Railway signal axle counting system

Abstract

A railway signal axle counting system, comprising axle counting sensors, trackside devices, and an indoor main unit (10). Each trackside device comprises an axle counting ground trackside apparatus and an information demodulation device. The axle counting sensors are disposed on steel rails, identify the number of axles when a train passes through and the travel direction of the train, and send said number of axles and said travel direction to the axle counting ground trackside apparatus in the form of an analog signal for parsing to obtain trackside axle counting information, and the trackside axle counting information is sent to the indoor main unit (10). The indoor main unit (10) determines that the state of each steel rail segment is an idle state or an occupied state. The information demodulation device uses distributed acoustic wave sensor technology, and uses the backscattering power of optical cables to detect a steel rail vibration acoustic wave signal, and obtains a detection result indicating whether the steel rails are broken. Whether the steel rail segments are occupied can be checked, and whether the steel rails are broken can be automatically checked, thereby discovering in a timely manner potential dangers existing during travel of the train.

Description

Railway signal axle counting system

[0001] This application claims priority to the Chinese patent application filed with the China Patent Office on January 12, 2024, with application number 202410050721.7 and invention name “A Railway Signal Axle Counter System”, the entire contents of which are incorporated by reference into this application. Technical Field

[0002] The present invention relates to the technical field of railway signals, and more particularly to a railway signal axle counting system. Background Art

[0003] Railway signal axle counting equipment, used to check track section occupancy and availability, uses axle counting sensors placed at both ends of a block section to detect train entry in real time. Before a train leaves, it continuously outputs information indicating a section is occupied. Once all trains have left the section, it outputs information indicating a section is free. Once a block section is occupied, the train control system prevents other trains from entering it, ensuring safe train operation.

[0004] However, when a railway rail breaks, it poses a potential danger to the train, but railway signal axle counting equipment cannot detect this situation and can only rely on manual visual inspection. This shortcoming limits the widespread application of railway signal axle counting equipment.

[0005] Summary of the Invention

[0006] In view of this, the present invention discloses a railway signal axle counting system, which can realize automatic inspection of whether the rail section is occupied and whether the rail is broken, thereby facilitating timely detection of potential dangers in the train operation process and realizing the widespread application of the railway signal axle counting system.

[0007] A railway signal axle counting system, comprising: an axle counting sensor, a trackside device, and an indoor host computer, wherein the trackside device is connected to the axle counting sensor and the indoor host computer respectively, and the trackside device comprises: a connected axle counting ground trackside device and an information demodulation device, wherein the number of the axle counting sensor and the number of the trackside device are both N, where N is a positive integer;

[0008] The axle counting sensor is arranged on the rail to sense the wheel status when a train passes by, identify the number of axles and the direction of the train when the train passes by, and output an analog signal to the axle counting ground trackside device, wherein two adjacent axle counting sensors are respectively arranged at the two ends of the same rail section;

[0009] The axle counting ground trackside device is communicatively connected to the indoor host, and is used to analyze the analog signal to obtain trackside axle counting information, and transmit the trackside axle counting information to the indoor host, wherein the trackside axle counting information includes: the number of axles and the direction of travel of the train;

[0010] The indoor host is used to determine whether the status of each rail section is idle or occupied based on each trackside axle counting information and rail section configuration information transmitted by each trackside axle counting device, wherein the rail section configuration information records the axle counting sensors installed at both ends of each rail section;

[0011] The information demodulation device is used to adopt distributed acoustic wave sensor technology, use the backscattered power of the optical cable to detect the rail vibration acoustic wave signal, and obtain the detection result of whether the rail is broken based on the rail vibration acoustic wave signal.

[0012] Optionally, the information demodulation device includes: a laser, a first coupler, a second coupler, a modulator, an optical circulator, and a balanced detector;

[0013] The laser is used to output continuous high-coherence laser light;

[0014] The first coupler is connected to the laser and is used to separate the high coherence laser into detection light and local oscillator light;

[0015] The modulator is connected to the first coupler and is used to modulate the continuous detection light into pulsed light;

[0016] The optical circulator is connected to the modulator and is used to input the pulse light into the optical cable under test, so that the pulse light is scattered in the optical cable under test to form scattered light and then passes through the optical circulator again; wherein the optical cable under test is the optical cable connecting the information demodulation equipment and the adjacent trackside equipment;

[0017] The second coupler is connected to the first coupler, the optical circulator and the balanced detector respectively, and is used to output the interference formed by the local oscillator light and the scattered light to the balanced detector;

[0018] The balanced detector is used to use the scattered light in the interference as the rail vibration sound wave signal, and obtain a detection result of whether the rail is broken based on the amplitude comparison result of the rail vibration sound wave signal and the local oscillation light.

[0019] Optionally, the balanced detector is specifically used for:

[0020] Calculating the amplitude difference between the rail vibration sound wave signal and the local oscillation light;

[0021] When the amplitude difference exceeds a preset difference, it is determined that the rail is broken;

[0022] When the amplitude difference does not exceed the preset difference, it is determined that the rail is not broken.

[0023] Optionally, the optical circulator is further configured to: after inputting the pulse light into the optical cable under test, obtain a time delay difference of the pulse light propagating in the optical cable under test;

[0024] The balance detector is further used to determine an abnormal position of the rail according to the time delay difference.

[0025] Optionally, the balanced detector is further used for:

[0026] determining a propagation speed of the pulsed light in the optical cable under test;

[0027] Taking the product of the propagation speed and the time delay difference to obtain a propagation distance;

[0028] The position of the axle counter sensor corresponding to the information demodulation device on the rail is taken as a starting point, and a position whose distance from the starting point is taken as the propagation distance is determined as the abnormal position of the rail.

[0029] Optionally, the balanced detector is further used for:

[0030] Filtering the rail vibration sound wave signal to obtain an intermediate rail vibration sound wave signal;

[0031] performing wavelet decomposition and reconstruction on the middle rail vibration acoustic wave signal, removing the overall trend quantity in the middle rail vibration acoustic wave signal, and obtaining a target rail vibration acoustic wave signal;

[0032] Selecting a light wave signal characteristic value from the target rail vibration sound wave signal;

[0033] The light wave signal characteristic value is input into a pre-trained neural network model to obtain a detection result of whether the rail is broken or not.

[0034] Optionally, the training process of the neural network model includes:

[0035] Initialize neural network weights and neural network biases;

[0036] Set the number of nodes in the input layer, hidden layer, and output layer to obtain the initial neural network model;

[0037] Inputting light wave signal characteristic value sample data as input values ​​into the input layer of the initial neural network model;

[0038] In the input layer, the light wave signal characteristic value sample data is added to the neural network bias according to the neural network weight to obtain intermediate sample data, and the intermediate sample data is input into the hidden layer;

[0039] After the hidden layer transforms the intermediate sample data using a transfer function, the data is output to the output layer;

[0040] Determine the error value and node error rate of each node in the output layer;

[0041] Based on the node error value and the node error rate, updating the neural network weight and the neural network bias;

[0042] Determine whether the current model iteration number reaches the preset iteration number;

[0043] If yes, the neural network model training is completed;

[0044] If not, return to step 1 and input the lightwave signal characteristic value sample data again.

[0045] Optionally, the information demodulation device further includes: a memory;

[0046] The memory is connected to the balance detector and is used to store the detection result of whether the rail is broken.

[0047] Optionally, the information demodulation device is further used to:

[0048] When it is determined that the rail is broken, the broken rail information is sent to the indoor host, and the indoor host outputs an alarm message.

[0049] From the above technical solution, it can be seen that the present invention discloses a railway signal axle counting system, which includes: an axle counting sensor, a trackside device and an indoor host. The trackside device is connected to the axle counting sensor and the indoor host respectively. Each trackside device includes: a connected axle counting ground trackside device and an information demodulation device. The axle counting sensor is set on the rail to identify the number of axles and the direction of train travel when the train passes, and sends it to the axle counting ground trackside device in the form of an analog signal. The axle counting ground trackside device parses the analog signal to obtain trackside axle counting information, including the number of axles and the direction of train travel, and sends the trackside axle counting information to the indoor host. The indoor host determines whether the status of each rail section is idle or occupied. The information demodulation device adopts distributed acoustic wave sensor technology, uses the backscattered power of the optical cable to detect the rail vibration acoustic wave signal, and obtains the detection result of whether the rail is broken based on the rail vibration acoustic wave signal. The railway signal axle counting system disclosed in the present invention can not only check whether a rail section is occupied, but also automatically check whether a rail is broken, thereby facilitating timely discovery of potential dangers in the process of train operation and realizing the widespread application of the railway signal axle counting system. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the disclosed drawings without any creative work.

[0051] FIG1 is a schematic diagram of a railway signal axle counting system disclosed in an embodiment of the present invention;

[0052] FIG2 is a schematic structural diagram of an information demodulation device disclosed in an embodiment of the present invention;

[0053] FIG3 is a schematic structural diagram of another information demodulation device disclosed in an embodiment of the present invention;

[0054] FIG4 is a schematic diagram of the structure of an initial neural network model disclosed in an embodiment of the present invention;

[0055] FIG5 is a schematic diagram of a neural network model training result disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0057] An embodiment of the present invention discloses a railway signal axle counting system, which includes: an axle counting sensor, a trackside device and an indoor host. The trackside device is connected to the axle counting sensor and the indoor host respectively. Each trackside device includes: a connected axle counting ground trackside device and an information demodulation device. The axle counting sensor is set on the rail to identify the number of axles and the train's direction of travel when a train passes, and sends the information in the form of an analog signal to the axle counting ground trackside device. The axle counting ground trackside device analyzes the analog signal to obtain trackside axle counting information, including the number of axles and the train's direction of travel, and sends the trackside axle counting information to the indoor host. The indoor host determines whether the status of each rail section is idle or occupied. The information demodulation device adopts distributed acoustic wave sensor technology, uses the backscattered power of the optical cable to detect the rail vibration acoustic wave signal, and obtains a detection result of whether the rail is broken based on the rail vibration acoustic wave signal. The railway signal axle counting system disclosed in the present invention can not only check whether a rail section is occupied, but also automatically check whether a rail is broken, thereby facilitating timely discovery of potential dangers in the process of train operation and realizing the widespread application of the railway signal axle counting system.

[0058] 1 , which is a schematic diagram of a railway signal axle counting system disclosed in an embodiment of the present invention, includes an axle counting sensor, trackside equipment, and an indoor host 10 , wherein the trackside equipment is connected to the axle counting sensor and the indoor host 10 , respectively.

[0059] Preferably, the trackside equipment is electrically connected to the axle counting sensor, and the trackside equipment is connected to the indoor host 10 via an optical cable.

[0060] In this embodiment, the number of axle counting sensors and trackside equipment is N, where N is a positive integer.

[0061] For example, see Figure 1, the trackside equipment 1 is connected to the axle counter sensor 1, the trackside equipment 2 is connected to the axle counter sensor 2, and the trackside equipment 3 is connected to the axle counter sensor 3. The axle counter sensor 1 and the axle counter sensor 2 are arranged at both ends of the rail section 1, and the axle counter sensor 2 and the axle counter sensor 3 are arranged at both ends of the rail section 2.

[0062] Each trackside equipment includes: a connected axle counting ground trackside device and an information demodulation device. For example, in Figure 1, trackside equipment 1 includes: a connected axle counting ground trackside device 1 and an information demodulation device 1, trackside equipment 2 includes: a connected axle counting ground trackside device 2 and an information demodulation device, and trackside equipment 3 includes: a connected axle counting ground trackside device 3 and an information demodulation device 3.

[0063] In practical applications, the axle counter sensor is set on the rail to sense the wheel status when a train passes by, identify the number of axles and the direction of the train when the train passes by, and output it to the axle counter ground trackside device in the form of an analog signal.

[0064] In this embodiment, the axle counter sensor is designed based on the principle of electromagnetic induction. It has a transmitting coil in the middle and a receiving coil on both sides. A 25kHz AC signal enters the transmitting coil, generating an alternating magnetic field. The receiving coil senses the voltage and outputs the voltage. When the wheels pass by the axle counter sensor, they block the receiving coils respectively, and the receiving coils sequentially change to low voltage output, thereby achieving the purpose of identifying the number of axles and determining the direction of the train.

[0065] Two adjacent axle counter sensors are located at either end of the same rail segment. For example, in Figure 1, when a train is traveling from left to right, axle counter sensor 1 and axle counter sensor 2 are located at either end of rail segment 1, each identifying the number of axles and the train's direction of travel as the train passes. Axle counter sensor 2 and axle counter sensor 3 are located at either end of rail segment 2, each identifying the number of axles and the train's direction of travel as the train passes.

[0066] In actual applications, the location of each axle counter sensor on the rail can be determined based on site needs. For example, if the length of rail section 1 is set to 500 meters, the distance between axle counter sensors 1 and 2 can be 500 meters. Depending on actual conditions, the length of rail section 1 can also be set to 300 meters, depending on actual needs.

[0067] The ground trackside axle counting device is in communication with the indoor host 10 and is used to analyze the analog signal to obtain trackside axle counting information and transmit the trackside axle counting information to the indoor host 10. The trackside axle counting information includes: the number of axles and the direction of train travel.

[0068] In actual applications, all the axle counting ground trackside devices in the axle counting system transmit the trackside axle counting information obtained by their respective analysis to the indoor host computer 10 via optical cables.

[0069] The indoor host 10 is used to determine whether the status of each rail section is idle or occupied based on the trackside axle counting information and rail section configuration information transmitted by each of the axle counting ground trackside devices, wherein the rail section configuration information records the axle counting sensors set at both ends of each rail section.

[0070] In this embodiment, the rail section configuration information is the information for configuring the rail section, which is completed on the indoor host 10 through the configuration table. After the configuration is completed, the indoor host 10 can know that the axle counting sensor 1 and the axle counting sensor 2 are the input sensor and output sensor of the rail section 1 respectively. In this way, when the indoor host 10 receives the number of axles identified by all the axle counting sensors, it compares whether the number of axles entering and exiting the rail section 1 is consistent. For example, if there are 5 axles entering and 4 axles exiting, it means that the train is in the rail section 1.

[0071] For determining whether the state of the rail section is idle or occupied, for example, assuming that rail section 1 enters 5 axles and outputs 0, then 0-5=-5, indicating that rail section 1 is in the occupied state; if rail section 1 enters 5 axles and outputs 5 axles, then 5-5=0, indicating that rail section 1 is in the idle state.

[0072] In practical applications, the axle counting sensors, the axle counting ground trackside device and the indoor host 10 together constitute an outdoor distributed axle counting device.

[0073] Each information demodulation device is used to adopt distributed acoustic wave sensor technology, use the backscattered power of the optical cable to detect the rail vibration acoustic wave signal, and obtain the detection result of whether the rail is broken based on the rail vibration acoustic wave signal.

[0074] In actual applications, the information demodulation devices in adjacent trackside equipment are also connected by optical cables.

[0075] In this embodiment, the information demodulation equipment is designed based on distributed acoustic wave sensing technology. It utilizes the high backscatter power of Rayleigh scattering in optical cables, resulting in a higher signal-to-noise ratio and faster response speed to detect rail vibration acoustic wave signals. Because the entire optical cable can be divided into a series of distributed pickups at high density intervals, the information demodulation equipment also features high resolution and large scale. Furthermore, the information demodulation equipment features continuous, long-term online monitoring, which also offers significant advantages in response time and operational maintenance.

[0076] In summary, the present invention discloses a railway signal axle counting system, which includes: an axle counting sensor, a trackside device and an indoor host 10. The trackside device is connected to the axle counting sensor and the indoor host 10 respectively. Each trackside device includes: a connected axle counting ground trackside device and an information demodulation device. The axle counting sensor is set on the rail to identify the number of axles and the direction of train travel when the train passes, and sends it to the axle counting ground trackside device in the form of an analog signal. The axle counting ground trackside device parses the analog signal to obtain trackside axle counting information, including the number of axles and the direction of train travel, and sends the trackside axle counting information to the indoor host. The indoor host determines whether the status of each rail section is idle or occupied. The information demodulation device adopts distributed acoustic wave sensor technology, uses the backscattered power of the optical cable to detect the rail vibration acoustic wave signal, and obtains the detection result of whether the rail is broken based on the rail vibration acoustic wave signal. The railway signal axle counting system disclosed in the present invention can not only check whether a rail section is occupied, but also automatically check whether a rail is broken, thereby facilitating timely discovery of potential dangers in the process of train operation and realizing the widespread application of the railway signal axle counting system.

[0077] To further optimize the above embodiment, refer to Figure 2, which is a structural diagram of an information demodulation device disclosed in an embodiment of the present invention. The information demodulation device includes: a laser 21, a first coupler 22, a second coupler 23, a modulator 24, an optical circulator 25 and a balanced detector 26.

[0078] The laser 21 is used to output continuous high-coherence laser light.

[0079] Preferably, the laser 21 is a narrow linewidth laser.

[0080] The first coupler 22 is connected to the laser 21 and is used to split the high-coherence laser light into detection light and local oscillator light.

[0081] In this embodiment, the first coupler 22 is a 1×2 coupler. In practical applications, the end with smaller optical power is used as the detection light, and the end with larger optical power is used as the local oscillator light.

[0082] The modulator 24 is connected to the first coupler 22 and is used to modulate the continuous detection light into pulsed light.

[0083] The optical circulator 25 is connected to the modulator 24 and is used to input the pulse light into the optical cable under test, so that the pulse light is scattered by Rayleigh in the optical cable under test and then passes through the optical circulator 25 again; wherein, the optical cable under test is an optical cable connecting the information demodulation equipment and the adjacent trackside equipment.

[0084] Referring to Figure 1, assuming that the current information demodulation device is located in the trackside device 1, the trackside device 2 is adjacent to the trackside device 1 and is connected to the trackside device 1 through an optical cable, the optical cable connected between the trackside device 1 and the trackside device 2 is used as the optical cable under test for inputting pulse light into the information demodulation device.

[0085] The second coupler 23 is connected to the first coupler 22 , the optical circulator 25 and the balanced detector 26 respectively, and is used to output the interference formed by the local oscillator light and the scattered light to the balanced detector 26 .

[0086] Preferably, the second coupler 23 may be a 2×2 coupler.

[0087] In this embodiment, the pulsed light is input into the optical cable under test through the optical circulator 25. The pulsed light forms scattered light after Rayleigh scattering in the optical cable under test. The scattered light passes through the optical circulator 25 again and interferes with the local oscillator light at the receiving end of the second coupler 23. The light is then input into the balanced detector 26 through the second coupler 23.

[0088] The balanced detector 26 is used to use the scattered light in the interference as the rail vibration sound wave signal, and obtain a detection result of whether the rail is broken based on the amplitude comparison result of the rail vibration sound wave signal and the local oscillation light.

[0089] In actual applications, when a train is running on rails, if the rails are broken, it will cause fluctuations in the position of the broken rails during the train's running, so that the optical cable corresponding to the broken rail position can sense this fluctuation, and the amplitude of the rail vibration sound wave signal obtained from the optical cable will change significantly. That is, the rail vibration sound wave signal can reflect the amplitude of the vibration sound received by the optical cable. Based on this amplitude, it can be determined whether the rails are broken.

[0090] Specifically, the balance detector 26 is used to:

[0091] Calculating the amplitude difference between the rail vibration sound wave signal and the local oscillation light;

[0092] When the amplitude difference exceeds a preset difference, it is determined that the rail is broken;

[0093] When the amplitude difference does not exceed the preset difference, it is determined that the rail is not broken.

[0094] The value of the preset difference is determined according to actual needs and is not limited in the present invention.

[0095] In summary, the present invention discloses an information demodulation device comprising: a laser 21, a first coupler 22, a second coupler 23, a modulator 24, an optical circulator 25, and a balanced detector 26. The highly coherent laser light output by the laser 21 is divided into a probe light and a local oscillator light by the first coupler 22. The probe light is processed by the modulator 24 to obtain a pulsed light. The pulsed light is input into the optical cable under test through the optical circulator 25. The pulsed light is scattered in the optical cable under test to form scattered light. The scattered light passes through the optical circulator 25 again and interferes with the local oscillator light at the receiving end of the second coupler 23. The scattered light is input into the balanced detector 26 through the second coupler 23. The balanced detector 26 detects whether the rail is broken by comparing the rail vibration sound wave signal with the amplitude of the local oscillator light. Therefore, the railway signal axle counting system disclosed in the present invention can not only check whether the rail section is occupied, but also automatically check whether the rail is broken, thereby facilitating the timely detection of potential dangers in the train operation process and realizing the wide application of the railway signal axle counting system.

[0096] To further optimize the above embodiment, the optical circulator 25 may also be used to obtain the time delay difference of the pulse light propagating in the optical cable under test after the pulse light is input into the optical cable under test.

[0097] The balance detector 26 is further used to determine the abnormal position of the rail according to the time delay difference.

[0098] The process of the balance detector 26 determining the abnormal position of the rail according to the time delay difference specifically includes:

[0099] Determine the propagation speed of pulsed light in the optical cable under test;

[0100] Taking the product of the propagation speed and the time delay difference to obtain a propagation distance;

[0101] The position of the axle counter sensor corresponding to the information demodulation device on the rail is taken as a starting point, and a position whose distance from the starting point is taken as the propagation distance is determined as the abnormal position of the rail.

[0102] The propagation speed of the pulse light in the optical cable under test can be determined according to actual conditions.

[0103] To further optimize the above embodiment, refer to FIG3 , which is a schematic structural diagram of another information demodulation device disclosed in an embodiment of the present invention. Based on the embodiment shown in FIG2 , the information demodulation device may further include: a memory 27 .

[0104] The memory 27 is connected to the balance detector 26 and is used to store the detection result of whether the rail is broken.

[0105] Based on the above discussion, it can be seen that the balance detector 26 can determine whether a rail is broken based on the amplitude comparison between the rail vibration acoustic wave signal and the local oscillation light. To avoid errors in the detection results caused by inaccurate rail vibration acoustic wave signals, the balance detector 26 of the present invention can also use a neural network model to further determine whether a rail is broken.

[0106] Therefore, the balanced detector 26 can also be used to:

[0107] (1) Filter the rail vibration sound wave signal to obtain the middle rail vibration sound wave signal.

[0108] In practical applications, optical information demodulation equipment utilizes distributed acoustic wave sensing technology. Because it is located trackside, it is susceptible to slight interference from the external environment, which can cause phase shifts in the optical cable being measured. Therefore, interference signals must be shielded.

[0109] The present invention utilizes a wavelet denoising method to suppress environmental noise. Based on the difference between the effective signal and the noise signal, the purpose of retaining the effective signal and removing the noise signal is achieved by selecting a threshold. The effective signal has a certain continuity, and its coefficient modulus in the wavelet domain is large, while the noise signal has no continuity, and its coefficient modulus after wavelet transformation is small. Therefore, by setting a threshold, the part above the threshold is retained, and the coefficient below the threshold is reduced or eliminated to complete the denoising. For the estimation of the threshold, a unified threshold denoising method is adopted, and the variance of the noise is used to obtain the optimal threshold. The noise is almost completely suppressed by using the wavelet transform threshold denoising method, and the characteristic peak points of the original signal can be well preserved.

[0110] In practical application scenarios, in addition to environmental noise, there is also noise generated by the vibration of internal equipment during train operation. This noise is mainly low-frequency noise, and the interference of this type of signal needs to be eliminated. The present invention proposes to use digital filtering methods for denoising. Combined with the typical structure of optical information demodulation equipment, it can be seen that the received data is a two-dimensional matrix. The present invention selects the order of the FIR (Finite Impulse Response, filter). An FIR filter of order (15) is used to achieve better filtering performance.

[0111] Preferably, a band-pass filter is selected as the filter type, and the filtering range is 1-500 Hz.

[0112] (2) performing wavelet decomposition and reconstruction on the middle rail vibration sound wave signal, removing the overall trend quantity in the middle rail vibration sound wave signal, and obtaining the target rail vibration sound wave signal.

[0113] This embodiment performs wavelet decomposition and reconstruction on the filtered intermediate rail vibration sound wave signal to remove the overall trend of the signal, thereby avoiding the increase and decrease trend of the sound information itself and the interference caused by the same type of environment.

[0114] Preferably, the present invention uses wavelet decomposition and reconstruction based on Dmey wavelet basis functions, which can focus on any details of the signal to perform multi-resolution time-frequency domain analysis.

[0115] (3) Selecting light wave signal characteristic values ​​from the target rail vibration sound wave signal.

[0116] Twelve indicators including standard deviation, variance, mean value, root mean square, kurtosis index, skewness, maximum value, root square amplitude, peak factor, margin factor, waveform factor and pulse index of the reconstructed detail signal are selected as characteristic values ​​of the light wave signal and combined with the expected results as training samples of the BP neural network.

[0117] Examples of light wave signal characteristic values ​​are shown in Table 1 below.

[0118] Table 1

[0119] (4) Inputting the characteristic value of the light wave signal into a pre-trained neural network model to obtain a detection result of whether the rail is broken or not.

[0120] Preferably, the neural network model in this embodiment is preferably a BP (Back-propagation) neural network.

[0121] To further optimize the above embodiment, the present invention also discloses a training process of a neural network model, which specifically includes:

[0122] (1) Initialize the neural network weights and neural network biases.

[0123] Neural network weights and neural network biases are two very important concepts in neural networks.

[0124] Neural network weights represent the strength of connections between neurons. Each neuron's connection to another neuron has a weight value, which determines the strength of the connection and the effectiveness of information transfer. The higher the weight value, the stronger the connection between the two neurons, and the greater the impact of the information transferred.

[0125] A neural network bias represents a neuron's activation threshold. A neuron is activated when its total input (the sum of all input connections) plus the bias is greater than a certain value (usually 0). The bias allows the neuron to activate even without any input.

[0126] Here's a simple example:

[0127] Suppose there are two neurons, neuron A and neuron B. The connection weight between them is 0.7, which means that the connection between them is relatively strong.

[0128] The input value of neuron A is 0.6 and the bias is 0.2;

[0129] Neuron B has an input value of 0.3 and a bias of 0.1.

[0130] So:

[0131] The total input of neuron A is the input value 0.6 + the connection weight with neuron B 0.7 * the input value of neuron B 0.3 = 0.6 + 0.21 = 0.81.

[0132] Since 0.81 + bias 0.2 > 0, neuron A is activated.

[0133] The total input of neuron B is the input value 0.3 + the connection weight with neuron A 0.7 * the input value of neuron A 0.6 = 0.3 + 0.42 = 0.72.

[0134] Since 0.72 + bias 0.1 > 0, neuron B is also activated.

[0135] As can be seen, the connection weights between neurons and their respective biases often jointly determine the activation of neurons and the signals they transmit. They are the basis for building neural networks. Through training, we can obtain the neural network weights and biases that make the neural network produce ideal outputs.

[0136] in short:

[0137] Neural network weights determine the strength of connections between neurons and are used to control the flow of information.

[0138] The neural network bias determines the activation threshold of the neuron itself and is used to control the activation of the neuron.

[0139] The use of neural network weights and neural network biases gives neural networks powerful expressive and learning capabilities.

[0140] (2) Set the number of nodes in the input layer, hidden layer, and output layer to obtain the initial neural network model.

[0141] Specifically, see Figure 4, which is a structural diagram of an initial neural network model disclosed in an embodiment of the present invention, where the number of nodes in the input layer is X1~Xn, the number of nodes in the hidden layer is Y1~Yn, and the number of nodes in the output layer is Z1~Zn.

[0142] (3) Inputting the light wave signal characteristic value sample data as input values ​​into the input layer of the initial neural network model.

[0143] (4) In the input layer, the light wave signal characteristic value sample data is accumulated according to the neural network weight and the neural network bias to obtain intermediate sample data, and the intermediate sample data is input into the hidden layer.

[0144] (5) After the hidden layer uses a transfer function to transform the intermediate sample data, it is output to the output layer.

[0145] (6) Determine the error value and node error rate of each node in the output layer.

[0146] (7) Based on the node error values ​​and the node error rates, the neural network weights and the neural network biases are updated.

[0147] (8) Determine whether the current model iteration number reaches the preset iteration number.

[0148] (9) If yes, the neural network model training is completed.

[0149] (10) If not, return to step 1 and input the light wave signal characteristic value sample data again.

[0150] It should be noted that the training process of the neural network model is divided into two stages. The first stage is the forward propagation of the signal, from the input layer through the hidden layer, and finally to the output layer; the second stage is the back propagation of the error, from the output layer to the hidden layer, and finally to the input layer, adjusting the weights and bias from the hidden layer to the output layer, and the weights and bias from the input layer to the hidden layer in turn.

[0151] The accuracy of the neural network model will be greatly improved after multiple training. Through implementation, it can be seen that the prediction accuracy of the neural network model in the present invention reaches more than 96% after 6000 iterations of training. The training results are shown in Figure 5.

[0152] In this embodiment, after the rail vibration sound wave signal is decomposed and reconstructed by wavelet decomposition, the 12 extracted eigenvalues ​​are used as the input values ​​of the neural network model, so the number of nodes in the input layer is 12, the number of nodes in the hidden layer is set to 6, the number of nodes in the output layer is 5, the learning rate is set to 0.0075, the number of iterations is 10000, tanh is used as the activation function of the first layer, sigmoid is used as the activation function of the second layer, and cross-entropy is used as the cost function. The output layer 00000 represents that the rail is normal, 00001 represents that the rail is 10 meters away from the sending end and the rail is broken or partially broken, 00010 represents that there is no rail broken or partially broken at a distance of 20 meters from the sending end, 00011 represents that the rail is broken or partially broken at a distance of 30 meters from the sending end, and so on.

[0153] To further optimize the above embodiment, the information demodulation device may also be used for:

[0154] When a rail break is detected, the rail break information is sent to the indoor host, which then outputs an alarm message so that technical personnel can take effective measures in a timely manner.

[0155] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0156] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0157] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A railway signal axle counter system, characterized in that, Including: Axle-counting sensors, trackside equipment, and an indoor host. The trackside equipment is respectively connected to the axle-counting sensors and the indoor host. The trackside equipment includes an axle-counting trackside device and an information demodulation device connected to each other. The number of both the axle-counting sensors and the trackside equipment is N, and N is a positive integer; The axle-counting sensors are arranged on the rails, used to sense the wheel state when a train passes, identify the number of axles and the train running direction when the train passes, and output in the form of an analog signal to the axle-counting trackside device. Among them, two adjacent axle-counting sensors are respectively arranged at both ends of the same rail section; The axle-counting trackside device is communicatively connected to the indoor host, used to analyze the analog signal to obtain trackside axle-counting information, and transmit the trackside axle-counting information to the indoor host. The trackside axle-counting information includes: the number of axles and the train running direction; The indoor host is used to determine the state of each rail section as an idle state or an occupied state according to the trackside axle-counting information transmitted by each axle-counting trackside device and the rail section configuration information. Among them, the rail section configuration information records the axle-counting sensors arranged at both ends of each rail section; The information demodulation device is used to adopt the distributed acoustic wave sensor technology, detect the rail vibration acoustic wave signal by using the backward scattering power of the optical cable, and obtain the detection result of whether the rail is broken based on the rail vibration acoustic wave signal.

2. The axle counter system for railway signals according to claim 1, wherein The information demodulation device includes: a laser, a first coupler, a second coupler, a modulator, an optical circulator, and a balanced detector; The laser is used to output continuous high-coherence laser light; The first coupler is connected to the laser and is used to divide the high-coherence laser light into probe light and local oscillator light; The modulator is connected to the first coupler and is used to modulate the continuous probe light into pulsed light; The optical circulator is connected to the modulator and is used to input the pulsed light into the optical cable to be measured, so that the pulsed light forms scattered light after Rayleigh scattering in the optical cable to be measured and then passes through the optical circulator again; among them, the optical cable to be measured is the optical cable connecting the information demodulation device and the adjacent trackside equipment; The second coupler is respectively connected to the first coupler, the optical circulator, and the balanced detector, and is used to output the interference formed by the local oscillator light and the scattered light to the balanced detector; The balanced detector is used to use the scattered light in the interference as the rail vibration acoustic wave signal, and obtain the detection result of whether the rail is broken based on the amplitude comparison result between the rail vibration acoustic wave signal and the local oscillator light.

3. The axle counter system for railway signals according to claim 2, characterized in that, The balanced detector is specifically used for: Calculating the amplitude difference between the rail vibration acoustic wave signal and the local oscillator light; When the amplitude difference exceeds a preset difference, it is determined that the rail is broken; When the amplitude difference does not exceed the preset difference, it is determined that the rail is not broken.

4. The axle counter system for railway signals according to claim 2, characterized in that, The optical circulator is further used for: after inputting the pulsed light into the optical cable to be measured, obtaining the time delay difference of the pulsed light propagating in the optical cable to be measured; The balance detector is further configured to: determine the abnormal position of the rail according to the time delay difference.

5. The axle counter system for railway signals according to claim 4, wherein, The balance detector is further configured to: determine the propagation speed of the pulsed light in the optical cable under test; multiply the propagation speed by the time delay difference to obtain the propagation distance; starting from the position of the axle counter sensor corresponding to the information demodulation device on the rail, determine the position of the rail where the distance from the starting point is the propagation distance as the abnormal position of the rail.

6. The axle counter system for railway signals according to claim 2, wherein The balance detector is further configured to: filter the rail vibration acoustic wave signal to obtain an intermediate rail vibration acoustic wave signal; perform wavelet decomposition and reconstruction on the intermediate rail vibration acoustic wave signal, and remove the overall trend quantity in the intermediate rail vibration acoustic wave signal to obtain a target rail vibration acoustic wave signal; select the optical wave signal eigenvalue from the target rail vibration acoustic wave signal; input the optical wave signal eigenvalue into a pre-trained neural network model to obtain a detection result of whether the rail is broken or not.

7. The axle counter system for railway signals according to claim 6, characterized in that, The training process of the neural network model includes: initializing the neural network weights and neural network biases; setting the number of nodes in the input layer, hidden layer, and output layer to obtain an initial neural network model; inputting the optical wave signal eigenvalue sample data as input values into the input layer of the initial neural network model; in the input layer, adding the optical wave signal eigenvalue sample data to the neural network bias according to the neural network weights to obtain intermediate sample data, and inputting the intermediate sample data into the hidden layer; after transforming the intermediate sample data using a transfer function in the hidden layer, outputting it to the output layer; determining the error values and error rates of each node in the output layer; updating the neural network weights and neural network biases based on the error values and error rates of each node; judging whether the current model iteration times reach the preset iteration times; if so, complete the training of the neural network model; if not, return to the step to input the optical wave signal eigenvalue sample data again.

8. The axle counter system for railway signals according to claim 2, wherein The information demodulation device further includes: a memory; the memory is connected to the balance detector and is used to store the detection result of whether the rail is broken or not.

9. The axle counter system for railway signals according to claim 1, characterized in that, The information demodulation device is further configured to: when it is determined that the rail is broken, send the broken rail information to the indoor host, and the indoor host outputs an alarm message.

Images