KR102238669B1 - System and method for predicting and early warning surface temperature of rail using deep learning - Google Patents

Abstract

The present invention provides a method for rail surface temperature prediction and early warning using deep learning which accurately predicts rail surface temperature and rail inside temperature. According to the present invention, the method for rail surface temperature prediction and early warning using deep learning is performed by a computer device and comprises: (a) a step of acquiring surface temperature information of a rail from a noncontact temperature sensor; (b) a step of training a first correlation model based on existing rail temperature information acquired from a server; (c) a step of deriving correction surface temperature information by applying the first correlation model based on the surface temperature information acquired from the noncontact temperature sensor; and (d) a step of calculating prediction surface temperature information by applying a second correlation model based on weather forecast information including air temperature information, prediction time information, and the correction surface temperature information. The existing rail temperature information includes existing rail inside temperature and existing rail surface temperature measured by an external rail temperature detection device.

Description

Rail surface temperature prediction and early warning system and method using deep learning {SYSTEM AND METHOD FOR PREDICTING AND EARLY WARNING SURFACE TEMPERATURE OF RAIL USING DEEP LEARNING}

The present invention relates to a rail surface temperature prediction and early warning system and method using deep learning.

In general, during train operation, the surface temperature and internal temperature of railroad rails are measured and managed, and maintenance work and train operation regulations are enforced if they meet the threshold values according to train operation standards. The rail surface temperature is measured by attaching a temperature sensor to the surface of the railroad rail, and the internal rail temperature (ie, rail longitudinal temperature) is measured only in some sections according to the installation conditions of the operating regulations.

Preferably, the operating regulations related to rail temperature in the railroad use the internal rail temperature, but the number of sensors installed on the existing rail is small, and thus the rail surface temperature data is being replaced.

However, there is a problem that the rail surface temperature deviates from the rail internal temperature. In particular, the sensor for measuring the temperature inside the rail is difficult to install on the running rail due to the railroad rail maintenance and rail stability problems.

In addition, even when a temperature sensor is installed to measure the surface temperature of the rail, there is a limitation that it should be installed only in the curved, sunny, and poorly ventilated section of the rail or in a place that does not interfere with maintenance work.

In this regard, Korean Patent Registration No. 10-0471346 discloses a remote branch line monitoring system using wired and wireless lines in a railroad turnout.

The present invention is to solve the problems of the prior art described above, artificial intelligence (deep learning, etc.) learning the atmospheric temperature, rail surface temperature, rail internal temperature, etc. at a reference position, and measured at multiple points using the learning result. It is intended to provide a rail surface temperature prediction and early warning system and method thereof using deep learning by predicting more accurate rail surface temperature and rail internal temperature using the rail surface temperature.

However, the technical problem to be achieved by the present embodiment is not limited to the technical problem as described above, and other technical problems may exist.

As a technical means for achieving the above-described technical problem, the rail surface temperature prediction and early warning method using deep learning, performed by the computer device according to the first aspect of the present invention, includes (a) the non-contact temperature sensor of the railroad rail. Obtaining surface temperature information; (b) learning a first correlation model based on the existing rail temperature information obtained from the server; (c) deriving corrected surface temperature information by applying a first correlation model based on the surface temperature information obtained from the non-contact temperature sensor; And (d) calculating predicted surface temperature information by applying a second correlation model based on weather forecast information including corrected surface temperature information and predicted time information and atmospheric temperature information, wherein the existing rail temperature information is It includes the existing rail internal temperature and the existing rail surface temperature measured by the external rail temperature detection device.

In addition, a rail surface temperature prediction and early warning system using deep learning according to a second aspect of the present invention includes a memory for storing a rail surface temperature prediction and early warning program using deep learning; And a processor executing the program, wherein the processor obtains surface temperature information of the railroad rail from the non-contact temperature sensor as the program is executed, and creates a first correlation model based on the existing rail temperature information obtained from the server. It learns and derives the corrected surface temperature information by applying the first correlation model based on the surface temperature information obtained from the non-contact temperature sensor, and provides weather forecast information including the corrected surface temperature information and predicted time information and atmospheric temperature information. As a basis, the rail surface temperature is calculated by applying the second correlation model, and the existing rail temperature information includes the existing rail internal temperature and the existing rail surface temperature measured by an external rail temperature detection device.

The present invention has the advantage of easy maintenance of a railroad rail in the future by installing a non-contact temperature sensor. In addition, by measuring the rail surface temperature through a non-contact temperature sensor that is relatively easy to install compared to installing a temperature sensor inside the rail, it can be easily applied during railroad operation.

1 is a block diagram of a rail surface temperature prediction and early warning system using deep learning according to an embodiment of the present invention.
2 is a schematic configuration diagram illustrating a process of learning a first correlation model based on existing rail temperature information according to an embodiment of the present invention.
3 is a schematic diagram illustrating a process of learning a second correlation model based on weather forecast information including corrected surface temperature information, predicted time information, and air temperature information according to an embodiment of the present invention. .
4 is a diagram schematically showing a state in which a non-contact temperature sensor according to an embodiment of the present invention is disposed on a railroad rail.
5 is a flowchart illustrating a method of predicting rail surface temperature and early warning using deep learning according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art can easily implement the present invention. However, the present invention may be implemented in various different forms and is not limited to the embodiments described herein. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the specification, when a part is said to be "connected" with another part, this includes not only "directly connected" but also "electrically connected" with another element interposed therebetween. . In addition, when a part "includes" a certain component, it means that other components may be further included rather than excluding other components unless specifically stated to the contrary.

In the present specification, the term "unit" includes a unit realized by hardware, a unit realized by software, and a unit realized using both. Further, one unit may be realized by using two or more hardware, or two or more units may be realized by one piece of hardware. Meanwhile,'~ unit' is not meant to be limited to software or hardware, and'~ unit' may be configured to be in an addressable storage medium or configured to reproduce one or more processors. Thus, as an example,'~ unit' refers to components such as software components, object-oriented software components, class components, and task components, processes, functions, properties, and procedures. , Subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays and variables. Components and functions provided in the'~ units' may be combined into a smaller number of elements and'~ units', or may be further separated into additional elements and'~ units'. In addition, components and'~ units' may be implemented to play one or more CPUs in a device or a security multimedia card.

The "system" mentioned below may be implemented as a computer or portable terminal that can access a server or other terminal through a network. Here, the computer includes, for example, a notebook equipped with a web browser, a desktop, and a laptop.

The portable terminal is, for example, a wireless communication device that guarantees portability and mobility, and may be any kind of handheld-based wireless communication device such as, for example, a smart phone, a tablet PC, or a notebook.

In addition, a network refers to a connection structure in which information exchange is possible between nodes such as terminals and servers, and includes a local area network (LAN), a wide area network (WAN), and the Internet (WWW). : World Wide Web), wired and wireless data networks, telephone networks, wired and wireless television networks, etc. Examples of wireless data networks include 3G, 4G, 5G, 3GPP (3rd Generation Partnership Project), LTE (Long Term Evolution), WIMAX (World Interoperability for Microwave Access), Wi-Fi, Bluetooth communication, infrared communication, and ultrasound. Communication, Visible Light Communication (VLC), LiFi, and the like are included, but are not limited thereto.

On the other hand, in the case of high-speed rail, rail temperature detection devices are installed (18 locations) to measure and manage the temperature of the rail in real time. If it is less than 70km/h, it is recommended to drive slowly at 70km/h or less, if it is 55℃ or more and less than 60℃, slow drive at 230km/h or less, and when the rail temperature is 50℃ or more and less than 55℃, it is recommended to drive carefully. In addition, in the case of general railroads, 76 locations per KORAIL facility office are measured (daily: 12, 14, and 16:00) to reinforce the inspection of trains when the temperature rises, and safety facilities are provided for use in hot weather. have.

1 is a block diagram of a rail surface temperature prediction and early warning system using deep learning according to an embodiment of the present invention.

Referring to FIG. 1, a rail surface temperature prediction and early warning system 100 using deep learning includes a memory 110 for storing a rail surface temperature prediction and early warning program using deep learning, and a processor 130 for executing the program. Including, but, as the program is executed, the processor 130 obtains the surface temperature information of the railroad rail from the non-contact temperature sensor 20, and a first correlation based on the existing rail temperature information obtained from the server 10 The model is trained, and based on the surface temperature information obtained from the non-contact temperature sensor, the first correlation model is applied to derive the corrected surface temperature information, and the weather forecast including the corrected surface temperature information, prediction time information, and atmospheric temperature information The rail surface temperature can be calculated by applying the second correlation model based on the information. Here, the existing rail temperature information includes the existing rail internal temperature and the existing rail surface temperature measured by the external rail temperature detection device.

Accordingly, it is possible to predict the internal temperature of the rail using a plurality of existing rail temperature detection devices. In addition, by predicting the rail surface temperature in advance using weather forecast information, it is possible to warn of danger early.

Meanwhile, the artificial intelligence technology applied to the present invention is classified into machine learning, knowledge reasoning, visual intelligence, and language intelligence, and the present invention is not limited thereto, although the deep learning method of machine learning may be applied as an example. Machine learning, clustering, etc. may be applied.

Hereinafter, the rail surface temperature prediction and early warning system 100 using deep learning will be described in detail.

Specifically, the rail surface temperature prediction and early warning system 100 using deep learning includes a memory 110, a communication unit 120 and a processor 130, and includes a desktop computer, a laptop computer, a server, and a portable terminal. It may be driven by an electronic device having a communication function such as.

The memory 110 stores a rail surface temperature prediction and early warning program (or at least one instruction) using deep learning. In this case, the memory 110 collectively refers to a nonvolatile storage device that continuously maintains stored information even when power is not supplied, and a volatile storage device that requires power to maintain the stored information.

The communication unit 120 includes one or more components for communicating with external devices such as a server 10 that provides existing rail temperature information and weather forecast information, and a non-contact temperature sensor 20 that measures the surface temperature of a railroad rail. can do. For example, the communication unit 120 is a LAN (Local Area Network) or Bluetooth, BLE (Bluetooth Low Energy), infrared (IrDA, infrared Data Association), Zigbee communication that performs wired or wireless data communication. It may include a component capable of performing at least one of. Alternatively, the communication unit 120 may transmit and receive wireless signals through a mobile communication network or a broadcast communication network.

The processor 130 controls the overall operation of the rail surface temperature prediction and early warning system 100 using deep learning. To this end, the processor 130 may include at least one of a random access memory (RAM), a read-only memory (ROM), a CPU, a graphic processing unit (GPU), and a bus.

In addition, the processor 130 performs a rail surface temperature prediction and early warning process by executing a program stored in the memory 110.

2 is a schematic configuration diagram illustrating a process of learning a first correlation model based on existing rail temperature information according to an embodiment of the present invention.

3 is a schematic configuration diagram for explaining a process of learning a second correlation model based on weather forecast information including corrected surface temperature information, predicted time information, and air temperature information according to an embodiment of the present invention. .

4 is a view schematically showing a state in which a non-contact temperature sensor according to an embodiment of the present invention is disposed on a railroad rail.

2 and 3, the processor 130 obtains surface temperature information of the railroad rail 40 from the non-contact temperature sensor 20, and first correlation based on the existing rail temperature information obtained from the server 10. The relationship model 210 may be trained.

In this case, the first correlation model 210 may learn a correlation between the rail inner temperature and the rail surface temperature based on the obtained surface temperature information and the existing rail temperature information.

That is, the existing rail temperature information may include the existing rail internal temperature and the existing rail surface temperature measured by the external rail temperature detection device.

As an example, the server 10 is an external server such as an external rail temperature detection device installed in the past, and may measure and manage rail temperature information. In addition, the server 10 may include a database for storing the existing rail internal temperature and the existing rail surface temperature.

For example, the first correlation model 210 may derive the corrected surface temperature information based on the surface temperature information of the rail 40 received from the non-contact temperature sensor 20 in real time.

For example, the processor 130 generates a plurality of learning data from the existing rail temperature information acquired through the communication unit 120 and the surface temperature information of the railroad rail 40 acquired in real time from the non-contact temperature sensor 20, and , A first correlation model 210 may be generated by learning a correlation between the rail inner temperature and the rail surface temperature through deep learning machine learning.

In a further embodiment, the processor 130 may collect the existing rail surface temperature and the existing rail internal temperature from the server 10, and collect the rail surface temperature in real time from the non-contact temperature sensor 20. Subsequently, preprocessing may be performed to be used for deep learning learning, and the preprocessed data may be generated as a plurality of training data.

In addition, the processor 130 may perform machine learning through the first correlation model 210 to define a correlation between the rail surface temperature and the rail internal temperature. In this case, when the accumulated number of internal/surface temperature data for each temperature range is greater than or equal to a predetermined reference value as machine learning is repeatedly performed, internal/surface temperature data may be set as reference data. At this time, the processor 130 applies the first correlation model 210 to correlate between any one rail internal temperature data and a plurality of rail surface temperature data based on the existing rail temperature information and surface temperature information corresponding to the reference data. You can derive a relationship. For example, the derived internal/surface temperature data may be stored in a database in a matrix form.

In addition, when obtaining a plurality of surface temperature information in real time from the non-contact temperature sensor 20, the processor 130 derives normalized data through normal distribution analysis of the reference data and the acquired rail surface temperature data, and Through comparative analysis of the degree of variance for, it is possible to derive the corrected rail surface temperature that matches the reference data.

Referring to FIG. 3, the processor 130 may learn the second correlation model 220 based on the corrected surface temperature information and air temperature information obtained from the server 10.

The second correlation model 220 may learn a correlation between the rail surface temperature and the air temperature based on the corrected surface temperature information and the air temperature information obtained from the server 10.

For example, the server 10 may be an external server such as a weather observation server, and may include a database storing weather forecast information. At this time, the weather forecast information may include predicted time information and air temperature information. For example, the weather forecast information refers to the forecast date and the air temperature for each time of the date.

For example, the second correlation model 220 may calculate the predicted rail surface temperature based on the corrected surface temperature information derived from the first correlation model 210 and the weather forecast information obtained from the server 10. have.

For example, the processor 130 generates a plurality of training data from the corrected surface temperature information derived from the first correlation model 210 and the weather forecast information obtained from the server 10, and through deep learning machine learning. A second correlation model 220 may be generated by learning a correlation between the rail surface temperature and the air temperature.

In a further embodiment, the processor 130 may collect the predicted time and the air temperature from the server 10 and collect the corrected rail surface temperature. In addition, the processor 130 may perform machine learning through the second correlation model 220 in order to define a correlation between the rail surface temperature and the air temperature. In this case, as the machine learning is repeatedly performed, when the cumulative number of surface/atmospheric temperature data for each temperature range is greater than or equal to a predetermined reference value, the surface/atmospheric temperature data may be set as reference data. At this time, the processor 130 applies the second correlation model 220 to correlate any one rail surface temperature data and a plurality of air temperature data based on the corrected surface temperature information and weather forecast information corresponding to the reference data. Can be derived. For example, the derived surface/atmospheric temperature data may be matched with the internal/surface temperature data derived through the first correlation model 210, and the matched internal/surface/atmospheric temperature data is in a matrix form. It can be stored in a database.

In addition, when receiving the weather forecast information from the external server 10, the processor 130 derives normalized data through normal distribution analysis of the reference data and the received air temperature data, and compares the degree of variance for the normalized data. Through the analysis, it is possible to derive the predicted rail surface temperature that meets the reference data.

Accordingly, the user may be provided with a predicted rail surface temperature based on weather forecast information for reference when planning daily, weekly and monthly train operation. Using such predicted rail surface temperature, early warning can be performed, and human and material resources can be efficiently distributed.

In addition, the processor 130 may provide early warning information when the calculated predicted surface temperature information matches a preset temperature range. At this time, the early warning information may include stopping the operation, slow driving and careful driving.

Exemplarily, the preset temperature range is: stop running when the rail temperature is above 64℃, slow driving at 70km/h or less when the rail temperature is more than 60℃ and less than 64℃, and slow driving at 230km/h or less when the rail temperature is 55℃ or more and less than 60℃. When the temperature is 50°C or more and less than 55°C, it may be set based on the rules for handling high-speed railways, which are prescribed for careful operation, but are not limited thereto.

Accordingly, even when the weather forecast information changes from time to time, the user may be provided with early warning information updated in real time to refer to train operation.

On the other hand, as shown in Figure 4, the non-contact temperature sensor 20 is disposed a plurality of spaced apart on the outer surface of a pair of rails 40 disposed between a plurality of sleepers 30, sensors including infrared or laser scanning method Can be For example, the non-contact temperature sensor 20 may measure a plurality of surface temperature information of the railroad rail 40 in real time, and the processor 130 may receive a plurality of measured surface temperature information.

Hereinafter, a description of a configuration that performs the same function among the configurations illustrated in FIGS. 1 to 4 will be omitted.

5 is a flowchart illustrating a method of predicting rail surface temperature and early warning using deep learning according to an embodiment of the present invention.

Referring to Figure 5, the rail surface temperature prediction and early warning method using deep learning of the present invention is the step of acquiring surface temperature information of the railroad rail 40 from the non-contact temperature sensor 20 (S110), from the server 10 Learning the first correlation model 210 based on the obtained existing rail temperature information (S120), applying the first correlation model 210 based on the surface temperature information obtained from the non-contact temperature sensor 20 By applying the second correlation model 220 based on the step of deriving the corrected surface temperature information (S130) and the weather forecast information including the corrected surface temperature information, the predicted time information, and the air temperature information, the predicted surface temperature information is calculated. It includes a step of calculating (S140).

Here, the existing rail temperature information may include the existing rail internal temperature and the existing rail surface temperature measured by the external rail temperature detection device.

The non-contact type temperature sensor 20 may be a temperature sensor including an infrared or laser scanning method, which is disposed a plurality of spaced apart on the outer surface of the pair of rails 40 disposed between the plurality of sleepers 30.

For example, the first correlation model 210 may learn a correlation between the rail inner temperature and the rail surface temperature based on the obtained surface temperature information and the existing rail temperature information. The second correlation model 220 may learn a correlation between the rail surface temperature and the air temperature based on the corrected surface temperature information and the air temperature information obtained from the server 10.

In addition, the rail surface temperature prediction and early warning method using deep learning of the present invention may further include providing early warning information when the calculated predicted surface temperature information matches a preset temperature range. At this time, the early warning information may include stopping the operation, slow driving and careful driving.

An embodiment of the present invention may also be implemented in the form of a recording medium including instructions executable by a computer, such as a program module executed by a computer. Computer-readable media can be any available media that can be accessed by a computer, and includes both volatile and nonvolatile media, removable and non-removable media. Further, the computer-readable medium may include a computer storage medium. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data.

Although the methods and systems of the present invention have been described in connection with specific embodiments, some or all of their components or operations may be implemented using a computer system having a general-purpose hardware architecture.

The above description of the present invention is for illustrative purposes only, and those of ordinary skill in the art to which the present invention pertains will be able to understand that other specific forms can be easily modified without changing the technical spirit or essential features of the present invention. will be. Therefore, it should be understood that the embodiments described above are illustrative in all respects and are not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as being distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be construed as being included in the scope of the present invention. do.

10: server
110: memory
120: communication department
130: processor
20: non-contact temperature sensor
210: first order correlation model
220: quadratic correlation model
30: sleepers
40: rail

Claims (10)

In the rail surface temperature prediction and early warning method using deep learning performed by a computer device,
(a) obtaining information on the surface temperature of the railroad rail from the non-contact temperature sensor;
(b) learning a first correlation model based on the existing rail temperature information obtained from the server;
(c) deriving corrected surface temperature information by applying the first correlation model based on the surface temperature information obtained from the non-contact temperature sensor; And
(d) calculating predicted surface temperature information by applying a second correlation model based on weather forecast information including the corrected surface temperature information and predicted time information and atmospheric temperature information,
The above existing rail temperature information is
Including the existing rail internal temperature and the existing rail surface temperature measured by the external rail temperature detection device,
Rail surface temperature prediction and early warning method using deep learning.
The method of claim 1,
The first correlation model is
Based on the obtained surface temperature information and the existing rail temperature information, the correlation between the rail inner temperature and the rail surface temperature is learned,
Rail surface temperature prediction and early warning method using deep learning.
The method of claim 1,
The second correlation model is
The correlation between the rail surface temperature and the air temperature is learned based on the corrected surface temperature information and the air temperature information obtained from the server,
Rail surface temperature prediction and early warning method using deep learning.
The method of claim 1,
(e) further comprising the step of providing early warning information when the calculated predicted surface temperature information matches a preset temperature range,
The early warning information is to include stop, slow driving and careful driving,
Rail surface temperature prediction and early warning method using deep learning.
The method of claim 1,
The non-contact temperature sensor
A plurality of spaced apart are arranged on the outer surface of a pair of rails arranged between a plurality of sleepers,
It is a temperature sensor including infrared or laser scanning method,
Rail surface temperature prediction and early warning method using deep learning.
In the rail surface temperature prediction and early warning system using deep learning,
A memory for storing rail surface temperature prediction and early warning programs using deep learning; And
Including a processor that executes the program,
The processor, as the program is executed,
Acquire the surface temperature information of the railroad rail from the non-contact temperature sensor,
Learning the first correlation model based on the existing rail temperature information obtained from the server,
Applying the first correlation model based on the surface temperature information obtained from the non-contact temperature sensor to derive the corrected surface temperature information,
The rail surface temperature is calculated by applying a second correlation model based on weather forecast information including the corrected surface temperature information, predicted time information, and air temperature information,
The above existing rail temperature information is
Including the existing rail internal temperature and the existing rail surface temperature measured by the external rail temperature detection device,
Rail surface temperature prediction and early warning system using deep learning.
The method of claim 6,
The first correlation model is
Based on the obtained surface temperature information and the existing rail temperature information, the correlation between the rail inner temperature and the rail surface temperature is learned,
Rail surface temperature prediction and early warning system using deep learning.
The method of claim 6,
The second correlation model is
The correlation between the rail surface temperature and the air temperature is learned based on the corrected surface temperature information and the air temperature information obtained from the server,
Rail surface temperature prediction and early warning system using deep learning.
The method of claim 6,
The processor is
When the calculated predicted surface temperature information matches a preset temperature range, early warning information is provided,
The early warning information includes stop, slow driving and careful driving,
Rail surface temperature prediction and early warning system using deep learning.
The method of claim 6,
The non-contact temperature sensor
A plurality of spaced apart are arranged on the outer surface of a pair of rails arranged between a plurality of sleepers,
It is a sensor including an infrared or laser scanning method,
Rail surface temperature prediction and early warning system using deep learning.

Images