CN110660184A - Adaboost-based railway perimeter early warning method of fiber laser radar - Google Patents

Abstract

The invention belongs to the technical field of railway safety monitoring, and particularly relates to a railway perimeter early warning method based on an Adaboost fiber laser radar, which comprises the following steps: s1, injecting pulse light into the sensing optical fiber; s2, receiving the backscattered light in the sensing optical fiber by using a coherent demodulation module; s3, light intensity signals received by the coherent demodulation module are collected through a collection card and are further processed by an upper computer, high-sensitivity sensing of the surrounding environment along the railway track is achieved through a mode of combining an optical fiber coherent Rayleigh principle and heterodyne detection, the precision is high, the structure is simple, the remote monitoring cost is low, the operation is convenient, the communication optical cable has double functions of sensing and transmission, third-party construction or detection of man-made and wild animal invasion events near the track is achieved through the optical fiber Rayleigh scattering and phase demodulation principle, and the positioning precision of construction events can reach 100 meters.

Description

A Railway Perimeter Early Warning Method Based on Adaboost Fiber Lidar

technical field

The invention belongs to the technical field of railway safety monitoring, in particular to a railway perimeter early warning method based on Adaboost fiber laser radar.

Background technique

With the leap-forward development of my country's railways, the degree of high-speed passenger transport and heavy-duty freight has been continuously improved, which poses new challenges to the railway running safety system. The railway perimeter protection and monitoring system is one of the important subsystems of the railway disaster prevention and safety monitoring system. It is mainly used to monitor the surrounding environment of high-speed railways and the handling of foreign matter intrusion in important stations and railway bridge crossings. The ability to observe and analyze the content of the scene, realize the automation and intelligence of monitoring, and has shown great development potential in railway engineering applications. The front-end equipment of the existing perimeter intrusion warning system is embedded with human image recognition algorithms, which can accurately detect personnel. , tracking, to achieve human body detection, analysis and identification, real-time warning of personnel intrusion events in the perimeter area. When a suspicious person enters the monitoring range, it can be automatically identified, that is, it can be captured and the image is transmitted to the management center, and an alarm signal is output in the management center, but the detection effect is not good.

Some intrusion incidents, such as illegal destruction or crossing of railway guardrails, placing foreign objects on the track, stealing cables along the railway line and other dangerous or malicious behaviors, will cause serious accidents to the safe operation of railways. Of course, environmental factors will also bring huge hidden dangers to railway safety, such as mudslides, landslides, collapses, earthquakes, etc., which will cause damage to the protection of the railway perimeter, and third-party factors such as manual construction, excavator operations, etc., although not malicious However, it may still bring security risks to the optical cables along the railway.

For the above-mentioned hidden dangers that may cause damage to the railway or the optical cable along the railway, we need an effective monitoring method to dynamically detect the safety status of the optical cable to ensure the safety of the railway communication system.

SUMMARY OF THE INVENTION

The present invention provides a railway perimeter early warning method based on Adaboost fiber laser radar, so as to solve the problems raised in the above background technology.

In order to achieve the above object, the present invention provides the following technical solutions: a railway perimeter early warning method based on Adaboost's fiber laser radar, comprising the following steps:

S1. Inject pulsed light into the sensing fiber;

S2. Use a coherent demodulation module to receive the backscattered light in the sensing fiber;

S3. Collect the light intensity signal received by the coherent demodulation module through the acquisition card, and hand it over to the host computer for further processing;

In S3, the local oscillator light interferes with the signal light backscattered by the optical fiber, is photoelectrically converted by the detector, amplified, and the difference calculation is performed on the Rayleigh signal curve before and after, and the position where the interference light intensity signal changes on the difference curve ;

For the photoelectrically converted signal of the detector, according to the time domain amplitude and duration of the signal, the frequency domain distribution and spectrum duration of the signal, and the characteristics of the signal frequency changing with time, the Adaboost algorithm is used, which is based on multiple weak classifiers. In the above, the spectrum and time-frequency comprehensive information features of threatening railway events are extracted and identified.

Preferably, in S1, the pulsed light is modulated and converted by a laser through an acousto-optic modulator (AOM), and the pulsed light is then amplified by an erbium-doped fiber amplifier (EDFA), and injected into the sensing fiber through a circulator.

Preferably, the laser is a narrow linewidth laser.

Preferably, in S2, during the forward propagation of the received pulse light along the optical fiber by the coherent demodulation module, due to the uneven distribution of the medium in the optical fiber, backward Rayleigh scattered light will be generated.

Preferably, due to the uneven distribution of the medium in the optical fiber, backward Rayleigh scattered light will be generated and propagated in the opposite direction along the sensing optical fiber to be received by the coherent demodulation module through the circulator.

Preferably, the position where the interference light intensity signal changes on the differential curve is displayed on the display screen by a waterfall chart through software.

Preferably, multiple weak classifiers are initially trained first. If a weak classifier has a high accuracy rate, its weight will be higher, otherwise, its weight will be lower.

Preferably, the initial training is divided into several rounds, and the classification error rate (weight error function) of each weak classifier on the training data set in this round is calculated: the weight error function focuses on the weight distribution of the data set in this round, not on the weight distribution of the data set in this round. The parameters inside the weak classifier; the error of the high probability distribution in this round (the data of focus) will be given a greater penalty; the model coefficient calculated according to the classification error of the data set by the weak classifier in this round: represents the current round. The importance of the weak classifier obtained in this round; the smaller the classification error rate in this round, the greater the role of the basic classifier in the final classifier; update the weight distribution of the next round of training data sets; in this round of training , the weights of misclassified samples by the basic classifier can be expanded, while the weights of correctly classified samples are reduced in the next round.

Compared with the prior art, the beneficial effects of the present invention are:

1. The present invention realizes highly sensitive perception of the surrounding environment along the railway track through the combination of optical fiber coherent Rayleigh principle and heterodyne detection, with high precision, simple structure, low cost for long-distance monitoring, and convenient operation.

2. The communication optical cable takes into account the dual functions of sensing and transmission. Through the principle of optical fiber Rayleigh scattering and phase demodulation, it realizes the detection of third-party construction or human and wild animal intrusion events near the track, and the positioning accuracy of construction events can reach 100 meters.

3. The technical method adopted in the present invention can realize the continuous monitoring of the speed of the train by means of a waterfall diagram.

4. The sensor is not charged and can be used in the strong electromagnetic radiation environment of the railway environment.

5. Adaboost has high classification accuracy as a classifier, which is beneficial to improve the accurate detection of threatening railway safety events.

6. Under the framework of Adaboost, various regression classification models can be used to construct weak learners, which is very flexible.

7. When used as a simple binary classifier, the structure is simple, compared with BP neural network, SVM support vector machine and other algorithms, it is not easy to over-fit, thus improving the recognition accuracy.

Description of drawings

The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the specification, and are used to explain the present invention together with the embodiments of the present invention, and do not constitute a limitation to the present invention. In the attached image:

Fig. 1 is the structural representation of the present invention;

FIG. 2 is a time-domain waterfall diagram (s/km) of the passing of the passenger car in Embodiment 1 of the present invention;

3 is a time-domain waterfall diagram (s/m) of a passenger car passing through in Embodiment 1 of the present invention;

4 is a time-domain signal and a short-time Fourier analysis spectrogram when a truck passes through in Embodiment 2 of the present invention;

5 is a waterfall diagram of excavator signal detection in Embodiment 3 of the present invention;

6 is a schematic diagram of a model for identifying the intrusion signal at the railway perimeter using the ADaboost algorithm according to the present invention;

FIG. 7 is a flowchart of the processing of the railway perimeter intrusion event by the Adaboost algorithm of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Example 1

1-3, the present invention provides the following technical solutions: a railway perimeter early warning method based on Adaboost fiber laser radar, comprising the following steps:

S1. Inject pulsed light into the sensing fiber;

S2. Use a coherent demodulation module to receive the backscattered light in the sensing fiber;

S3. Collect the light intensity signal received by the coherent demodulation module through the acquisition card, and hand it over to the upper computer for further processing.

Specifically, in S1, the pulsed light is modulated and converted by a laser through an acousto-optic modulator (AOM), and the pulsed light is then amplified by an erbium-doped fiber amplifier (EDFA), and injected into the sensing fiber through a circulator; The laser adopts a narrow linewidth laser; in S2, the coherent demodulation module receives the pulsed light in the process of forward propagation along the fiber, due to the uneven medium distribution in the fiber, back Rayleigh scattering light will be generated; due to the uneven distribution in the fiber In S3, the local oscillator light interferes with the signal light backscattered back through the fiber, and the detector is received by the coherent demodulation module. Photoelectric conversion, amplification processing, the difference calculation is performed on the Rayleigh signal curve before and after the time, the position where the interference light intensity signal changes on the differential curve; the position where the interference light intensity signal changes on the differential curve is displayed on the display screen through the software. display.

In this embodiment, the continuous light wave output by the continuous narrow linewidth laser is modulated and converted into pulsed light by an acousto-optic modulator (AOM), then the power is amplified by an erbium-doped fiber amplifier (EDFA), and injected into the sensing fiber through a circulator . During the forward propagation of the pulsed light along the fiber, backward Rayleigh scattering light will be generated due to the uneven distribution of the medium in the fiber. The scattered light propagates backward along the sensing fiber and is received by the coherent demodulation module through the circulator. The light intensity signal is collected by the acquisition card, and then sent to the upper computer for further processing.

Fiber laser radar uses ultra-narrow linewidth lasers to achieve the interference effect between back Rayleigh scattered light within the pulse width range. When there is external intrusion interference somewhere along the fiber line, the refractive index of the fiber at the corresponding position will change. , resulting in a change in the optical phase at this position. Due to the change of the phase of the interference effect, the intensity of the back-scattered Rayleigh light will change. The local oscillator light interferes with the signal light backscattered by the fiber, and is converted into photoelectricity by the detector and amplified. The difference calculation is performed on the Rayleigh signal curve before and after the time, and the position where the interference light intensity signal changes on the difference curve corresponds to the position where the disturbance occurs.

In this embodiment, a feature library of railway perimeter signals is established by collecting signals of a large number of threat events and noise events. Threat events mainly include: third-party construction around the railway, illegal entry into the railway perimeter area, and jumping over the railway fence; noise events mainly include: noise from highways and other railways along the railway perimeter, and small animal invasion along the railway.

The railway safety threat events and noise events can be described by the following signal characteristics: 1. The time domain amplitude and duration of the signal, the frequency domain distribution and spectrum duration of the signal, and the characteristics of the signal frequency changing with time;

Using the Adaboost algorithm, based on multiple weak classifiers, the spectrum and time-frequency comprehensive information features of threatening railway events are extracted and identified, as shown in Figure 6;

AdaBoost algorithm is used for the detection of railway perimeter threat events. Threat events at the railway perimeter mainly refer to third-party construction and illegal climbing along the railway line; the main sources of noise interference include: river transmission and crossing along the railway line, parallel highway vehicle vibration, agricultural machinery farming, etc. The specific implementation method is divided into the following steps, learning a series of weak classifiers or basic classifiers from the database along the railway line collected by the lidar system, and linearly combining these weak classifiers into a strong classifier.

According to the AdaBoost algorithm, if the accuracy of a base classifier is high, then its weight will be higher, and vice versa. First, initialize the collected intrusion vibration signals, and the weight distribution of the data (N represents the number of samples):

D={w11,w1i...w1N},i=1,2...N

1. It is assumed that these vibration signal data have a uniform weight distribution, that is, each training sample has the same role in the learning of the basic classifier. This assumption ensures that the first step can learn the basic classifier G1(x) on the original train signal. ;

2. Assuming that the number of training rounds for railway intrusion signals is M (until a predetermined small enough error rate is reached or the maximum number of iterations specified in advance is reached), the following processing is performed for m=1, 2, 3...M :

a Use the training data set with the weight distribution Dm (the data set corresponding to the weight distribution of this round) to learn to obtain the basic classifier of this round.

b Calculate the classification error rate (weight error function) of each weak classifier on the current round of training data sets: The weight error function focuses on the weight distribution of the current round of data sets, rather than the parameters inside the weak classifier. Errors with high probability distributions (data of focus) in this round are given a larger penalty.

c Model coefficient calculated according to the classification error of the data set by the weak classifier in this round: it represents the importance of the weak classifier obtained in this round. The smaller the classification error rate in this round, the greater the role of the basic classifier in the final classifier.

d Update the weight distribution of the training dataset for the next round. In this round of training, the weights of misclassified samples by the basic classifier are enlarged, while the weights of correctly classified samples are reduced in the next round. Comparing the two, the weights of the misclassified intrusion signal samples are enlarged, so the misclassified samples play a greater role in the next round of learning. It does not change the given training data itself, but constantly changes the distribution of the weights of the training data, so that the training data is continuously optimized for weak classification in the learning of the basic classifier, so as to finally achieve the purpose of a strong classification model.

See Figure 7;

In each round of training, the weight distribution of the training samples is constantly changing, and at the same time 1. The weight distribution plays a proportional role in the importance of the weak classifier in this round in the final linear classifier combination; 2. The next round of The sample weight adjustment works inversely proportional.

Figure 2 shows that there is a train signal about 6.5 kilometers away from the first station. The running track of the train can be seen by zooming in on Figure 3, and the speed of the train can be calculated from the slope of the track.

Example 2

Please refer to Figures 1 and 4, the present invention provides the following technical solutions: a railway perimeter early warning method based on Adaboost's fiber laser radar, comprising the following steps:

S1. Inject pulsed light into the sensing fiber;

S2. Use a coherent demodulation module to receive the backscattered light in the sensing fiber;

S3. Collect the light intensity signal received by the coherent demodulation module through the acquisition card, and hand it over to the upper computer for further processing.

Specifically, in S1, the pulsed light is modulated and converted by a laser through an acousto-optic modulator (AOM), and the pulsed light is then amplified by an erbium-doped fiber amplifier (EDFA), and injected into the sensing fiber through a circulator; The laser adopts a narrow linewidth laser; in S2, the coherent demodulation module receives the pulsed light in the process of forward propagation along the fiber, due to the uneven medium distribution in the fiber, back Rayleigh scattering light will be generated; due to the uneven distribution in the fiber In S3, the local oscillator light interferes with the signal light backscattered back through the fiber, and the detector is received by the coherent demodulation module. Photoelectric conversion, amplification processing, the difference calculation is performed on the Rayleigh signal curve before and after the time, the position where the interference light intensity signal changes on the differential curve; the position where the interference light intensity signal changes on the differential curve is displayed on the display screen through the software. display.

In this embodiment, the continuous light wave output by the continuous narrow linewidth laser is modulated and converted into pulsed light by an acousto-optic modulator (AOM), then the power is amplified by an erbium-doped fiber amplifier (EDFA), and injected into the sensing fiber through a circulator . During the forward propagation of the pulsed light along the fiber, backward Rayleigh scattering light will be generated due to the uneven distribution of the medium in the fiber. The scattered light propagates backward along the sensing fiber and is received by the coherent demodulation module through the circulator. The light intensity signal is collected by the acquisition card, and then sent to the upper computer for further processing.

Fiber laser radar uses ultra-narrow linewidth lasers to achieve the interference effect between back Rayleigh scattered light within the pulse width range. When there is external intrusion interference somewhere along the fiber line, the refractive index of the fiber at the corresponding position will change. , resulting in a change in the optical phase at this position. Due to the change of the phase of the interference effect, the intensity of the back-scattered Rayleigh light will change. The local oscillator light interferes with the signal light backscattered by the fiber, and is converted into photoelectricity by the detector and amplified. The difference calculation is performed on the Rayleigh signal curve before and after the time, and the position where the interference light intensity signal changes on the difference curve corresponds to the position where the disturbance occurs.

In this embodiment, a feature library of railway perimeter signals is established by collecting signals of a large number of threat events and noise events. Threat events mainly include: third-party construction around the railway, illegal entry into the railway perimeter area, and jumping over the railway fence; noise events mainly include: noise from highways and other railways along the railway perimeter, and small animal invasion along the railway.

The railway safety threat events and noise events can be described by the following signal characteristics: 1. The time domain amplitude and duration of the signal, the frequency domain distribution and spectrum duration of the signal, and the characteristics of the signal frequency changing with time;

Using the Adaboost algorithm, based on multiple weak classifiers, the spectrum and time-frequency comprehensive information features of threatening railway events are extracted and identified, as shown in Figure 6;

AdaBoost algorithm is used for the detection of railway perimeter threat events. Threat events at the railway perimeter mainly refer to third-party construction and illegal climbing along the railway line; the main sources of noise interference include: river transmission and crossing along the railway line, parallel highway vehicle vibration, agricultural machinery farming, etc. The specific implementation method is divided into the following steps, learning a series of weak classifiers or basic classifiers from the database along the railway line collected by the lidar system, and linearly combining these weak classifiers into a strong classifier.

According to the AdaBoost algorithm, if the accuracy of a base classifier is high, then its weight will be higher, and vice versa. First, initialize the collected intrusion vibration signals, and the weight distribution of the data (N represents the number of samples):

D={w11,w1i...w1N},i=1,2...N

1. It is assumed that these vibration signal data have a uniform weight distribution, that is, each training sample has the same role in the learning of the basic classifier. This assumption ensures that the first step can learn the basic classifier G1(x) on the original train signal. ;

2. Assuming that the number of training rounds for railway intrusion signals is M (until a predetermined small enough error rate is reached or the maximum number of iterations specified in advance is reached), the following processing is performed for m=1, 2, 3...M :

a Use the training data set with the weight distribution Dm (the data set corresponding to the weight distribution of this round) to learn to obtain the basic classifier of this round.

b Calculate the classification error rate (weight error function) of each weak classifier on the current round of training data sets: The weight error function focuses on the weight distribution of the current round of data sets, rather than the parameters inside the weak classifier. Errors with high probability distributions (data of focus) in this round are given a larger penalty.

c Model coefficient calculated according to the classification error of the data set by the weak classifier in this round: it represents the importance of the weak classifier obtained in this round. The smaller the classification error rate in this round, the greater the role of the basic classifier in the final classifier.

d Update the weight distribution of the training dataset for the next round. In this round of training, the weights of misclassified samples by the basic classifier are enlarged, while the weights of correctly classified samples are reduced in the next round. Comparing the two, the weights of the misclassified intrusion signal samples are enlarged, so the misclassified samples play a greater role in the next round of learning. It does not change the given training data itself, but constantly changes the distribution of the weights of the training data, so that the training data is continuously optimized for weak classification in the learning of the basic classifier, so as to finally achieve the purpose of a strong classification model.

See Figure 7;

In each round of training, the weight distribution of the training samples is constantly changing, and at the same time 1. The weight distribution plays a proportional role in the importance of the weak classifier in this round in the final linear classifier combination; 2. The next round of The sample weight adjustment works inversely proportional.

Figure 4 shows the vibration signal of the optical fiber when the freight train passes by. The left side is the original signal, and the right side is the time domain differential signal. The signal is periodically segmented, and the period length is about 1s. Considering the train speed is about 15m/s, 54km/h, The corresponding length is 15 meters, which is equivalent to the length of a general freight train carriage, so the period should be caused by the vibration of the connection between the carriages.

Example 3

Please refer to Figures 1 and 5, the present invention provides the following technical solutions: a railway perimeter early warning method based on Adaboost's fiber laser radar, comprising the following steps:

S1. Inject pulsed light into the sensing fiber;

S2. Use a coherent demodulation module to receive the backscattered light in the sensing fiber;

S3. Collect the light intensity signal received by the coherent demodulation module through the acquisition card, and hand it over to the upper computer for further processing.

Specifically, in S1, the pulsed light is modulated and converted by a laser through an acousto-optic modulator (AOM), and the pulsed light is then amplified by an erbium-doped fiber amplifier (EDFA), and injected into the sensing fiber through a circulator; The laser adopts a narrow linewidth laser; in S2, the coherent demodulation module receives the pulsed light in the process of forward propagation along the fiber, due to the uneven medium distribution in the fiber, back Rayleigh scattering light will be generated; due to the uneven distribution in the fiber In S3, the local oscillator light interferes with the signal light backscattered back through the fiber, and the detector is received by the coherent demodulation module. Photoelectric conversion, amplification processing, the difference calculation is performed on the Rayleigh signal curve before and after the time, the position where the interference light intensity signal changes on the differential curve; the position where the interference light intensity signal changes on the differential curve is displayed on the display screen through the software. display.

In this embodiment, a feature library of railway perimeter signals is established by collecting signals of a large number of threat events and noise events. Threat events mainly include: third-party construction around the railway, illegal entry into the railway perimeter area, and jumping over the railway fence; noise events mainly include: noise from highways and other railways along the railway perimeter, and small animal invasion along the railway.

The railway safety threat events and noise events can be described by the following signal characteristics: 1. The time domain amplitude and duration of the signal, the frequency domain distribution and spectrum duration of the signal, and the characteristics of the signal frequency changing with time;

Using the Adaboost algorithm, based on multiple weak classifiers, the spectrum and time-frequency comprehensive information features of threatening railway events are extracted and identified, as shown in Figure 6;

AdaBoost algorithm is used for the detection of railway perimeter threat events. Threat events at the railway perimeter mainly refer to third-party construction and illegal climbing along the railway line; the main sources of noise interference include: river transmission and crossing along the railway line, parallel highway vehicle vibration, agricultural machinery farming, etc. The specific implementation method is divided into the following steps, learning a series of weak classifiers or basic classifiers from the database along the railway line collected by the lidar system, and linearly combining these weak classifiers into a strong classifier.

According to the AdaBoost algorithm, if the accuracy of a base classifier is high, its weight will be higher, and vice versa. First, initialize the collected intrusion vibration signals for training, and the weight distribution of the data (N represents the number of samples):

D={w11,w1i...w1N},i=1,2...N

1. It is assumed that these vibration signal data have a uniform weight distribution, that is, each training sample has the same role in the learning of the basic classifier. This assumption ensures that the first step can learn the basic classifier G1(x) on the original train signal. ;

2. Assuming that the number of training rounds for railway intrusion signals is M (until a predetermined small enough error rate is reached or the maximum number of iterations specified in advance is reached), the following processing is performed for m=1, 2, 3...M :

a Use the training data set with the weight distribution Dm (the data set corresponding to the weight distribution of this round) to learn to obtain the basic classifier of this round.

b Calculate the classification error rate (weight error function) of each weak classifier on the current round of training data sets: The weight error function focuses on the weight distribution of the current round of data sets, rather than the parameters inside the weak classifier. Errors with high probability distributions (data of focus) in this round are given a larger penalty.

c Model coefficient calculated according to the classification error of the data set by the weak classifier in this round: it represents the importance of the weak classifier obtained in this round. The smaller the classification error rate in this round, the greater the role of the basic classifier in the final classifier.

d Update the weight distribution of the training dataset for the next round. In this round of training, the weights of misclassified samples by the basic classifier are enlarged, while the weights of correctly classified samples are reduced in the next round. Comparing the two, the weights of the misclassified intrusion signal samples are enlarged, so the misclassified samples play a greater role in the next round of learning. It does not change the given training data itself, but constantly changes the distribution of the weights of the training data, so that the training data is continuously optimized for weak classification in the learning of the basic classifier, so as to finally achieve the purpose of a strong classification model.

See Figure 7;

In each round of training, the weight distribution of the training samples is constantly changing, and at the same time 1. The weight distribution plays a proportional role in the importance of the weak classifier in this round in the final linear classifier combination; 2. The next round of The sample weight adjustment works inversely proportional.

Figure 5 is a continuous waterfall diagram of the ground vibration signal along the 1500-meter-long optical cable collected by the fiber laser radar early warning system with time. It can be seen that there is a continuous strong vibration signal 250 meters away from the starting end of the system, which is an excavator. Operation events, through the joint amplitude of the signal in the time domain and frequency domain, the frequency distribution changes, determine the intrusion behavior.

Finally, it should be noted that the above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, the The technical solutions described in the foregoing embodiments may be modified, or some technical features thereof may be equivalently replaced. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims (7)

1. a railway perimeter early warning method based on the fiber laser radar of Adaboost, is characterized in that: comprise the following steps: S1. Inject pulsed light into the sensing fiber; S2. Use a coherent demodulation module to receive the backscattered light in the sensing fiber; S3. Collect the light intensity signal received by the coherent demodulation module through the acquisition card, and hand it over to the host computer for further processing; In S3, the local oscillator light interferes with the signal light backscattered by the optical fiber, is photoelectrically converted by the detector, amplified, and the difference calculation is performed on the Rayleigh signal curve before and after, and the position where the interference light intensity signal changes on the difference curve ; For the photoelectrically converted signal of the detector, according to the time domain amplitude and duration of the signal, the frequency domain distribution and spectrum duration of the signal, and the characteristics of the signal frequency changing with time, the Adaboost algorithm is used, which is based on multiple weak classifiers. In the above, the spectrum and time-frequency comprehensive information features of threatening railway events are extracted and identified. 2. a kind of railway perimeter early warning method based on the fiber laser radar of Adaboost according to claim 1, is characterized in that: in S1, pulsed light is converted into by laser through acousto-optic modulator (AOM) modulation, and The pulsed light is then amplified by an erbium-doped fiber amplifier (EDFA) and injected into the sensing fiber through a circulator. 3. a kind of railway perimeter early warning method based on the fiber laser radar of Adaboost according to claim 1, is characterized in that: in S2, the coherent demodulation module receives the pulse light in the forward propagation process along the fiber, because the fiber Inhomogeneous medium distribution in the medium will produce back Rayleigh scattered light. 4. a kind of railway perimeter early warning method based on the fiber laser radar of Adaboost according to claim 1, is characterized in that: the position where the interference light intensity signal changes on the differential curve is displayed by software on the display screen by waterfall chart . 5. A kind of railway perimeter early warning method based on Adaboost fiber laser radar according to claim 1, it is characterized in that: carry out initial training to a plurality of weak classifiers, if the accuracy rate of a weak classifier is high, then it The weight will be higher, otherwise the weight will be lower. 6. A kind of railway perimeter early warning method based on Adaboost's fiber laser radar according to claim 5, it is characterized in that: initial training is divided into several rounds, calculate the classification of each weak classifier on the training data set of this round Error rate: The weight error function focuses on the weight distribution of the data set in this round, and does not pay attention to the parameters inside the weak classifier. 7. A kind of railway perimeter early warning method based on Adaboost's fiber laser radar according to claim 6, is characterized in that: the error of this round of high probability distribution will be given greater punishment; according to the weak classification of this round The model coefficient calculated by the classifier on the classification error of the data set: it represents the importance of the weak classifier obtained in this round; the smaller the classification error rate in this round, the greater the role of the basic classifier in the final classifier; update the next The weight distribution of the training data set for one round; in this round of training, the weights of the samples misclassified by the basic classifier are enlarged, while the weights of the correctly classified samples are reduced in the next round.

Images