A rail abnormal point detection system, method and device

Abstract

The present invention provides a system, method and device for detecting abnormal points on rails, including an acquisition module, an analysis module and a judgment module. The acquisition module is used to collect and store the real-time temperature on the rail, the propagation distance between the transducer emission point and the corresponding receiving point, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength. The analysis module is used to obtain the corresponding fifth signal strength and sixth signal strength according to the real-time temperature, the propagation distance between the transducer emission point and the corresponding receiving point, the first signal strength and the second signal strength. The judgment module compares the second signal strength and the fifth signal strength, and compares the fourth signal strength and the sixth signal strength at the same time. If the second signal strength is less than the fifth signal strength and the fourth signal strength is less than the sixth signal strength, it is judged that there is an abnormal point on the rail. The present invention improves the efficiency, accuracy, reliability and real-time performance in rail detection and maintenance work, and effectively prevents the occurrence of traffic accidents.

Description

Technical Field

[0001] The present invention relates to the field of rail transportation technology, and in particular to a rail abnormal point detection system, method and device. Background Art

[0002] Accurate detection of abnormal points on rails is an indispensable task in the operation of subway maintenance companies. Its core lies in timely discovering and confirming hidden dangers of cracks inside or on the surface of the rails. If such cracks are not dealt with in time, they are likely to cause serious traffic accidents and pose a major threat to public safety.

[0003] At present, the Chinese patent with the authorization announcement number CN104020221B discloses a real-time broken rail detection and positioning method based on ultrasonic guided waves, and uses a real-time broken rail detection and positioning system based on ultrasonic guided waves, which is characterized in that a detection method is adopted in which a transmitting positioning node and a receiving positioning node are arranged alternately and a terminal node is set at one end of the detection area. On the rail, an excitation signal is transmitted to an ultrasonic probe fixed on the rail waist through the transmitting positioning node, and the excitation signal is converted into an ultrasonic guided wave signal through the piezoelectric ceramic in the ultrasonic probe. The ultrasonic guided wave signal is transmitted over a long distance of the rail and received by the ultrasonic probe at the receiving positioning node of the receiving node. The probe converts the vibration signal into an electrical signal, and the receiving board circuit then filters and amplifies the signal and analyzes the signal characteristics to determine the integrity of the rail between the receiving module and the transmitting module. However, this method of detecting cracks by unidirectionally transmitting ultrasonic guided wave signals will have a large error due to different crack positions, and the rail, as a metal material, has thermal expansion. With the change of temperature, the length, shape and internal stress of the rail will change, which will affect the propagation path and reflection characteristics of the ultrasonic guided wave, thereby affecting the accuracy of crack detection. At the same time, ultrasonic guided wave detection cannot accurately calculate the location, length and width of the crack, and there is no alarm to remind maintenance personnel to perform maintenance or inspection according to the specific situation.

[0004] Therefore, subway maintenance companies need a device that can accurately detect abnormal points on rails to meet the needs of rapid rail maintenance to ensure safe and reliable rail transportation. Summary of the invention

[0005] The technical problem to be solved by the present invention is to overcome the defects of the prior art and provide a rail abnormal point detection system, method and device to improve the efficiency, accuracy, reliability and real-time performance of rail detection and maintenance work, and effectively prevent the occurrence of traffic accidents.

[0006] The present invention provides a method for detecting abnormal points on a rail. The detection group includes a transducer A and a transducer B disposed at opposite ends of the rail. The detection method includes:

[0007] The transducers A and B at both ends of the detection group periodically transmit and receive ultrasonic signals to each other, and collect and store real-time parameters on the rail at the same time, wherein the real-time parameters include real-time temperature, propagation distance between the transducer transmitting point and the corresponding receiving point, first signal strength, second signal strength, third signal strength and fourth signal strength; the first signal strength is the signal strength transmitted by transducer A to transducer B, the second signal strength is the signal strength received by transducer B to transducer A, the third signal strength is the signal strength transmitted by transducer B to transducer A, and the fourth signal strength is the signal strength received by transducer A to transducer B; according to the Real-time temperature, propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength and the third signal strength, obtain the fifth signal strength corresponding to the first signal strength and the sixth signal strength corresponding to the third signal strength; the fifth signal strength is the signal strength when transducer B receives transducer A when the rail is in a normal state, and the sixth signal strength is the signal strength when transducer A receives transducer B when the rail is in a normal state; compare the second signal strength and the fifth signal strength, and compare the fourth signal strength and the sixth signal strength at the same time. If the second signal strength is less than the fifth signal strength and the fourth signal strength is less than the sixth signal strength, it is determined that there is an abnormal point on the rail.

[0008] The detection method further includes calculating the length and width of the abnormal point when an abnormal point exists on the rail, specifically including:

[0009] Collect several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, distance between transducer A and abnormal point, distance between transducer B and abnormal point, first signal strength, second signal strength, third signal strength, fourth signal strength, abnormal point length and abnormal point width; introduce an initial model to train a neural network model, take the historical temperature, distance between transducer A and abnormal point, distance between transducer B and abnormal point, first signal strength, second signal strength, third signal strength and fourth signal strength data at historical moments as input, take the abnormal point length and abnormal point width data corresponding to the historical moment as output, train the neural network model to obtain a model for calculating the length and width of the abnormal point of the rail at real-time temperature, the model for calculating the length and width of the abnormal point of the rail at real-time temperature is based on a formula for calculating the length and width of the abnormal point of the rail at real-time temperature, and is used to calculate the length and width of the abnormal point when the rail is in a normal state, and the formula for calculating the length and width of the abnormal point at real-time temperature is:

[0010]

[0011] Among them, x i is the input vector, including the real-time temperature, the distance between transducer A and the abnormal point, the distance between transducer B and the abnormal point, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength, n is the number of neurons, wi represents the weight of the i-th neuron, b i represents the bias of the i-th neuron, Y represents the output vector including the abnormal point length and the abnormal point width, and i refers to the i-th neuron; according to the real-time temperature abnormal point length and the abnormal point width, the rail warning risk level indication information is generated.

[0012] While adopting the above technical solutions, the present invention may also adopt or combine the following technical solutions: the detection method obtains a fifth signal strength corresponding to the first signal strength and a sixth signal strength corresponding to the third signal strength according to the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength and the third signal strength, and further includes:

[0013] Collecting several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, propagation distance between the transducer emission point and the corresponding receiving point, first signal strength, third signal strength, fifth signal strength and sixth signal strength; introducing an initial model exponential decay model, taking the historical temperature at the historical moment, propagation distance between the transducer emission point and the corresponding receiving point, first signal strength, and third signal strength data as input, taking the fifth signal strength and sixth signal strength and data corresponding to the historical moment as output, training the exponential decay model to obtain a real-time temperature normal signal strength model, the real-time temperature normal signal strength model is based on a real-time temperature normal signal strength formula, and is used to predict the corresponding normal signal strength received by the transducer when the rail is in a normal state, and the real-time temperature normal signal strength formula includes a real-time temperature fifth signal strength formula and a real-time temperature sixth signal strength formula;

[0014] The fifth signal strength formula of real-time temperature is:

[0015] E A (d) = E A0 e -α(T)L

[0016] Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point in meters (m), e is a constant, and E A0 Indicates the first signal strength, E A (d) represents the fifth signal strength, α(T) represents the attenuation coefficient of the signal transmitted from transducer A to transducer B at temperature T;

[0017] The real-time temperature sixth signal strength formula is:

[0018] E B (d) = E B0 e -α(T)L

[0019] Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point in meters (m), e is a constant, and EB0 Indicates the third signal strength, E B (d) represents the sixth signal strength, and α′(T) represents the attenuation coefficient of the signal transmitted by transducer B to transducer A at temperature T.

[0020] As a preferred technical solution of the present invention, the detection method further includes determining the position of the abnormal point when there is an abnormal point state on the rail, specifically including:

[0021] The real-time parameters on the rail are collected and stored, wherein the real-time parameters include the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength, the second signal strength, the third signal strength, and the fourth signal strength, and the attenuation coefficient of the signal transmitted by the real-time temperature transducer A to the transducer B and the attenuation coefficient of the signal transmitted by the real-time temperature transducer B to the transducer A are obtained; the mathematical model constructed by the physical mechanism of the initial model is introduced, and the real-time temperature of the real-time parameters, the distance between the transducer A and the abnormal point, the distance between the transducer B and the abnormal point, the first signal strength, the third signal strength, the temperature T The attenuation coefficient of the signal transmitted from transducer A to transducer B, the attenuation coefficient of the signal transmitted from transducer B to transducer A at temperature T, and the attenuation coefficient data of the signal passing through the abnormal point are taken as input, and the propagation distance between the transducer transmitting point and the corresponding receiving point, the second signal strength, and the fourth signal strength data are taken as output. The mathematical model constructed by training the physical mechanism is obtained to obtain a model for determining the position of the abnormal point of the rail at real-time temperature. The model for determining the position of the abnormal point of the rail at real-time temperature is based on the formula for determining the position of the abnormal point of the rail at real-time temperature, and is used to determine the position of the abnormal point of the rail. The formula for determining the position of the abnormal point of the rail at real-time temperature is:

[0022]

[0023]

[0024] L=X1+X2

[0025] Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point, in meters (m), X1 represents the distance between transducer A and the abnormal point, in meters (m), X2 represents the distance between transducer B and the abnormal point, in meters (m), e is a constant, β represents the attenuation coefficient of the signal passing through the abnormal point, E A (c) represents the second signal strength, E A0 represents the first signal strength, α(T) represents the attenuation coefficient of the signal transmitted from transducer A to transducer B at temperature T, and E B (c) represents the fourth signal strength, E B0 represents the third signal strength, and α′(T) represents the attenuation coefficient of the signal transmitted by transducer B to transducer A at temperature T.

[0026] As a preferred technical solution of the present invention, a rail abnormal point detection system of a rail abnormal point detection method includes an acquisition module, an analysis module, and a judgment module; the transducers A and B at both ends of the detection group periodically transmit and receive ultrasonic signals to each other, and the acquisition module is used to collect and store real-time parameters on the rail, wherein the real-time parameters include real-time temperature, propagation distance between the transducer emission point and the corresponding receiving point, first signal strength, second signal strength, third signal strength and fourth signal strength; the analysis module is used to obtain a fifth signal strength corresponding to the first signal strength and a sixth signal strength corresponding to the third signal strength according to the real-time temperature, the propagation distance between the transducer emission point and the corresponding receiving point, the first signal strength and the third signal strength; the judgment module is used to compare the second signal strength and the fifth signal strength, and at the same time compare the fourth signal strength and the sixth signal strength, if the second signal strength is less than the fifth signal strength and the fourth signal strength is less than the sixth signal strength, it is judged that there is an abnormal point on the rail.

[0027] As a preferred technical solution of the present invention, the detection system also includes a training module:

[0028] The acquisition module is used to collect several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, propagation distance between the transducer emission point and the corresponding receiving point, first signal strength, third signal strength, fifth signal strength and sixth signal strength; the training module introduces an initial model exponential decay model, takes the historical temperature at the historical moment, the propagation distance between the transducer emission point and the corresponding receiving point, first signal strength, third signal strength data as input, takes the fifth signal strength and sixth signal strength data corresponding to the historical moment as output, trains the exponential decay model to obtain a real-time temperature normal signal strength model, the real-time temperature normal signal strength model is based on the real-time temperature normal signal strength formula, and is used to predict the corresponding normal signal strength received by the transducer when the rail is in normal state, and the real-time temperature normal signal strength formula includes the real-time temperature fifth signal strength formula and the real-time temperature sixth signal strength formula.

[0029] As a preferred technical solution of the present invention, the detection system also includes a calculation module:

[0030] When there is an abnormal point on the rail, the acquisition module and the calculation module are also used to calculate the position of the abnormal point: the acquisition module is used to collect and store real-time parameters on the rail, wherein the real-time parameters include real-time temperature, propagation distance between the transducer transmitting point and the corresponding receiving point, first signal strength, second signal strength, third signal strength, and fourth signal strength, and obtain the attenuation coefficient of the real-time temperature transducer A transmitting the signal to transducer B, and the attenuation coefficient of the real-time temperature transducer B transmitting the signal to transducer A; the calculation module is used to introduce the mathematical model constructed by the physical mechanism of the initial model, and compare the real-time temperature of the real-time parameters, transducer A and the abnormal point. The distance between the normal point, the distance between transducer B and the abnormal point, the first signal strength, the third signal strength, the attenuation coefficient of the signal transmitted by transducer A to transducer B at temperature T, the attenuation coefficient of the signal transmitted by transducer B to transducer A at temperature T, and the attenuation coefficient data of the signal passing through the abnormal point are taken as input, and the propagation distance between the transducer transmitting point and the corresponding receiving point, the second signal strength, and the fourth signal strength data are taken as output. The mathematical model constructed by training the physical mechanism is obtained to obtain a model for determining the position of the abnormal point of the rail by real-time temperature. The model for determining the position of the abnormal point of the rail by real-time temperature is based on the formula for determining the position of the abnormal point of the rail by real-time temperature, and is used to determine the position of the abnormal point of the rail.

[0031] As a preferred technical solution of the present invention, in the detection system, when there is an abnormal point on the rail, the acquisition module, the training module and the calculation module are also used to calculate the length and width of the abnormal point:

[0032] The acquisition module is used to collect several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, distance between transducer A and the abnormal point, distance between transducer B and the abnormal point, first signal strength, second signal strength, third signal strength, fourth signal strength, abnormal point length and abnormal point width;

[0033] The training module is used to introduce the initial model to train the neural network model, take the historical temperature at the historical moment, the distance between the transducer A and the abnormal point, the distance between the transducer B and the abnormal point, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength data as input, take the length of the abnormal point and the width of the abnormal point corresponding to the historical moment as output, train the neural network model to obtain a model for calculating the length and width of the abnormal point of the rail at real-time temperature, the model for calculating the length and width of the abnormal point of the rail at real-time temperature is based on a formula for calculating the length and width of the abnormal point of the rail at real-time temperature, and is used to calculate the length and width of the abnormal point when the rail is in a normal state;

[0034] The calculation module is used to calculate the abnormal point length and width model according to the real-time temperature to obtain the abnormal point length and abnormal point width.

[0035] As a preferred technical solution of the present invention, the detection system further includes an alarm module, and the alarm module is used to generate rail early warning risk level indication information according to the real-time temperature abnormal point length and abnormal point width.

[0036] As a preferred technical solution of the present invention, a rail abnormal point detection device includes a plurality of detection groups and a terminal of the rail abnormal point detection system arranged on each side of the rail, wherein the detection group includes a transducer A, a transducer B, an analog-to-digital converter, a wireless communicator and a temperature sensor arranged at both ends; the transducer A and the transducer B are used to periodically transmit ultrasonic signals to each other and receive corresponding ultrasonic signals; the analog-to-digital converter is connected to the transducer A and the transducer B respectively, and is used to convert the various received signal strengths into digital signals and send them to the wireless communicator; the temperature sensor is used to collect the real-time temperature of the rail and convert it into a digital signal and send it to the wireless communicator; one end of the wireless communicator is connected to the temperature sensor and the other end is connected to the temperature sensor. The first end is connected to the digital-to-analog converter, and the other end is connected to the terminal, for receiving digital information and outputting digital information to the terminal; the terminal is connected to the wireless communicator, for comprehensively calculating the received real-time temperature, digital signal, initial signal strength emitted by transducer A and transducer B and the corresponding attenuated signal strength digital signal after propagation, and the propagation distance between the transducer transmitting point and the corresponding receiving point; the signal strength after propagation when the real-time temperature rail is in a normal state is obtained, and compared with the attenuated signal strength after actual propagation, to determine whether there is an abnormal point on the rail, and then calculate the position, width and length of the abnormal point on the rail at the real-time temperature; the terminal includes an acquisition module, a training module, an analysis module, a judgment module, a calculation module and an alarm module.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] The temperature sensor of the detection group in the detection device of the present invention collects temperature in real time, and transducer A and transducer B periodically and interactively transmit the initial ultrasonic signal strength and receive the attenuated signal strength after propagation. The bidirectional ultrasonic signal transmission of transducer A and transducer B can detect the rail from two directions, and can more comprehensively cover all parts of the rail, thereby improving the detection efficiency, and the bidirectional ultrasonic detection method can mutually verify the detection results, thereby improving the reliability of the detection. Through the collection of the rail temperature and the ambient temperature by the temperature sensor, whether under high or low temperature conditions, the bidirectional ultrasonic signal transmission method can effectively detect the needs and more accurately detect the cracks in the rail, which helps to timely discover potential safety hazards and ensure the safety and smoothness of railway transportation.

[0039] During the detection process at real-time temperature, the detection system will adjust and calculate the signal strength of the corresponding rail in normal state according to the real-time temperature to ensure the accuracy of judging whether there are abnormal points in the detected rails, improve the detection efficiency, and provide multiple real and effective data for the subsequent abnormal point analysis and calculation of the rail detection system, thereby further improving the accuracy of judging and identifying abnormal points on the rails. It can improve the detection accuracy, strong adaptability, reduce false alarms and missed alarms, and improve the detection efficiency. The calculation module in the detection system of the present invention can accurately calculate the specific location, length and width of the abnormal point. This precise positioning and size calculation can help maintenance personnel to quickly lock the position of the abnormal point, accurately assess the degree of damage to the abnormal point, and immediately take appropriate repair measures to effectively prevent the expansion of abnormal point damage and the occurrence of potential accidents.

[0040] The alarm module in the detection system of the present invention can automatically identify which risk level interval the length and width data of the abnormal point fall into, and generate corresponding alarms and maintenance instructions, which enhances the accuracy of maintenance personnel's rapid risk judgment. When multiple abnormal points are detected, the detection system can also prioritize the abnormal points according to the risk level interval of the abnormal points, guide maintenance personnel to efficiently arrange maintenance plans, ensure the timeliness and pertinence of maintenance work, and greatly reduce the safety risks caused by rail abnormalities. At the same time, timely rail maintenance and overhaul effectively reduces the downtime of trains caused by maintenance and improves the efficiency of subway operations.

[0041] The rail abnormal point detection system, method and device of the present invention improve the work efficiency, detection accuracy, reliability and real-time performance in rail detection and maintenance work, and effectively prevent the occurrence of traffic accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0043] Figure 1 It is a schematic diagram of the flow of the rail abnormal point detection method of the present invention;

[0044] Figure 2 It is a functional module diagram of the rail abnormal point detection system of the present invention;

[0045] Figure 3 It is a schematic diagram of the structure of the transducer of the rail abnormal point detection device of the present invention;

[0046] Figure 4 It is a structural schematic diagram of a detection group of a rail abnormal point detection device of the present invention;

[0047] Figure 5 It is a structural schematic diagram of a rail abnormal point detection device of the present invention;

[0048] In the figure: detection group 1; transducer A 11; transducer B 12; temperature sensor 13; analog-to-digital converter 14; wireless communicator 15; ultrasonic signal 16; terminal 2; acquisition module 21; training module 22; analysis module 23; judgment module 24; calculation module 25; alarm module 26; abnormal point 3. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0050] In the present technical solution, transducers A11 and B12 at both ends of the rail detection group 1 periodically transmit and receive ultrasonic signals 16 to each other. One-way ultrasonic signal transmission can only be detected from one direction, and multiple scans may be required to ensure the detection coverage of the entire rail. Moreover, one-way detection may lead to misjudgment due to factors such as equipment failure and signal interference, and may also cause detection difficulties due to changes in the attenuation of the signal in these areas and the reflection path. Bidirectional ultrasonic signal transmission can better adapt to complex detection environments. For example, when there are structural changes such as bends, joints or welds in the rails, bidirectional detection can more effectively capture and identify abnormal signals in these areas. The bidirectional ultrasonic signal transmission of transducers A and B can detect the rails from two directions, which can more comprehensively cover all parts of the rails, thereby improving the detection efficiency. Moreover, the bidirectional ultrasonic detection method can verify the detection results with each other, thereby improving the reliability of the detection. The historical parameters are used to collect historical temperatures, the propagation distance between the transducer transmitting point and the corresponding receiving point, the distance between transducer A11 and abnormal point 3, the distance between transducer B12 and abnormal point 3, the initial signal strength mutually emitted by transducer A11 and transducer B12, and the attenuated signal strength data after corresponding propagation to train the real-time temperature normal signal strength model formula, the real-time temperature determination formula of the abnormal point 3 position of the rail, and the real-time temperature calculation formula of the abnormal point 3 length and width. By bringing the data collected at the real-time temperature into the required different formulas, it is possible to determine whether there is an abnormal point 3 on the rail at the real-time temperature, determine the position of the abnormal point 3 on the rail at the real-time temperature, calculate the length and width of the abnormal point 3 at the real-time temperature, and generate the rail early warning risk level indication information through the length and width of the abnormal point 3 at the real-time temperature, so as to remind the maintenance personnel to perform maintenance or inspection, thereby avoiding damage to the rail during use and effectively improving the durability and safety of the subway rail.

[0051] In this embodiment, Figure 1 As shown, the present invention provides a method for detecting rail abnormal points 3, comprising:

[0052] S10, transducer A11 and transducer B12 at both ends of the rail detection group 1 periodically transmit and receive ultrasonic signals 16 to each other, and simultaneously collect and store real-time parameters on the rail, wherein the real-time parameters include real-time temperature, propagation distance between the transducer transmitting point and the corresponding receiving point, and simultaneously collect and store the first signal strength, the second signal strength, the third signal strength and the fourth signal strength; the first signal strength is the signal strength transmitted by transducer A to transducer B, the second signal strength is the signal strength received by transducer B to transducer A, the third signal strength is the signal strength transmitted by transducer B to transducer A, and the fourth signal strength is the signal strength received by transducer A to transducer B. The transducer includes a vibration source and a piezoelectric crystal, the vibration source is used to generate vibration to output the ultrasonic signal 16, and the piezoelectric crystal is used to receive the ultrasonic signal 16 after propagation.

[0053] S20, according to the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength and the third signal strength, obtain the fifth signal strength corresponding to the first signal strength and the sixth signal strength corresponding to the third signal strength; the fifth signal strength is the signal strength when transducer B receives transducer A when the rail is in a normal state, and the sixth signal strength is the signal strength when transducer A receives transducer B when the rail is in a normal state.

[0054] S30, by calculating the attenuation coefficient of the signal transmitted by transducer A11 to transducer B12 at the real-time temperature, the fifth signal strength, i.e., the signal strength of transducer B receiving transducer A when the rail is in a normal state, can be obtained according to the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, and the first signal strength. The attenuation coefficient of the signal transmitted by transducer B12 to transducer A11 at the real-time temperature can also be calculated, so according to the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, and the second signal strength, the sixth signal strength, i.e., the signal strength of transducer A receiving transducer B when the rail is in a normal state, can be obtained. Abnormal point 3 includes breakpoints, damage, cracks, and fissures. When the ultrasonic signal 16 encounters abnormal point 3, an abnormally large attenuation phenomenon will occur. The second signal strength and the fifth signal strength are compared, and the fourth signal strength and the sixth signal strength are compared. If the second signal strength is less than the fifth signal strength and the fourth signal strength is less than the sixth signal strength, it is determined that there is an abnormal point 3 on the rail. Transducer A11 and transducer B12 are used as the transmitter of initial signal strength and the receiver of attenuated signal strength, respectively, and can detect the abnormal point 3 more accurately, ensuring the accuracy of judging whether there is an abnormal point 3 on the rail. If only transducer A11 transmits the initial signal strength and transducer B12 receives the attenuated signal strength, when the abnormal point 3 is close to transducer B12, because the ultrasonic signal 16 has undergone severe signal strength attenuation on the rail, when the signal strength attenuation of the abnormal point 3 is encountered, the signal strength attenuation caused by the abnormal point 3 is not obvious. Therefore, transducer A11 and transducer B12 are used as the signal source for transmitting the initial signal strength, and transducer B12 and transducer A11 are used as the corresponding receiver of the attenuated signal strength. When compared with the attenuated signal strength when the rail is in a normal state, the accuracy of judging the abnormal point 3 on the rail can be improved.

[0055] On the basis of the above embodiment, the detection method further includes calculating the length and width of the abnormal point 3 when there is an abnormal point 3 on the rail, specifically including:

[0056] S40, collecting several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, distance between transducer A and abnormal point 3, distance between transducer B and abnormal point 3, first signal strength, second signal strength, third signal strength, fourth signal strength, length of abnormal point 3, and width of abnormal point 3;

[0057] S50, introduce the initial model to train the neural network model, take the historical temperature at the historical moment, the distance between the transducer A11 and the abnormal point 3, the distance between the transducer B12 and the abnormal point 3, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength data as input, take the length of the abnormal point 3 and the width of the abnormal point 3 corresponding to the historical moment as output, train the neural network model to obtain a model for calculating the length and width of the abnormal point 3 of the rail at real-time temperature, the model for calculating the length and width of the abnormal point 3 of the rail at real-time temperature is based on a formula for calculating the length and width of the abnormal point 3 of the rail at real-time temperature, and is used to calculate the length and width of the abnormal point 3 when the rail is in a normal state, and the formula for calculating the length and width of the abnormal point 3 at real-time temperature is:

[0058]

[0059] Among them, x i is the input vector, including the real-time temperature, the distance between transducer A and abnormal point 3, the distance between transducer B and abnormal point 3, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength, n is the number of neurons, w i represents the weight of the i-th neuron, b i represents the bias of the i-th neuron, Y represents the output vector including the outlier 3 length and the outlier 3 width, and i refers to the i-th neuron.

[0060] According to the real-time temperature, the distance between transducer A11 and abnormal point 3, the distance between transducer B12 and abnormal point 3, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength data are substituted into the formula for the length and width of abnormal point 3 at real-time temperature to obtain the length and width of abnormal point 3.

[0061] It should be noted that the topology of the neural network model refers to the configuration of each layer in the model, including the number of neurons in the input layer, hidden layer, and output layer. Minimum root mean square error refers to minimizing the difference between the predicted value and the actual value through iterative optimization algorithm during model training. The specific calculation process of the failure probability involves how the model converts input features into output failure probability.

[0062] Preferably, the following operation steps can be further refined: first, define input features, which may include historical temperatures at historical moments, the distance between transducer A11 and abnormal point 3, the distance between transducer B12 and abnormal point 3, the first signal strength, the second signal strength, the third signal strength, and the fourth signal strength data, and determine the weight and bias of each feature. Secondly, select a suitable activation function, such as sigmoid or ReLU function, to increase the nonlinear expression ability of the model. Then, determine the optimal model parameters and structure through methods such as cross-validation to improve the generalization ability of the model. In addition, regularization techniques, such as L1 or L2 regularization, can be used to prevent model overfitting.

[0063] More specifically, the following refinements or alternatives can be provided for the model training process: First, optimization algorithms such as batch gradient descent or stochastic gradient descent can be used to train the model to minimize the loss function. Second, a learning rate decay strategy can be set to ensure the stability and convergence of model training. Different neural network architectures, such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs), can also be explored to adapt to different types of data features and failure modes.

[0064] Based on the above embodiment, S20 further includes:

[0065] S21, collecting several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, propagation distance between a transducer transmitting point and a corresponding receiving point, a first signal strength, a third signal strength, a fifth signal strength, and a sixth signal strength;

[0066] S22, introducing an initial model exponential decay model, taking the historical temperature at the historical moment, the propagation distance between the transducer emission point and the corresponding receiving point, the first signal strength, and the third signal strength data as input, taking the fifth signal strength and the sixth signal strength and data corresponding to the historical moment as output, training the exponential decay model to obtain a real-time temperature normal signal strength model, the real-time temperature normal signal strength model is based on a real-time temperature normal signal strength formula, and is used to predict the corresponding normal signal strength received by the transducer when the rail is in a normal state, and the real-time temperature normal signal strength formula includes a real-time temperature fifth signal strength formula and a real-time temperature sixth signal strength formula;

[0067] The formula for the real-time temperature and the fifth signal strength is:

[0068] E A (d) = E A0 e -α(T)L

[0069] Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point in meters (m), e is a constant, and E A0Indicates the first signal strength, E A (d) represents the fifth signal strength, α(T) represents the attenuation coefficient of the signal transmitted from transducer A to transducer B at temperature T;

[0070] Substituting the historical temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength and the fifth signal strength into the above formula, the attenuation coefficient of the signal transmitted by transducer A11 to transducer B12 at the real-time temperature is calculated.

[0071] The formula for the real-time temperature sixth signal strength is:

[0072] E B (d) = E B0 e -α′(T)L

[0073] Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point in meters (m), e is a constant, and E B0 Indicates the third signal strength, E B (d) represents the sixth signal strength, and α′(T) represents the attenuation coefficient of the signal transmitted by transducer B to transducer A at temperature T.

[0074] S23, the historical temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the second signal strength and the first signal strength are substituted into the above formula to calculate the attenuation coefficient of the signal transmitted by transducer B12 to transducer A11 at the real-time temperature.

[0075] Under different temperature conditions, the attenuation characteristics of the ultrasonic signal 16 will show significant differences, that is, the ultrasonic signal attenuation coefficient will change accordingly. In order to accurately assess the health of the rail, this physical phenomenon must be taken into account, and the fifth signal strength and the sixth signal strength corresponding to the temperature must be dynamically calculated to ensure that no matter how the rail temperature fluctuates, the rail detection standard can maintain a high degree of adaptability and accuracy, thereby more accurately identifying the abnormal state of the rail. The attenuation coefficient of the ultrasonic signal 16 is used to comprehensively consider the historical temperature at the recorded historical moment, the propagation distance between the transducer emission point and the corresponding receiving point, and the first signal strength, the second signal strength, the fifth signal strength and the sixth signal strength. The attenuation coefficient of the signal strength is obtained through precise calculation. This process aims to accurately quantify the degree of signal strength reduction caused by various factors during the propagation process, such as inherent structures in the rail such as screw holes, wire holes, welds, switches, joints, etc. Subsequently, the training module 22 will further retrieve or calculate the fifth signal strength and the sixth signal strength that match the current conditions based on the calculated attenuation coefficient, and substitute the real-time temperature, the propagation distance between the transducer emission point and the corresponding receiving point, the first signal strength data, and the attenuation coefficient of the signal transmitted from transducer A11 to transducer B12 at temperature T into the real-time temperature fifth signal strength formula to obtain the fifth signal strength. At the same time, substitute the real-time temperature, the propagation distance between the transducer emission point and the corresponding receiving point, the third signal strength, and the attenuation coefficient data of the signal transmitted from transducer B12 to transducer A11 at temperature T into the real-time temperature sixth signal strength formula to obtain the sixth signal strength. Provide real, effective, objective, and large amounts of data support for the rail detection system, so that subsequent judgment, analysis, and calculation processes can be carried out based on this accurate training information, and improve the accuracy of calculating and obtaining the information of the abnormal point 3 of the rail detection.

[0076] On the basis of the above embodiment, the detection method further includes determining the position of the abnormal point 3 when there is an abnormal point 3 on the rail, specifically including:

[0077] S60, collecting and storing real-time parameters on the rail, wherein the real-time parameters include real-time temperature, propagation distance between a transducer transmitting point and a corresponding receiving point, first signal strength, second signal strength, third signal strength, and fourth signal strength, and obtaining an attenuation coefficient of a signal transmitted from the real-time temperature transducer A11 to the transducer B12, and an attenuation coefficient of a signal transmitted from the real-time temperature transducer B12 to the transducer A11;

[0078] S70, introduce the mathematical model constructed by the physical mechanism of the initial model, take the real-time temperature of the real-time parameters, the distance between the transducer A11 and the abnormal point 3, the distance between the transducer B12 and the abnormal point 3, the first signal strength, the third signal strength, the attenuation coefficient of the signal transmitted from the transducer A11 to the transducer B12 at temperature T, the attenuation coefficient of the signal transmitted from the transducer B12 to the transducer A11 at temperature T, and the attenuation coefficient data of the signal passing through the abnormal point 3 as input, take the propagation distance between the transducer transmitting point and the corresponding receiving point, the second signal strength, and the fourth signal strength data as output, train the mathematical model constructed by the physical mechanism to obtain a model for determining the position of the abnormal point 3 of the rail by real-time temperature, the model for determining the position of the abnormal point 3 of the rail by real-time temperature is based on the formula for determining the position of the abnormal point 3 of the rail by real-time temperature, and is used to determine the position of the abnormal point 3 of the rail, and the formula for determining the position of the abnormal point 3 of the rail by real-time temperature is:

[0079]

[0080]

[0081] L=X1+X2

[0082] Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point, in meters (m), X1 represents the distance between transducer A and abnormal point 3, in meters (m), X2 represents the distance between transducer B and abnormal point 3, in meters (m), e is a constant, β represents the attenuation coefficient of the signal passing through abnormal point 3, E A (c) represents the second signal strength, E A0 represents the first signal strength, α(T) represents the attenuation coefficient of the signal transmitted from transducer A to transducer B at temperature T, and E B (c) represents the fourth signal strength, E B0 represents the third signal strength, α′(T) represents the attenuation coefficient of the signal transmitted by transducer B to transducer A at temperature T;

[0083] S80, substitute the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength, the second signal strength, the third signal strength, the fourth signal strength, the attenuation coefficient of the signal transmitted from transducer A11 to transducer B12 at temperature T, and the attenuation coefficient of the signal transmitted from transducer B12 to transducer A11 at temperature T into the formula for determining the position of the rail abnormal point 3 using the real-time temperature, and obtain the distance between transducer A11 and the abnormal point 3, the distance between transducer B12 and the abnormal point 3, and the attenuation coefficient of the signal passing through the abnormal point 3.

[0084] S90: Generate rail warning risk level indication information according to the length and width of abnormal point 3 under the real-time temperature. The indication information may be presented in the form of sound, light signal, text or number, so as to remind maintenance personnel to carry out maintenance or repair.

[0085] In order to accurately assess the rail status and respond to potential risks in a timely manner, the following steps are implemented: the rail abnormal point 3 detection system collects several groups of rail abnormal point 3 length and width data at real-time temperature at historical moments. Subsequently, a two-dimensional risk matrix model is constructed using these data, in which the horizontal axis of the matrix represents the length of abnormal point 3 and the vertical axis represents the width of abnormal point 3. Through this matrix, we subdivide the risks that the rail may face into multiple areas, and each area clearly identifies different risk levels, such as: "low risk", "medium risk" and "high risk" risk level intervals, to intuitively reflect the severity of the impact of abnormal point 3 on the rail. At the same time, several risk level intervals are matched with different combinations of the length and width of abnormal point 3. When the length and width data of abnormal point 3 calculated at real-time temperature fall into a certain interval, the alarm module 26 can automatically identify the risk level interval in which the abnormal point 3 falls, and generate corresponding alarms and maintenance instructions. For example, "high risk" requires a red alarm light and urgent maintenance immediately, "medium risk" requires an orange alarm light and maintenance after the operation is completed on the same day, and "low risk" requires a yellow alarm light and maintenance after the operation is completed on the same day. The rail alarm maintenance process not only improves the accuracy of risk identification by maintenance personnel, but also ensures the timeliness and pertinence of inspection and maintenance work, effectively reducing potential safety risks caused by rail abnormalities.

[0086] In this embodiment, Figure 2 As shown, the present invention provides a rail abnormal point 3 detection system, including a collection module 21, an analysis module 23, and a judgment module 24. The transducers A11 and B12 at both ends of the detection group periodically transmit and receive ultrasonic signals to each other. The collection module 21 is used to collect and store real-time parameters on the rail, wherein the real-time parameters include real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, and the collection module 21 simultaneously collects and stores the first signal strength and the third signal strength, and obtains the fifth signal strength corresponding to the first signal strength and the sixth signal strength corresponding to the third signal strength. The transducer includes a vibration source and a piezoelectric crystal, the vibration source is used to generate vibration to output an ultrasonic signal 16, and the piezoelectric crystal is used to receive the ultrasonic signal 16 after propagation. The analysis module 23 is used to obtain the fifth signal strength corresponding to the first signal strength and the sixth signal strength corresponding to the third signal strength according to the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength and the third signal strength. The judgment module 24 is used to compare the second signal strength and the fifth signal strength, and to compare the fourth signal strength and the sixth signal strength. If the second signal strength is less than the fifth signal strength and the fourth signal strength is less than the sixth signal strength, it is judged that there is an abnormal point 3 on the rail.

[0087] The analysis module 23 is used to obtain a fifth signal strength corresponding to the first signal strength and a sixth signal strength corresponding to the third signal strength according to the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength and the third signal strength.

[0088] By calculating the attenuation coefficient of the signal transmitted by transducer A11 to transducer B12 at the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point and the first signal strength, the fifth signal strength, i.e., the signal strength of transducer B receiving transducer A when the rail is in a normal state, can be obtained. By calculating the attenuation coefficient of the signal transmitted by transducer B12 to transducer A11 at the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point and the second signal strength, the sixth signal strength, i.e., the signal strength of transducer A receiving transducer B when the rail is in a normal state, can be obtained. The abnormal point 3 includes a breakpoint, damage, crack and fissure, etc. When the ultrasonic signal 16 encounters the abnormal point 3, an abnormally large attenuation phenomenon will occur. The judgment module 24 is used to compare the second signal strength and the fifth signal strength, and at the same time compare the fourth signal strength and the sixth signal strength. If the second signal strength is less than the fifth signal strength and the fourth signal strength is less than the sixth signal strength, it is judged that there is an abnormal point 3 on the rail.

[0089] Based on the above embodiment, the detection system further includes a training module 22:

[0090] The collection module 21 is used to collect several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, propagation distance between the transducer transmitting point and the corresponding receiving point, first signal strength, third signal strength, fifth signal strength and sixth signal strength;

[0091] Training module 22 introduces an initial model exponential decay model, takes the historical temperature at the historical moment, the propagation distance between the transducer emission point and the corresponding receiving point, the first signal strength, and the third signal strength data as input, and takes the fifth signal strength and sixth signal strength and data corresponding to the historical moment as output, and trains the exponential decay model to obtain a real-time temperature normal signal strength model. The real-time temperature normal signal strength model is based on the real-time temperature normal signal strength formula, and is used to predict the corresponding normal signal strength received by the transducer when the rail is in a normal state. The real-time temperature normal signal strength formula includes the real-time temperature fifth signal strength formula and the real-time temperature sixth signal strength formula.

[0092] The formula for the real-time temperature and the fifth signal strength is:

[0093] E A (d) = E A0 e -α(T)L

[0094] Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point in meters (m), e is a constant, and E A0 Indicates the first signal strength, E A (d) represents the fifth signal strength, α(T) represents the attenuation coefficient of the signal transmitted from transducer A to transducer B at temperature T;

[0095] Substituting the historical temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength and the fifth signal strength into the above formula, the attenuation coefficient of the signal transmitted by transducer A11 to transducer B12 at the real-time temperature is calculated.

[0096] The formula for the real-time temperature sixth signal strength is:

[0097] E B (d) = E B0 e -α′(T)L

[0098] Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point in meters (m), e is a constant, and E B0 Indicates the third signal strength, E B (d) represents the sixth signal strength, and α′(T) represents the attenuation coefficient of the signal transmitted by transducer B to transducer A at temperature T.

[0099] Substituting the historical temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the second signal strength and the first signal strength into the above formula, the attenuation coefficient of the signal transmitted by transducer B12 to transducer A11 at the real-time temperature is calculated.

[0100] Substitute the real-time temperature, the propagation distance between the transducer emission point and the corresponding receiving point, the first signal strength data, and the attenuation coefficient of the signal transmitted from transducer A11 to transducer B12 at temperature T into the real-time temperature fifth signal strength formula to obtain the fifth signal strength. At the same time, substitute the real-time temperature, the propagation distance between the transducer emission point and the corresponding receiving point, the third signal strength, and the attenuation coefficient data of the signal transmitted from transducer B12 to transducer A11 at temperature T into the real-time temperature sixth signal strength formula to obtain the sixth signal strength. Provide real, effective, objective, and large amounts of data support for the rail detection system, so that subsequent judgment, analysis, and calculation processes can be based on this accurate training information, and improve the accuracy of calculating and obtaining the information of the rail detection abnormal point 3.

[0101] Based on the above embodiment, the detection system further includes a calculation module 25:

[0102] When there is an abnormal point 3 on the rail, the acquisition module 21 and the calculation module 25 are also used to calculate the position of the abnormal point 3: the acquisition module 21 is used to collect and store real-time parameters on the rail, wherein the real-time parameters include real-time temperature, propagation distance between the transducer transmitting point and the corresponding receiving point, first signal strength, second signal strength, third signal strength, fourth signal strength, and (analysis module) obtain the attenuation coefficient of the real-time temperature transducer A transmitting the signal to the transducer B, and the attenuation coefficient of the real-time temperature transducer B transmitting the signal to the transducer A;

[0103] The calculation module 25 is used to introduce the mathematical model constructed by the physical mechanism of the initial model, and takes the real-time temperature of the real-time parameters, the distance between the transducer A and the abnormal point 3, the distance between the transducer B and the abnormal point 3, the first signal strength, the third signal strength, the attenuation coefficient of the signal transmitted by the transducer A to the transducer B at the temperature T, the attenuation coefficient of the signal transmitted by the transducer B to the transducer A at the temperature T, and the attenuation coefficient data of the signal passing through the abnormal point 3 as input, and takes the propagation distance between the transducer transmitting point and the corresponding receiving point, the second signal strength, and the fourth signal strength data as output, and trains the mathematical model constructed by the physical mechanism to obtain a model for determining the position of the abnormal point 3 of the rail by real-time temperature. The model for determining the position of the abnormal point 3 of the rail by real-time temperature is based on the formula for determining the position of the abnormal point 3 of the rail by real-time temperature, and is used to determine the position of the abnormal point 3 of the rail. The formula for determining the position of the abnormal point 3 of the rail by real-time temperature is:

[0104]

[0105]

[0106] L=X1+X2

[0107] Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point, in meters (m), X1 represents the distance between transducer A and abnormal point 3, in meters (m), X2 represents the distance between transducer B and abnormal point 3, in meters (m), e is a constant, β represents the attenuation coefficient of the signal passing through abnormal point 3, E A (c) represents the second signal strength, E A0 represents the first signal strength, α(T) represents the attenuation coefficient of the signal transmitted from transducer A to transducer B at temperature T, and E B (c) represents the fourth signal strength, E B0 represents the third signal strength, and α′(T) represents the attenuation coefficient of the signal transmitted by transducer B to transducer A at temperature T.

[0108] The real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength, the second signal strength, the third signal strength, the fourth signal strength, the attenuation coefficient of the signal transmitted from transducer A11 to transducer B12 at temperature T, and the attenuation coefficient of the signal transmitted from transducer B12 to transducer A11 at temperature T are substituted into the formula for determining the position of the rail abnormal point 3 by the real-time temperature to obtain the distance between transducer A11 and the abnormal point 3, the distance between transducer B12 and the abnormal point 3, and the attenuation coefficient of the signal passing through the abnormal point 3.

[0109] On the basis of the above embodiment, when there is an abnormal point 3 on the rail, the acquisition module 21, the training module 22 and the calculation module 25 are also used to calculate the length and width of the abnormal point 3:

[0110] The acquisition module 21 collects several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, the distance between transducer A and abnormal point 3, the distance between transducer B and abnormal point 3, the first signal strength, the second signal strength, the third signal strength, the fourth signal strength, the length of abnormal point 3 and the width of abnormal point 3.

[0111] The training module 22 introduces the initial model to train the neural network model, takes the historical temperature at the historical moment, the distance between the transducer A11 and the abnormal point 3, the distance between the transducer B12 and the abnormal point 3, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength data as input, takes the length of the abnormal point 3 and the width of the abnormal point 3 corresponding to the historical moment as output, trains the neural network model to obtain a model for calculating the length and width of the abnormal point 3 of the rail at real-time temperature, the model for calculating the length and width of the abnormal point 3 of the rail at real-time temperature is based on the formula for calculating the length and width of the abnormal point 3 of the rail at real-time temperature, and is used to calculate the length and width of the abnormal point 3 when the rail is in a normal state, and the formula for calculating the length and width of the abnormal point 3 at real-time temperature is:

[0112]

[0113] Among them, x i is the input vector, including the real-time temperature, the distance between transducer A and abnormal point 3, the distance between transducer B and abnormal point 3, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength, n is the number of neurons, w i represents the weight of the i-th neuron, b i represents the bias of the i-th neuron, Y represents the output vector including the outlier 3 length and the outlier 3 width, and i refers to the i-th neuron.

[0114] The calculation module 25 is used to substitute the real-time temperature, the distance between the transducer A11 and the abnormal point 3, the distance between the transducer B12 and the abnormal point 3, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength data into the formula for the length and width of the abnormal point 3 at the real-time temperature to obtain the length and width of the abnormal point 3.

[0115] Based on the above embodiment, the detection system further includes an alarm module 26.

[0116] The alarm module 26 generates rail early warning risk level indication information according to the length and width of the abnormal point 3 under the real-time temperature. The indication information may be presented in the form of sound, light signal, text or number, so as to remind the maintenance personnel to carry out maintenance or repair.

[0117] In this embodiment, Figure 5 As shown, the present invention also provides a rail abnormal point 3 detection device, including a plurality of detection groups 1 and a rail abnormal point 3 detection system terminal 2 arranged on each side of the rail. Figure 3 As shown, a detection group 1 is arranged at a set distance on one side of the rail, and a pair of transducers are set at both ends of a detection group 1, so that the entire rail is divided into several small sections for detection. The small section detection method can reduce data and calculation errors and improve the accuracy of the detection results of the detection device. For example: in order to avoid thermal expansion and contraction, the rails are designed in sections, and a pair of ultrasonic transducers of the detection group 1 can be set at the starting point and end point of each section of the rail.

[0118] like Figure 4-5 As shown, the detection group 1 includes a transducer A11, a transducer B12, an analog-to-digital converter 14, a wireless communicator 15 and a temperature sensor 13 arranged at both ends. The transducer A11 and the transducer B12 are used to periodically transmit ultrasonic signals 16 to each other and receive corresponding ultrasonic signals 16. The analog-to-digital converter 14 is connected to the transducer A11 and the transducer B12 respectively, and is used to convert the various received signal strengths into digital signals and send them to the wireless communicator 15. It is used to convert the received initial signal strength emitted by the transducer B12 and the corresponding attenuated signal strength after propagation into digital signals and send them to the wireless communicator 15. The temperature sensor 13 is used to collect the real-time temperature of the rail and convert it into a digital signal and send it to the wireless communicator 15. The wireless communicator 15 is connected to the temperature sensor 13 and the digital-to-analog converter at one end and to the terminal 2 at the other end, and is used to receive digital information and output digital information to the terminal 2. The inspection group and the terminal 2 perform remote wireless communication.

[0119] Terminal 2 is connected to wireless communicator 15, and is used to perform comprehensive calculation on the received real-time temperature, digital signal, initial signal strength transmitted by transducer A11 and transducer B12, corresponding attenuated signal strength digital signal after propagation, and propagation distance between the transducer transmitting point and the corresponding receiving point, and obtain the attenuated signal strength after propagation when the real-time temperature rail is in normal state, and compare it with the attenuated signal strength after actual propagation, and judge whether there is an abnormal point 3 on the rail, and then calculate the position, width and length of the abnormal point 3 on the rail at the real-time temperature. Terminal 2 is equipped with a processing chip for data analysis and calculation processing, and terminal 2 includes acquisition module 21, training module 22, analysis module 23, judgment module 24, calculation module 25 and alarm module 26.

[0120] It should be noted that the acquisition module 21 may include a non-volatile memory, which stores several groups of real-time temperatures collected at historical moments, the initial signal strength emitted by transducer A11 and transducer B12 to each other and the attenuated signal strength after propagation, the propagation distance, the distance between transducer A11 and the abnormal point 3, and the distance between transducer B12 and the abnormal point 3, etc.

[0121] The training module 22 is a model training program pre-configured on the processing chip. It first introduces the initial model, then calls several groups of historical moment detection group 1 data stored in the non-volatile memory as input, and uses the rail abnormal point 3 data at the historical moment as output, and retrains the initial model to obtain a preset attenuation signal strength threshold model at real-time temperature, a rail abnormal point 3 position model at real-time temperature, and a model for identifying the rail abnormal point 3 position at real-time temperature.

[0122] The alarm module 26 can automatically identify which risk level interval the length and width data of the abnormal point 3 fall into, and generate corresponding alarms and maintenance instructions, which enhances the accuracy of risk identification and judgment by maintenance personnel. When multiple abnormal points 3 are detected, the detection system can also prioritize the abnormal points 3 according to the risk level interval of the abnormal points 3, guide maintenance personnel to efficiently arrange maintenance plans, ensure the timeliness and pertinence of maintenance work, and greatly reduce the safety risks caused by rail abnormalities. The alarm module 26 can be provided with a display as needed to facilitate relevant personnel to read the rail warning risk level indication information through the display.

[0123] On the basis of the above-mentioned embodiment, a rail abnormal point 3 detection device also includes an inspector, which is respectively connected to transducer A11 and transducer B12. Before the detection process is started, the inspector verifies the efficiency of the transducer in emitting ultrasonic waves and accurately receives the ultrasonic signal 16 returned after propagation. The transducer needs to be inspected by the inspector before use. This is mainly to ensure the performance and safety of the transducer and to avoid failure or damage during use, so as to ensure that the transducer is in the best working state, thereby ensuring the accuracy and reliability of subsequent data collection.

[0124] The attenuation coefficient mentioned above is used to balance the weight of each data in the formula calculation, so as to promote the accuracy of the calculation result. The size of the coefficient is to quantify each parameter to obtain a specific value, which is convenient for subsequent comparison. The size of the coefficient depends on the amount of sample data and the technical personnel in this field preliminarily set the corresponding weight factor coefficient for each group of sample data, as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0125] The above descriptions are only some preferred embodiments of the present invention and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the above features are replaced with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention to form a technical solution.

Claims

1. A method for detecting abnormal rail points, characterized in that The detection group includes transducers A and B located at opposite ends of the rail, and the detection method includes: The transducers A and B at both ends of the detection group periodically transmit and receive ultrasonic signals to each other, and collect and store real-time parameters on the rail at the same time, wherein the real-time parameters include real-time temperature, propagation distance between the transducer transmitting point and the corresponding receiving point, first signal strength, second signal strength, third signal strength and fourth signal strength; the first signal strength is the signal strength transmitted by transducer A to transducer B, the second signal strength is the signal strength received by transducer B to transducer A, the third signal strength is the signal strength transmitted by transducer B to transducer A, and the fourth signal strength is the signal strength received by transducer A to transducer B; According to the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength and the third signal strength, a fifth signal strength corresponding to the first signal strength and a sixth signal strength corresponding to the third signal strength are obtained; the fifth signal strength is the signal strength received by transducer B from transducer A when the rail is in a normal state, and the sixth signal strength is the signal strength received by transducer A from transducer B when the rail is in a normal state; comparing the second signal strength and the fifth signal strength, and comparing the fourth signal strength and the sixth signal strength at the same time, and if the second signal strength is less than the fifth signal strength and the fourth signal strength is less than the sixth signal strength, determining that an abnormal point exists on the rail; The detection method also includes calculating the length and width of the abnormal point when an abnormal point state exists on the rail, specifically including: collecting several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, the distance between transducer A and the abnormal point, the distance between transducer B and the abnormal point, the first signal strength, the second signal strength, the third signal strength, the fourth signal strength, the length of the abnormal point and the width of the abnormal point; The initial model is introduced to train the neural network model, and the historical temperature at the historical moment, the distance between the transducer A and the abnormal point, the distance between the transducer B and the abnormal point, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength data are used as inputs, and the length of the abnormal point and the width of the abnormal point corresponding to the historical moment are used as outputs. The neural network model is trained to obtain a model for calculating the length and width of the abnormal point of the rail at real-time temperature. The model for calculating the length and width of the abnormal point of the rail at real-time temperature is based on a formula for calculating the length and width of the abnormal point of the rail at real-time temperature, and is used to calculate the length and width of the abnormal point when the rail is in a normal state. The formula for calculating the length and width of the abnormal point at real-time temperature is: Among them, x i is the input vector, including the real-time temperature, the distance between transducer A and the abnormal point, the distance between transducer B and the abnormal point, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength, n is the number of neurons, w i represents the weight of the i-th neuron, b i represents the bias of the i-th neuron, Y represents the output vector including the outlier length and outlier width, and i refers to the i-th neuron; According to the real-time temperature abnormal point length and abnormal point width, rail early warning risk level indication information is generated.

2. The detection method according to claim 1, characterized in that: According to the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength and the third signal strength, a fifth signal strength corresponding to the first signal strength and a sixth signal strength corresponding to the third signal strength are obtained, further comprising: Collecting several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, propagation distance between a transducer transmitting point and a corresponding receiving point, a first signal strength, a third signal strength, a fifth signal strength, and a sixth signal strength; An initial model exponential decay model is introduced, and the historical temperature at the historical moment, the propagation distance between the transducer emission point and the corresponding receiving point, the first signal strength, and the third signal strength data are used as inputs, and the fifth signal strength and sixth signal strength data corresponding to the historical moment are used as outputs. The exponential decay model is trained to obtain a real-time temperature normal signal strength model. The real-time temperature normal signal strength model is based on a real-time temperature normal signal strength formula, and is used to predict the corresponding normal signal strength received by the transducer when the rail is in a normal state. The real-time temperature normal signal strength formula includes a real-time temperature fifth signal strength formula and a real-time temperature sixth signal strength formula; The fifth signal strength formula of real-time temperature is: E A (d) = E A0 e -α(T)L Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point in meters (m), e is a constant, and E A0 Indicates the first signal strength, E A (d) represents the fifth signal strength, α(T) represents the transducer at temperature T A Directional transducer B The attenuation coefficient of the transmitted signal; The real-time temperature sixth signal strength formula is: E B (d) = E B0 e -α′(T)L Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point in meters (m), e is a constant, and E B0 Indicates the third signal strength, E B (d) represents the sixth signal strength, and α′(T) represents the attenuation coefficient of the signal transmitted by transducer B to transducer A at temperature T.

3. The detection method according to claim 2, characterized in that: The detection method further includes determining the position of the abnormal point when an abnormal point exists on the rail, specifically comprising: Collect and store real-time parameters on the rail, wherein the real-time parameters include real-time temperature, propagation distance between a transducer transmitting point and a corresponding receiving point, first signal strength, second signal strength, third signal strength, and fourth signal strength, and obtain an attenuation coefficient of a signal transmitted from real-time temperature transducer A to transducer B and an attenuation coefficient of a signal transmitted from real-time temperature transducer B to transducer A; The mathematical model constructed by the physical mechanism of the initial model is introduced, and the real-time temperature of the real-time parameters, the distance between transducer A and the abnormal point, the distance between transducer B and the abnormal point, the first signal strength, the third signal strength, the attenuation coefficient of the signal transmitted from transducer A to transducer B at temperature T, the attenuation coefficient of the signal transmitted from transducer B to transducer A at temperature T, and the attenuation coefficient data of the signal passing through the abnormal point are taken as input, and the propagation distance between the transducer transmitting point and the corresponding receiving point, the second signal strength, and the fourth signal strength data are taken as output. The mathematical model constructed by the physical mechanism is trained to obtain a model for determining the position of the abnormal point of the rail by real-time temperature. The model for determining the position of the abnormal point of the rail by real-time temperature is based on the formula for determining the position of the abnormal point of the rail by real-time temperature, and is used to determine the position of the abnormal point of the rail. The formula for determining the position of the abnormal point of the rail by real-time temperature is: L=X1+X2 Where L represents the propagation distance between the transducer transmitting point and the corresponding receiving point, in meters (m), X1 represents the distance between transducer A and the abnormal point, in meters (m), X2 represents the distance between transducer B and the abnormal point, in meters (m), e is a constant, β represents the attenuation coefficient of the signal passing through the abnormal point, E A (c) represents the second signal strength, E A0 represents the first signal strength, α(T) represents the attenuation coefficient of the signal transmitted from transducer A to transducer B at temperature T, and E B (c) represents the fourth signal strength, E B0 represents the third signal strength, and α'(T) represents the attenuation coefficient of the signal transmitted by transducer B to transducer A at temperature T.

4. A rail abnormal point detection system based on the rail abnormal point detection method according to any one of claims 1 to 3, characterized in that: Including acquisition module, analysis module and judgment module; The transducers A and B at both ends of the detection group periodically transmit and receive ultrasonic signals to each other, and the acquisition module is used to collect and store real-time parameters on the rail, wherein the real-time parameters include real-time temperature, propagation distance between the transducer emission point and the corresponding receiving point, first signal strength, second signal strength, third signal strength and fourth signal strength; The analysis module is used to obtain a fifth signal strength corresponding to the first signal strength and a sixth signal strength corresponding to the third signal strength according to the real-time temperature, the propagation distance between the transducer transmitting point and the corresponding receiving point, the first signal strength and the third signal strength; The judgment module is used to compare the second signal strength and the fifth signal strength, and to compare the fourth signal strength and the sixth signal strength. If the second signal strength is less than the fifth signal strength and the fourth signal strength is less than the sixth signal strength, it is judged that there is an abnormal point on the rail.

5. The detection system according to claim 4, characterized in that The detection system also includes a training module: The acquisition module is used to collect several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, propagation distance between a transducer transmitting point and a corresponding receiving point, a first signal strength, a third signal strength, a fifth signal strength, and a sixth signal strength; The training module introduces an initial model exponential decay model, takes the historical temperature at the historical moment, the propagation distance between the transducer emission point and the corresponding receiving point, the first signal strength, and the third signal strength data as input, and takes the fifth signal strength and the sixth signal strength and data corresponding to the historical moment as output, and trains the exponential decay model to obtain a real-time temperature normal signal strength model. The real-time temperature normal signal strength model is based on the real-time temperature normal signal strength formula, and is used to predict the corresponding normal signal strength received by the transducer when the rail is in a normal state. The real-time temperature normal signal strength formula includes the real-time temperature fifth signal strength formula and the real-time temperature sixth signal strength formula.

6. The detection system according to claim 5, characterized in that: The detection system also includes a computing module: When there is an abnormal point on the rail, the acquisition module and the calculation module are also used to calculate the position of the abnormal point: The acquisition module is used to acquire and store real-time parameters on the rail, wherein the real-time parameters include real-time temperature, propagation distance between a transducer transmitting point and a corresponding receiving point, first signal strength, second signal strength, third signal strength, and fourth signal strength, and obtain an attenuation coefficient of a signal transmitted from real-time temperature transducer A to transducer B, and an attenuation coefficient of a signal transmitted from real-time temperature transducer B to transducer A; The calculation module is used to introduce a mathematical model constructed by the physical mechanism of the initial model, and takes the real-time temperature of the real-time parameters, the distance between the transducer A and the abnormal point, the distance between the transducer B and the abnormal point, the first signal strength, the third signal strength, the attenuation coefficient of the signal transmitted by the transducer A to the transducer B at the temperature T, the attenuation coefficient of the signal transmitted by the transducer B to the transducer A at the temperature T, and the attenuation coefficient data of the signal passing through the abnormal point as input, and takes the propagation distance between the transducer transmitting point and the corresponding receiving point, the second signal strength, and the fourth signal strength data as output, and trains the mathematical model constructed by the physical mechanism to obtain a model for determining the position of the abnormal point of the rail by real-time temperature. The model for determining the position of the abnormal point of the rail by real-time temperature is based on a formula for determining the position of the abnormal point of the rail by real-time temperature, and is used to determine the position of the abnormal point of the rail.

7. The detection system according to claim 6, characterized in that: When there is an abnormal point on the rail, the acquisition module, training module and calculation module are also used to calculate the length and width of the abnormal point: The acquisition module is used to collect several groups of historical parameters at historical moments, wherein the historical parameters include historical temperature, distance between transducer A and the abnormal point, distance between transducer B and the abnormal point, first signal strength, second signal strength, third signal strength, fourth signal strength, abnormal point length and abnormal point width; The training module is used to introduce the initial model to train the neural network model, take the historical temperature at the historical moment, the distance between the transducer A and the abnormal point, the distance between the transducer B and the abnormal point, the first signal strength, the second signal strength, the third signal strength and the fourth signal strength data as input, take the length of the abnormal point and the width of the abnormal point corresponding to the historical moment as output, train the neural network model to obtain a model for calculating the length and width of the abnormal point of the rail at real-time temperature, the model for calculating the length and width of the abnormal point of the rail at real-time temperature is based on a formula for calculating the length and width of the abnormal point of the rail at real-time temperature, and is used to calculate the length and width of the abnormal point when the rail is in a normal state; The calculation module is used to calculate the abnormal point length and width model according to the real-time temperature to obtain the abnormal point length and abnormal point width.

8. The detection system according to claim 7, characterized in that: The detection system also includes an alarm module, which is used to generate rail early warning risk level indication information based on the real-time temperature abnormal point length and abnormal point width.

9. A rail abnormal point detection device, characterized in that: The invention comprises a plurality of detection groups arranged on each side of the rail and a terminal of the rail abnormal point detection system according to claim 8, The detection group includes a transducer A, a transducer B, an analog-to-digital converter, a wireless communication device and a temperature sensor arranged at both ends; the transducer A and the transducer B are used to periodically transmit ultrasonic signals to each other and receive corresponding ultrasonic signals; The analog-to-digital converter is connected to the transducer A and the transducer B respectively, and is used to convert various received signal strengths into digital signals and send them to the wireless communicator; The temperature sensor is used to collect the real-time temperature of the rail and convert it into a digital signal, which is then sent to the wireless communicator; The wireless communicator has one end connected to the temperature sensor and the digital-to-analog converter respectively, and the other end connected to the terminal, for receiving digital information and outputting the digital information to the terminal; The terminal is connected to the wireless communicator and is used to perform comprehensive calculation on the received real-time temperature, digital signal, initial signal strength transmitted by transducer A and transducer B and corresponding attenuated signal strength digital signal after propagation, and propagation distance between the transducer transmitting point and the corresponding receiving point; obtain the signal strength after propagation when the real-time temperature rail is in a normal state, compare it with the attenuated signal strength after actual propagation, determine whether there is an abnormal point on the rail, and then calculate the position, width and length of the abnormal point on the rail at the real-time temperature; The terminal comprises a collection module, a training module, an analysis module, a judgment module, a calculation module and an alarm module.

Images