Railway switch machine indication gap real-time monitoring system and monitoring method thereof

Abstract

The invention provides a railway point switch indication gap real-time monitoring system and a monitoring method thereof, and relates to the technical field of railway signals. The railway switch machine indication gap real-time monitoring system comprises a camera; the lower computer is used for acquiring the scale images, acquired by the image acquisition module, of the outer side of the switch machine on the indication rod, and the scale images of the outer side of the switch machine on the indication rod comprise standard position scale reading images and real-time scale images of the outer side of the switch machine on the indication rod; taking a scale image on an indication rod on the outer side of the switch machine when the reference position is positioned and the reference position is reversed as a standard position scale reading image, obtaining a standard position scale reading, obtaining a scale real-time reading, and calculating to obtain an actual gap offset; and the upper computer is used for receiving the actual gap offset and comparing the actual gap offset with a set threshold value. The system monitors the indication gap of the railway switch machine in real time so as to ensure normal operation of a railway signal system and safe passing of a train.

Description

Technical Field

[0001] The invention relates to the technical field of railway signals, in particular to a railway switch machine indication gap real-time monitoring system and a monitoring method thereof. Background Art

[0002] The background technology of railway switch gap monitoring mainly involves the safety and reliability of railway signal equipment. The switch is an important part of the railway signal system, which is used to control the switching and locking of turnouts to ensure that trains run along the predetermined route. The gap is a key component in the switch, and its state directly affects the normal operation of the switch and the safe operation of the train.

[0003] The traditional switch gap monitoring system mainly relies on manual inspection and regular maintenance, which has problems such as low efficiency, poor real-time performance, and susceptibility to human factors. With the rapid development of railway transportation and the increase in train operation density, the safety and reliability requirements of railway signal equipment are becoming higher and higher. Therefore, it is urgent to develop a system that can monitor the switch gap status in real time. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the shortcomings of the prior art, the present invention provides a real-time monitoring system and a monitoring method for the railway switch marking gap. An industrial-grade camera is used to capture images of the outer marking rod of the switch, and then a deep learning model is used to recognize the scale image of the outer marking rod of the switch, so as to solve the problem of accurate and real-time monitoring of the close contact status of the switch.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0008] In a first aspect, a railway switch machine indication gap real-time monitoring system is provided, comprising:

[0009] A camera is fixed on the sleeve of the outer side indicator rod of the switch machine and is used to collect the scale image on the outer side indicator rod of the switch machine in real time;

[0010] The lower computer obtains the scale image on the outer side of the switch machine represented by the image acquisition module, wherein the scale image on the outer side of the switch machine represented by the image acquisition module includes a standard position scale reading image and a real-time scale image on the outer side of the switch machine represented by the image acquisition module, takes the scale image on the outer side of the switch machine represented by the image at the positioning reference position and the reverse reference position as the standard position scale reading image, identifies and processes the standard position scale reading image to obtain the standard position scale reading, identifies and processes the real-time scale image on the outer side of the switch machine represented by the image acquisition module to obtain the real-time scale reading, compares the real-time scale reading with the standard position scale reading to obtain the image offset scale, and calculates the actual offset of the gap through the image offset scale.

[0011] The process of identifying and processing the scale image on the outer side of the switch machine includes:

[0012] The scale image on the outer side of the switch machine is gray-scaled and the collected RGB image is converted into a gray-scale image. The conversion formula is as follows:

[0013] Gray=A1*R+A2*G+A3*B

[0014] Among them, A1, A2, and A3 are the calculation coefficients of the three-color values ​​of the RGB image. Specifically, the grayscale image Gray is smoothed by Gaussian filtering, the vertical edge of the scale line is enhanced, and the horizontal edge of the scale line is suppressed;

[0015] Perform non-maximum suppression on the image enhanced by the edge filter to refine the edges of the scale lines in the image;

[0016] The host computer is used to receive the actual offset of the gap and compare the actual offset of the gap with the set threshold. If the actual offset of the gap exceeds the set threshold, an alarm signal is issued. If the actual offset of the gap is within the set threshold range, normal operation is performed.

[0017] Preferably, the process of identifying and processing the scale image on the outer side of the switch machine includes:

[0018] The scale image on the outer side of the switch machine is gray-scaled and the collected RGB image is converted into a gray-scale image. The conversion formula is as follows:

[0019] Gray=A1*R+A2*G+A3*B

[0020] Among them, A1, A2, and A3 are the calculation coefficients of the three-color values ​​of the RGB image, and their values ​​can be obtained empirically. The grayscale image is enhanced by edge filter processing. Specifically: the grayscale image Gray is smoothed by Gaussian filtering, the vertical edge of the scale line is enhanced, and the horizontal edge of the scale line is suppressed;

[0021] Perform non-maximum suppression on the image enhanced by the edge filter to refine the edges of the scale lines in the image;

[0022] The edge-enhanced and refined images are subjected to image regression prediction tasks through a deep learning model (such as the TOLO model) to achieve real-time target detection.

[0023] Preferably, the calculation formula for obtaining the actual offset of the gap by calculating the image offset scale is as follows:

[0024] L=|L1-L2|

[0025] L'=αL

[0026] Wherein, L is the image offset scale, L1 is the real-time scale reading, L2 is the standard position scale reading, L' is the actual offset of the notch, and α is the offset coefficient.

[0027] Preferably, it also includes a communication module for realizing information transmission inside the monitoring system; the communication module includes a network layer, a transport layer and an application layer.

[0028] Preferably, the network layer design is:

[0029] Assign a unique address to each slave machine;

[0030] The host computer uses the broadcast address 255, which is used for the host computer to send messages to all the slave computers at the same time and monitor them;

[0031] Set up the Lora module and configure the address of the Lora network to achieve two-way communication between the upper computer and the lower computer;

[0032] Set up a 4G module, configure a public IP on the host computer, and implement one-way communication from the lower computer to the upper computer through the 4G network;

[0033] Set up the Bluetooth module, design a dedicated mobile phone App, and use Bluetooth as the information transmission channel between the mobile phone and the lower computer.

[0034] Preferably, the transport layer design is:

[0035] The message packet in the transport layer includes address information, sequence number and confirmation number, packet size, flag, check code and data content; the address information is used to ensure that the message is correctly sent to a specific lower computer; the sequence number and confirmation number realize the unpacking, reassembly and confirmation of the message; the packet size indicates the boundary of the data packet to solve the problem of data sticking; the flag bit describes the current data packet; the check code is used to verify the integrity of the data to ensure that the data is not damaged during wireless transmission; the data content includes actual distance data, heartbeat reply and image data.

[0036] Preferably, the application layer design is:

[0037] It adopts a design combining fixed head and extended head;

[0038] The fixed header includes protocol identifier, content type, transmission mode, text length, and extended header information;

[0039] The extension header is used to identify the next extension header and also to indicate the end position of the current extension header and a byte stream.

[0040] In a second aspect, a monitoring method based on a railway switch machine indication gap real-time monitoring system is provided, comprising:

[0041] After the system is started, the mobile phone App controls the lower computer via Bluetooth to perform the acquisition of positioning and reverse reference images, and uploads the reference images to the upper computer via wireless networking; the lower computer is responsible for controlling the camera to perform cyclic image acquisition of the scale of the indicator rod outside the switch machine, and uses the built-in deep learning algorithm to perform scale recognition; the acquired images and processed distance data are stored in the cache area, waiting for polling by the upper computer;

[0042] When the host computer sends a distance request, the switch machine reads the distance data in the data buffer, packages it and sends it without waiting for a confirmation response; if the host computer does not receive the data or the data is parsed incorrectly, it will re-initiate the request and retry it up to three times; if all three attempts fail, the switch machine is considered to have lost connection, and an alarm will be given, requiring manual intervention and repair; when the host computer makes an image request, the switch machine reads the image data in the buffer, splits it into data packets and sends it. After receiving the data packet, the host computer replies with an ACK packet until the data reception process is completed.

[0043] In a third aspect, a computer-readable storage medium storing one or more programs is provided, wherein the one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform the method described.

[0044] According to a third aspect, a computing device is provided, including:

[0045] One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing the method.

[0046] (III) Beneficial effects

[0047] (1) The present invention provides a railway switch gap real-time monitoring system and a monitoring method thereof, which is a switch gap monitoring method based on image recognition and processing technology, and has many characteristics such as high detection accuracy, low detection error, adjustment instructions, easy installation and construction, and small maintenance work.

[0048] (2) The present invention provides a railway switch indicator gap real-time monitoring system and monitoring method thereof. By pasting a measurement target on the corresponding indicator rod within the shooting range of the video gap sensor, the offset generated is calculated using image recognition processing technology to obtain the offset of the actual indicator gap, thereby completing intelligent online detection of the switch indicator gap. This real-time monitoring technology can promptly discover and handle possible faults or hidden dangers, improve the reliability and stability of the railway signal system, and ensure the safe passage of trains. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a structural schematic diagram of a railway switch gap real-time monitoring system of the present invention;

[0050] Figure 2 It is a detailed structural schematic diagram of the railway switch machine gap real-time monitoring system of the present invention;

[0051] Figure 3 This is a schematic diagram of the turnout gap detection principle of the present invention;

[0052] Figure 4 This is a diagram of the communication process between the host computer, the slave computer and the mobile phone terminal of the present invention;

[0053] Figure 5 It is the reference position for positioning and reversing the turnout of the present invention. DETAILED DESCRIPTION

[0054] The following will be combined with the accompanying drawings 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.

[0055] Example

[0056] like Figure 1 As shown, an embodiment of the present invention provides a railway switch machine indication gap real-time monitoring system, comprising:

[0057] Switch machine, with scales engraved on the outer side of the switch machine;

[0058] A camera is fixed on the sleeve of the outer side indicator rod of the switch machine and is used to collect the scale image on the outer side indicator rod of the switch machine in real time;

[0059] The lower computer obtains the scale image on the outer side of the switch machine represented by the image acquisition module, wherein the scale image on the outer side of the switch machine represented by the image acquisition module includes a standard position scale reading image and a real-time scale image on the outer side of the switch machine represented by the image acquisition module, takes the scale image on the outer side of the switch machine represented by the image at the positioning reference position and the reverse reference position as the standard position scale reading image, identifies and processes the standard position scale reading image to obtain the standard position scale reading, identifies and processes the real-time scale image on the outer side of the switch machine represented by the image acquisition module to obtain the real-time scale reading, compares the real-time scale reading with the standard position scale reading to obtain the image offset scale, and calculates the actual offset of the gap through the image offset scale.

[0060] The host computer is used to receive the actual offset of the gap and compare the actual offset of the gap with the set threshold. If the actual offset of the gap exceeds the set threshold, an alarm signal is issued. If the actual offset of the gap is within the set threshold range, normal operation is performed.

[0061] like Figure 5 As shown, the tip of the turnout point rail is at two fitting positions, A and B, which are the positioning and reverse reference positions respectively. The lower computer collects the scale image on the outer side of the switch machine at the positioning and reverse reference positions through a camera as the standard position scale reading image and records it in the lower computer.

[0062] Furthermore, the process of identifying and processing the scale image on the outer side of the switch machine includes:

[0063] Image preprocessing first converts the collected RGB image into a grayscale image. Grayscale processing of the image has two effects: on the one hand, it can reduce the amount of calculation required for algorithm prediction; on the other hand, since only the scale reading needs to be recognized, removing color information can remove redundant features, thereby better improving the recognition effect. The image conversion formula is as follows:

[0064] Gray=A1*R+A2*G+A3*B

[0065] Among them, A1, A2, and A3 are the calculation coefficients of the three-color values ​​of the RGB image, and their values ​​can be obtained empirically. By further smoothing the grayscale image with Gaussian filtering, the noise in the image can be effectively suppressed, the detail interference in the non-edge area can be reduced, and the stability and accuracy of the edge detection algorithm can be improved;

[0066] In scale image recognition, since scale lines are vertical edges, it is very important to use edge filters to enhance vertical edges. By using edge enhancement algorithms or operators to enhance vertical edges and suppress horizontal edges, the edge structure of scale lines can be more effectively located, providing a clearer and more reliable information basis for subsequent image processing algorithms, thereby improving the accuracy and efficiency of distance prediction.

[0067] The thickness of the scale line edge has a significant impact on the detection results; if the edge is too thick, the detection algorithm cannot accurately locate the pixel position of the edge, resulting in prediction errors;

[0068] Non-maximum suppression is an important technique for edge refinement. The core of this method is to find the local maximum along the gradient direction and suppress the points of non-local maximum to achieve edge refinement. Non-maximum suppression can help remove redundant pixels to obtain finer and more accurate edge lines. Non-maximum suppression reduces the change of edge width while maintaining edge continuity, making the edge sharper and clearer.

[0069] After edge enhancement and refinement of the image, a deep neural network model (such as the YOLO model) is used to perform image regression prediction tasks.

[0070] like Figure 3 As shown, a scale is engraved on the outer side of the switch machine, and the offset of the gap is translated to the end position of the indicator rod outside the switch machine by using the rigidity of the indicator rod. The scale image on the outer side of the switch machine is collected in real time, and the scale reading and the reference position scale reading are identified to obtain the gap distance data by comparison. By adopting the monitoring method of scale reading, the measurement error caused by the change in the relative position of the camera and the indicator rod is effectively avoided. After the turnout is switched, the scale on the collected image will be offset. The calculation formula for the actual offset of the gap is obtained by calculating the image offset scale as follows:

[0071] L=|L1-L2|

[0072] L'=αL

[0073] Wherein, L is the image offset scale, L1 is the real-time scale reading, L2 is the standard position scale reading, L' is the actual offset of the notch, α is the offset coefficient, and α is a constant, whose size is determined by the relationship between the image offset scale and the actual offset of the notch in actual circumstances.

[0074] like Figure 2As shown in the figure, the host computer is installed inside the station and is mainly responsible for operating the monitoring and alarm system. The host computer reads distance data from the lower computer through wireless networking polling, performs storage, display, statistics, tabulation, judgment and alarm and other functions, and efficiently realizes the management and control of the station monitoring information. The host computer also provides an optimized human-machine interface to facilitate operators to manage effectively. When the received data exceeds the specified range and an alarm is required, the system not only provides visual prompts through the alarm light, but also notifies relevant personnel through text messages to ensure timely response to emergencies, greatly improving the safety management efficiency of the station. Furthermore, the monitoring system also includes a communication module for realizing information transmission within the monitoring system; the communication module includes a network layer, a transport layer and an application layer.

[0075] Furthermore, the network layer design is as follows:

[0076] Assign a unique address to each slave machine;

[0077] The host computer uses the broadcast address 255, which is used for the host computer to send messages to all the slave computers at the same time and monitor them;

[0078] Set up the Lora module and configure the address of the Lora network to achieve two-way communication between the upper computer and the lower computer; set up the 4G module, configure the public IP on the upper computer, and achieve one-way communication from the lower computer to the upper computer through the 4G network; set up the Bluetooth module, design a dedicated mobile phone App, and use Bluetooth as the information transmission channel between the mobile phone and the lower computer.

[0079] Furthermore, the transport layer design is as follows:

[0080] Since the LoRa module and the Bluetooth module only implement the transparent transmission function equivalent to the data link layer, a special communication protocol needs to be designed for message identification. When designing the communication protocol, a layered approach is adopted. The core idea is to decompose complex communication tasks into smaller and more manageable parts. At the transport layer, you only need to focus on ensuring the accurate delivery of the message content without involving the specific details of the message content. Based on this design principle, the packet header of the transport layer is set to a fixed byte size. This design not only simplifies the structure of the protocol, making it easier to manage and maintain, but also effectively saves bandwidth resources, which is essential for maintaining efficient and stable network communication.

[0081] In this study, two communication modes were designed based on computer network communication protocols for different transmission requirements: TCP mode and UDP mode. In these two modes, the communication starts at the station. The station clearly specifies the selected communication mode and the specific information it carries in the first data packet sent, and then the switch machine is responsible for data transmission.

[0082] Table 1 Transport layer protocol design example

[0083]

[0084]

[0085] As shown in Table 1, the message packet in the transport layer includes address information, sequence number and confirmation number, packet size, flag, checksum and data content; the address information is used to ensure that the message is correctly sent to a specific lower computer; the sequence number and confirmation number realize the unpacking, reassembly and confirmation of the message; the packet size indicates the data packet boundary to solve the problem of data sticking; the flag bit describes the current data packet; the checksum is used to verify the integrity of the data to ensure that the data is not damaged during wireless transmission; the data content includes the actual distance data, heartbeat reply and picture data.

[0086] The transport layer header is M bytes in total. When M=10, the header information of each data packet is reduced by 10 bytes compared with the communication packet header (20 bytes) of the TCP transport protocol. Considering that the communication data is mainly floating point type distance data, and each floating point number occupies 4 bytes, the average data volume of each distance data packet is reduced by 41.6%.

[0087] Furthermore, the application layer design is as follows:

[0088] On the premise that the transport layer protocol guarantees the integrity and accuracy of data transmission, the application layer protocol is further designed. The core responsibility of the application layer protocol is to process specific data and content, which requires high efficiency in information judgment and processing, and also requires consideration of the extensibility of the protocol header. Based on this, referring to the IPv6 communication protocol, a design combining fixed headers and extended headers is adopted, which flexibly utilizes the characteristics of fixed headers and extended headers to adapt to different communication needs and improve communication efficiency and flexibility.

[0089] Table 2 Application layer protocol fixed header design implementation example

[0090]

[0091]

[0092] As shown in Table 2, a design combining a fixed head and an extended head was adopted;

[0093] The fixed header includes the protocol identifier, content type, transmission mode, text length, and extended header information; the protocol identifier occupies 4 bits, which is enough to represent 16 different protocol versions, facilitating future protocol updates. Since the content type mainly involves text and monitoring images in the turnout monitoring scenario, only 2 bits are needed to represent it. The transmission mode (select one of LoRa, 4G, or Bluetooth) is represented by 2 bits. The text length occupies 16 bits, which is used to indicate the length of the data packet content (excluding the packet header). 16 bits can represent up to 65535 bytes, which is enough to meet the transmission requirements of a single data packet. Finally, an 8-bit extended header information is used to indicate the type of the first extended header;

[0094] The extension header includes an 8-bit field for identifying the next extension header, another 8-bit field for indicating the end position of the current extension header, and a byte stream. In order to distinguish different semantic contents, " / " is used as a separator. In addition, some extension headers and their corresponding values ​​are shown in Table 3. By designing a dedicated application layer protocol, the traditional key-value pair header in the HTTP protocol is abandoned, thereby reducing the data bandwidth loss caused by the key value in the header.

[0095] Table 3 Application layer protocol extension header part design implementation example

[0096]

[0097] After the monitoring system is installed and debugged on site to meet the standards, the lower computer is controlled by the mobile phone App to collect and record the positioning and reverse standard images; to ensure the image quality, an industrial-grade camera is used for image acquisition; the camera is fixed on the outer sleeve of the switch machine to collect the scale image of the switch machine rod in real time; the camera is connected to the lower computer through the USB serial port to transmit the collected image data to the lower computer uninterruptedly;

[0098] The lower computer has a built-in deep learning model. After receiving the image data, it performs recognition processing to obtain the real-time scale reading, compares it with the reference position scale reading to obtain the gap distance data and stores it in the buffer.

[0099] like Figure 4 As shown, another embodiment of the present invention provides a monitoring method based on a railway switch machine indication gap real-time monitoring system, comprising:

[0100] After the system is started, the mobile phone App controls the lower computer via Bluetooth to perform the acquisition of positioning and reverse reference images, and uploads the reference images to the upper computer via wireless networking; the lower computer is responsible for controlling the camera to perform cyclic image acquisition of the scale of the indicator rod outside the switch machine, and uses the built-in deep learning algorithm to perform scale recognition; the acquired images and processed distance data are stored in the cache area, waiting for polling by the upper computer;

[0101] When the host computer sends a distance request, the switch machine reads the distance data in the data buffer, packages it and sends it without waiting for a confirmation response; if the host computer does not receive the data or the data parsing error occurs, it will re-initiate the request and retry up to three times; if all three attempts fail, the switch machine is considered to have lost connection, an alarm will be given, and manual intervention is required to repair it; when the host computer makes an image request, the switch machine reads the image data in the buffer, splits it into data packets and sends it, and the host computer replies with an ACK packet after receiving the data packet until the data reception process is completed.

[0102] Yet another embodiment of the present invention provides a computer-readable storage medium storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform a method.

[0103] Yet another embodiment of the present invention provides a computing device, comprising:

[0104] One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing the method.

[0105] The monitoring technologies for the switch machine gap status are mainly divided into three types: photoelectric detection, mechanical contact and video monitoring. The switch machine gap monitoring method based on image recognition and processing technology has many characteristics such as high detection accuracy, low detection error, adjustment instructions, easy installation and construction, and small maintenance work. At the same time, due to its strong compatibility, it is conducive to future upgrades and modifications. This technology pastes the measurement target on the corresponding indication rod within the shooting range of the video gap sensor, and uses image recognition and processing technology to calculate the offset of the actual indication gap, thereby completing the intelligent online detection of the switch machine gap. This real-time monitoring technology can promptly discover and deal with possible faults or hidden dangers, improve the reliability and stability of the railway signal system, and ensure the safe passage of trains.

[0106] The embodiment of the present application can be provided as method or computer program product.Therefore, the application can adopt the form of complete hardware embodiment, complete software embodiment or the embodiment in combination with software and hardware.And, the application can adopt the form of computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.The scheme in the embodiment of the present application can adopt various computer languages ​​to realize, for example, object-oriented programming language Java and literal scripting language JavaScript etc.

[0107] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0108] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0109] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0110] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

Claims

1. A railway switch machine indication gap real-time monitoring system, characterized in that: include: A camera is fixed on the sleeve of the outer side indicator rod of the switch machine and is used to collect the scale image on the outer side indicator rod of the switch machine in real time; The lower computer obtains the scale image on the outer side of the switch machine represented by the image acquisition module, wherein the scale image on the outer side of the switch machine represented by the image acquisition module includes a standard position scale reading image and a real-time scale image on the outer side of the switch machine represented by the image acquisition module, takes the scale image on the outer side of the switch machine represented by the image at the positioning reference position and the reverse reference position as the standard position scale reading image, identifies and processes the standard position scale reading image to obtain the standard position scale reading, identifies and processes the real-time scale image on the outer side of the switch machine represented by the image acquisition module to obtain the real-time scale reading, compares the real-time scale reading with the standard position scale reading to obtain the image offset scale, and calculates the actual offset of the gap through the image offset scale. The process of identifying and processing the scale image on the outer side of the switch machine includes: The scale image on the outer side of the switch machine is gray-scaled and the collected RGB image is converted into a gray-scale image. The conversion formula is as follows: Gray=A1*R+A2*G+A3*B Among them, A1, A2, and A3 are the calculation coefficients of the three-color values ​​of the RGB image. Specifically, the grayscale image Gray is smoothed by Gaussian filtering, the vertical edge of the scale line is enhanced, and the horizontal edge of the scale line is suppressed; Perform non-maximum suppression on the image enhanced by the edge filter to refine the edges of the scale lines in the image; The image after edge enhancement and refinement is subjected to image regression prediction task through a deep neural network model to achieve real-time target detection; The host computer is used to receive the actual offset of the gap and compare the actual offset of the gap with the set threshold. If the actual offset of the gap exceeds the set threshold, an alarm signal is issued. If the actual offset of the gap is within the set threshold range, normal operation is performed.

2. A railway switch machine indication gap real-time monitoring system according to claim 1, characterized in that: The calculation formula for the actual offset of the gap calculated by the image offset scale is as follows: L=|L1-L2 L'=αL Wherein, L is the image offset scale, L1 is the real-time scale reading, L2 is the standard position scale reading, L' is the actual offset of the notch, and α is the offset coefficient.

3. A railway switch machine indication gap real-time monitoring system according to claim 1, characterized in that: It also includes a communication module for realizing information transmission inside the monitoring system; the communication module includes a network layer, a transport layer and an application layer.

4. A railway switch machine indication gap real-time monitoring system according to claim 3, characterized in that: The network layer design is: Assign a unique address to each slave machine; The host computer uses the broadcast address 255, which is used for the host computer to send messages to all the slave computers at the same time and monitor them; Set up the LoRa module and configure the address of the LoRa network to achieve two-way communication between the upper computer and the lower computer; Set up a 4G module, configure a public IP on the host computer, and implement one-way communication from the lower computer to the upper computer through the 4G network; Set up the Bluetooth module, design a dedicated mobile phone App, and use Bluetooth as the information transmission channel between the mobile phone and the lower computer.

5. A railway switch machine indication gap real-time monitoring system according to claim 4, characterized in that: The transport layer design is: The message packet in the transport layer includes address information, sequence number and confirmation number, packet size, flag bit, checksum and data content; the address information is used to ensure that the message is correctly sent to a specific lower computer; the sequence number and confirmation number realize the unpacking, reassembly and confirmation of the message; the packet size indicates the data packet boundary to solve the problem of data sticking; the flag bit describes the current data packet; The checksum is used to verify the integrity of the data and ensure that the data is not damaged during wireless transmission; the data content includes the actual distance data, heartbeat reply and image data; the transport layer header is a total of M bytes, which contains the following information: protocol identifier, content type, transmission mode, text length and extended header information.

6. A railway switch machine indication gap real-time monitoring system according to claim 5, characterized in that: The application layer design is: It adopts a design combining fixed head and extended head; The fixed header includes: protocol identifier, content type, transmission method, text length, and extended header information; The extension header is used to identify the next extension header and to indicate the end position of the current extension header and a byte stream.

7. A monitoring method based on the railway switch machine indication gap real-time monitoring system according to any one of claims 1 to 6, characterized in that: include: After the system is started, the mobile phone App controls the lower computer via Bluetooth to perform the acquisition of positioning and reverse reference images, and uploads the reference images to the upper computer via wireless networking; the lower computer is responsible for controlling the camera to perform cyclic image acquisition of the scale of the indicator rod outside the switch machine, and uses the built-in deep learning algorithm to perform scale recognition; the acquired images and processed distance data are stored in the cache area, waiting for polling by the upper computer; When the host computer sends a distance request, the switch machine reads the distance data in the data buffer, packages it and sends it without waiting for a confirmation response; if the host computer does not receive the data or the data is parsed incorrectly, it will re-initiate the request and retry it up to three times; if all three attempts fail, the switch machine is considered to have lost connection, and an alarm will be given, requiring manual intervention and repair; when the host computer makes an image request, the switch machine reads the image data in the buffer, splits it into data packets and sends it. After receiving the data packet, the host computer replies with an ACK packet until the data reception process is completed.

8. A computer-readable storage medium storing one or more programs, characterized in that: The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform the method of claim 7 .

9. A computing device, characterized in that include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing the method according to claim 7.

Images