditions of the train, the identification threshold of the vibration point and It is also dynamically adjusted:

[0236]

[0237]

[0238] Where γ and δ are dynamic adjustment coefficients, γ ≥ 1 and δ ≤ 5, to ensure the accuracy of vibration point identification under different speed conditions. V(t) represents the current speed, and R(t) represents the curvature radius of the track.

[0239] γ and δ are dynamic adjustment coefficients used to adjust the vibration point identification threshold. These coefficients dynamically adjust the thresholds for vibrations caused by acceleration and angular velocity based on the train's speed (V(t)) and the track's radius of curvature (R(t)). This prevents large vibrations caused by turning or acceleration from being misidentified as abnormal vibrations.

[0240] When the train speed is low or the track is straight, the recognition threshold is lower so that smaller vibration changes can be detected.

[0241] Preferably, a point in the time-distance curve at which the rate of change of the angular velocity and the rate of change of the acceleration are greater than an identification threshold is determined as an abnormal vibration point.

[0242] S430: Mark the vibration point in a visual manner.

[0243] Each detected abnormal vibration point is not only marked with its time and mileage, but also uses color to indicate the severity of the vibration. The present invention determines the size and color of the visual vibration point based on the vibration intensity, thereby indicating the severity of the vibration, allowing users to see the specific information of each vibration point at a glance.

[0244] Figure 8The graph in the figure shows the absolute velocity changing with time, which is also a visual diagram of the abnormal vibration point. The horizontal axis of the curve represents time (seconds) and the vertical axis represents the absolute velocity (m / s). Figure 8 As shown in the curve, the circular points on the curve represent abnormal train vibration points, and the triangular points represent vibration changes caused by the train turning (which are normal and can be ignored). By visualizing the abnormal vibration points, users can directly view the number and frequency of abnormal train vibrations and also directly identify the abnormal vibration points.

[0245] The use of nonlinear correction models and high-order derivative analysis not only improves the accuracy of turning moment identification but also reduces the accumulation of mileage errors. Furthermore, third-order derivative analysis of acceleration and angular velocity enables more sensitive identification of abnormal vibration points, providing more accurate track maintenance information.

[0246] Preferably, the present invention can ensure that at least three mileage corrections are performed at each station by performing mileage corrections at turning points. Compared with the prior art method of only performing mileage corrections at the end, the time-mileage curve obtained by the present invention is closer to the time-mileage curve of train operation under actual conditions.

[0247] As mentioned above, due to the presence of a turning point, the Z-axis angular velocity data measured by the inertial measurement unit (IMU) when the train passes through this point will be significantly different from that when the train is moving in a straight line. Based on this principle, the moment when the train enters the turning point can be determined, and the mileage information calculated after data processing can be corrected to the information at the starting point of the turn.

[0248] The present invention also provides sample data obtained by testing the subway train positioning system, such as Figure 11 As shown. Figure 11 The test chart of the total mileage data of the preliminary correction is as follows Figure 9 As shown ( Figure 9 The triangle mark in the figure can be ignored); the test graph of the calculated total mileage data is as follows Figure 10 shown.

[0249] The method and system of the present invention can achieve the following objectives:

[0250] First, improve positioning accuracy and reduce error accumulation.

[0251] This method combines an inertial navigation system with subway track feature points (such as turning points, stops, and points of sudden acceleration at switches) and uses an error Kalman filter algorithm to correct the inertial navigation unit (INS) data. Using track feature points as a natural calibration reference, it achieves high-precision positioning in GPS-free environments, keeping the error within a reasonable range (5%).

[0252] Second, get rid of external signal dependence, reduce cost.

[0253] Compared with the existing GPS and UWB scheme, the application does not need to rely on external signals or install additional base station equipment, completely eliminates the signal limitation problem of GPS and Beidou in the underground environment of the subway, avoids the high installation and maintenance cost of the base station, and has good economy and operability.

[0254] Third, improve the convenience and effectiveness of train track detection.

[0255] The positioning and stability detector 200 adopted by the application has high integration, cooperates with the error Kalman filtering algorithm and the calibration strategy of the track feature point, can generate a time-mile curve of the train closer to the real one, is convenient for precise detection and maintenance of the subway line track abnormal point in the GPS-free environment, and greatly improves the precision and effectiveness of the train stability detection.

[0256] By achieving the above-mentioned objectives, the application provides a low-cost and high-precision solution for precise positioning and track detection in underground environment, effectively solves the defects of the positioning technology in the existing GPS-free and base station-uninstallable conditions.

[0257] It should be noted that the above specific embodiments are exemplary, and those skilled in the art can come up with various solutions under the inspiration of the disclosure of the application, and these solutions also belong to the disclosed range of the application and fall within the protection scope of the application. Those skilled in the art should understand that the specification and drawings of the application are illustrative and not constitute a limitation on the claims. The protection scope of the application is defined by the claims and their equivalents. The specification of the application contains multiple inventive concepts, such as "preferably", "according to a preferred embodiment", which means that the corresponding paragraph discloses an independent concept, and the applicant reserves the right to file a divisional application according to each inventive concept.

Claims

1. A subway train positioning system, characterized by, comprising a processor (100) configured to: process the multi-source data of the train collected by the inertial navigation unit based on an error Kalman filtering algorithm to preliminarily correct the acceleration and angular velocity, wherein the processing based on the error Kalman filtering algorithm comprises: in an initial stage of running of the processor (100), setting a state vector according to initial acceleration, angular velocity and position data of the inertial navigation unit, and initializing a state error covariance matrix according to characteristics of the inertial measurement unit, predicting a next time state of the train based on a preset motion model and the acceleration and angular velocity of the train collected by the inertial navigation unit, correcting the predicted state information of the train based on a Kalman gain and updating an error covariance matrix related to the prediction, thereby correcting the predicted state of the train; constructing a time-mileage curve of the train and correcting the time-mileage curve based on a first correction ratio; identifying a dynamic change feature of a turn based on the angular velocity collected by the inertial navigation unit, the processor (100) being configured to identify the dynamic change feature of the turn in the following manner: performing nonlinear fitting on the angular velocity curve, calculating a rate of change of the angular velocity and an angular acceleration, setting a dynamic threshold based on a current speed of the train and a curve radius of a track, determining a start time in a turn section of the train in a case where the angular velocity is greater than the dynamic threshold, determining an end time in the turn section of the train based on a negative change of the rate of change of the angular velocity and the angular acceleration, wherein a total mileage in the turn section is calculated, and the total mileage in the turn section is corrected based on a second correction ratio of the turn section; identifying an abnormal vibration feature based on a high-order rate of change of the acceleration and the angular velocity of the train and marking the abnormal vibration feature in a visual manner.

2. The system of claim 1, wherein, The processor (100) is further configured to: perform data filtering and smoothing processing on the preliminarily corrected acceleration and angular velocity before constructing the time-mileage curve of the train, so as to remove noise in the data.

3. The system of claim 2, wherein, The processor (100) is configured to correct the time-mileage curve in the following manner: convert the acceleration and the angular velocity into position and speed to construct the time-mileage curve; take a ratio of the mileage to a first target mileage as the first correction ratio, perform equal-ratio scaling on the mileage data at each time point based on the first correction ratio, so as to obtain corrected mileage, thereby correcting the time-mileage curve.

4. The system of claim 3, wherein, The processor (100) is configured to correct the total mileage in the turn section in the following manner: calculate the total mileage in the turn section based on the speed of the train; take a ratio of the total mileage in the turn section to a second target mileage in the turn section as the second correction ratio; perform equal-ratio scaling on the total mileage in the turn section based on the second correction ratio.

5. The system of claim 4, wherein, The processor (100) is further configured to correct the total mileage in the turn section in the following manner: smoothly calibrate the corrected total mileage in the turn section by using a transition function, so that the corrected time-mileage curve is continuous and smoothly connected at a calibration point.

6. The system of claim 5, wherein, The processor (100) is configured to identify the abnormal vibration feature in the following manner: The third derivative functions of the acceleration and the angular velocity of the train are calculated respectively, and the absolute values of the third derivative functions are vibration intensities; The identification threshold of the vibration point is dynamically adjusted based on the speed of the train; The position corresponding to the vibration intensity greater than the identification threshold is an abnormal vibration point.

7. A method of positioning a subway train, characterized by, The method comprises: The acceleration and the angular velocity of the train collected by the inertial navigation unit are processed based on an error Kalman filtering algorithm to preliminarily correct the acceleration and the angular velocity, wherein the processing based on the error Kalman filtering algorithm comprises: A state vector is set according to the initial acceleration, the angular velocity and the position data of the inertial navigation unit, and a state error covariance matrix is initialized according to the characteristics of the inertial measurement unit, The next time state of the train is predicted based on a preset motion model and the acceleration and the angular velocity of the train collected by the inertial navigation unit, The predicted state information is corrected and the error covariance matrix related to the prediction is updated based on a Kalman gain, so as to correct the predicted state of the train; The time-mileage curve of the train is constructed and the time-mileage curve is corrected based on a first correction ratio; The dynamic change characteristics of the turning are identified based on the angular velocity collected by the inertial navigation unit, comprising: The angular velocity curve is nonlinearly fitted, The rate of change of the angular velocity and the angular acceleration are calculated, The dynamic threshold is set based on the current speed of the train and the curve radius of the track, The start time in the turning section of the train is determined in the case that the angular velocity is greater than the dynamic threshold, The end time in the turning section of the train is determined based on the negative change of the rate of change of the angular velocity and the angular acceleration, Wherein, the total mileage in the turning section is calculated, and the total mileage in the turning section is corrected based on a second correction ratio of the turning section. The abnormal vibration characteristics are identified based on the high-order rates of change of the acceleration and the angular velocity of the train, and are marked in a visual manner.

8. The method of claim 7, wherein, The method further comprises: Before the time-mileage curve of the train is constructed, the preliminarily corrected acceleration and the angular velocity are subjected to data filtering and smoothing processing to remove the noise in the data.

Images