CN114743166A - Method for detecting brake of railway wagon - Google Patents

Abstract

The invention discloses a method for detecting a brake of a railway freight car, comprising: acquiring parameter information of the railway freight car; acquiring the positioning information of the RGV car in real time when the RGV car travels in a first direction, and obtaining the positioning information of the RGV car in real time based on the parameter information and the The positioning information generates the first target detection area of multiple carriages in real time, and the RGV car collects a plurality of first detection images in the first target detection area; The detection image is input into the neural network model to generate a first defect detection result; based on the first defect detection result, a second target detection area is generated when the RGV trolley travels in the second direction, and the RGV trolley collects the second target Detecting a plurality of second detection images in the area, inputting the plurality of second detection images into the neural network model to generate a second defect detection result; generating a second defect detection result based on the first defect detection result and the second defect detection result Brake defect inspection report.

Description

Brake detection method of a railway freight car

technical field

The invention relates to the technical field of rail vehicle detection, in particular to a brake detection method of a railway freight vehicle.

Background technique

To ensure safe operation, fault detection plays a major role in the transportation sector. There are many typical examples of applying fault detection in railway, aviation, marine and highway bridge maintenance. Due to the heavy responsibility in the field of transportation, once the important equipment fails, it will cause huge loss of personnel and property. Therefore, many countries in the world have invested a lot of manpower, material resources and financial resources in the research of fault detection. Fault detection has become one of the research hotspots in the field of transportation today.

For railway freight cars, the operation and maintenance party has set up a large number of train inspection yards to test the freight cars to ensure safe operation. At present, the inspection field mainly relies on manual maintenance, which has problems such as heavy workload and inability to standardize test results. For this problem, our company's previous research idea is: based on the RGV trolley, the collection of the bottom pictures of the railway freight car is continuously carried out, and then the pictures containing the brakes are found from the pictures, and finally the fault diagnosis is carried out. However, there are two problems in this research idea: all the detection images are screened on the RGV trolley, but this requires high computing power on the trolley, and it is difficult to meet the requirements in terms of heat dissipation, computing power, or size (computing power Strong, large size), according to this idea, you can only reduce the speed of the car to adapt to the problem of insufficient computing power of the car, but the efficiency will be reduced; the detection image is transmitted to the server, and the computing power of the server is used for screening, However, this method has too high requirements for network transmission bandwidth, and although the processing speed becomes faster, the transmission takes too long and cannot meet the real-time requirements.

To sum up, the existing brake detection methods for railway freight cars have the problems of low detection efficiency and poor real-time performance.

SUMMARY OF THE INVENTION

In view of this, the present invention provides a brake detection method for railway freight cars, which solves the problems of low detection efficiency and poor real-time performance of traditional railway freight car brake detection methods by improving data acquisition and data processing methods.

In order to solve the above problems, the technical solution of the present invention is to adopt a method for detecting a brake of a railway freight car, which includes: obtaining parameter information of the railway freight car; and obtaining the positioning information of the RGV car in real time when the RGV car travels in the first direction, Based on the parameter information and the positioning information, the first target detection area of multiple carriages is generated in real time, and the RGV car collects multiple first detection images in the first target detection area; obtains a pre-trained neural network model , input a plurality of the first detection images into the neural network model to generate a first defect detection result; generate a second target detection area when the RGV trolley travels in the second direction based on the first defect detection results, the The RGV trolley collects multiple second detection images in the second target detection area, and inputs the multiple second detection images into the neural network model to generate second defect detection results; based on the first defect detection results and The second defect detection result generates a brake motor defect detection report.

Optionally, acquiring the parameter information of the railway freight car includes: when the railway freight car enters the inspection yard, extracting the vehicle number information of the railway freight car and the vehicle wheelbase information and the vehicle axle corresponding to the vehicle number information based on the vehicle number identification unit. When the railway freight car enters the inspection yard, extract the number of axles of the railway freight car that has entered the inspection yard based on the axle counting sensor; based on the number of axles, the vehicle number information, the vehicle The wheelbase information and the vehicle axle number information constitute the parameter information.

Optionally, obtaining the positioning information of the RGV car in real time includes: obtaining travel distance information based on an encoder of the RGV car; generating information on the number of wheels of the RGV car that has passed the railway freight car based on the wheel sensor of the RGV car, and using it as auxiliary correction information; The positioning information of the RGV car compared to the railway freight car is generated based on the travel distance information, the auxiliary correction information and the initial position information of the RGV car, wherein the initial position of the RGV car is the beginning of the front of the railway freight car .

Optionally, generating the first target detection area in real time based on the parameter information and the positioning information includes: a brake is arranged between the second axle and the third axle of each carriage of the railway freight car. in the case of generating the first target detection for detecting the area between the second axle and the third axle of each carriage of the railway freight car based on the positioning information, the vehicle wheelbase information and the vehicle axle number information area.

Optionally, pre-training the neural network model includes: constructing an initialization network model, wherein the network model includes a semantic segmentation model; acquiring a training data set and a test data set composed of brake image samples marked with multiple categories, wherein , the labeling category includes brakes and multiple types of defects of the brakes; the neural network model is trained and tested based on the training data set and the test data set.

Optionally, inputting a plurality of the first detection images into the neural network model to generate a first defect detection result includes: inputting a plurality of the first detection images of the same carriage into the neural network model one by one, and When the neural network model generates the brake detection frame for the first time, the image recognition of the remaining first detection images in the same car is stopped, and the first detection images of the next car are input into the neural network model one by one , and at the same time, if other types of brake defect detection frames are simultaneously generated when the neural network model generates the brake detection frame for the first time, then the brake detection frame and the brake defect detection frame are generated based on the brake detection frame and the brake defect detection frame. The defect detection result of the carriage; when the neural network model traverses the first detection images of all the carriages, the first defect detection result is generated based on the defect detection results of all the carriages.

Optionally, generating a second target detection area when the RGV trolley travels in the second direction based on the first defect detection result includes: based on the first defect detection result, the brake detection frame of each row of carriages includes: The area information and coordinate information of the detection frame are generated to generate a second target detection area in the first target detection area of each train whose area is smaller than the area of the first target detection area and larger than the area of the detection frame, and the second target detection area includes all The area to which the brake detection frame belongs.

Optionally, inputting a plurality of the second detection images into the neural network model to generate a second defect detection result includes: inputting a plurality of the second detection images of the same carriage into the neural network model one by one, and then When the neural network model generates the brake detection frame for the first time, the image recognition of the remaining second detection images in the same car is stopped, and multiple second detection images of the next car are input into the neural network model one by one , and at the same time, if other types of brake defect detection frames are simultaneously generated when the neural network model generates the brake detection frame for the first time, then the brake detection frame and the brake defect detection frame are generated based on the brake detection frame and the brake defect detection frame. The defect detection result of the carriage; when the neural network model traverses the second detection images of all the carriages, the second defect detection result is generated based on the defect detection results of all the carriages.

The primary improvement of the present invention is the detection method of the brake of the railway freight car. By collecting the parameter information of the railway freight car and the positioning information of the RGV car compared to the railway train, the axle area where the brake is located is accurately located as the first target detection area , which effectively avoids the problem that the RGV trolley needs to continuously collect photos of the bottom of the train in real time during the driving process, resulting in a large number of photos and a huge computing power load on the data processing unit, so that the RGV trolley only needs to intermittently detect the first target detection during the driving process. The area is enough, which reduces the amount of data input to the RGV car data processing unit, and realizes the real-time detection of brake defects of the RGV car. At the same time, since the brake cylinder of each car may be located on the left or right side of the car body, when the RGV car is detected in the first direction, the brake can only be generated based on the axle position of each car and the positioning information of the RGV car. The axle area where it is located is used as the first target detection area, but after generating the first defect detection result, the data processing unit can determine whether the brake cylinder of each car is set on the left or right side of the car body from the brake detection frame information, In this way, the detection area is further reduced, and a more accurate second target detection area is generated, thereby further reducing the amount of data input to the RGV trolley data processing unit, realizing real-time and accurate detection of the brakes of railway freight cars, and solving the problem of traditional railway freight cars. The existing brake detection method has the problems of low detection efficiency and poor real-time performance.

Description of drawings

FIG. 1 is a simplified flow chart of a method for detecting a brake of a railway freight car according to the present invention.

Detailed ways

In order to make those skilled in the art better understand the technical solutions of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

As shown in FIG. 1 , a method for detecting a brake of a railway freight car includes: acquiring parameter information of the railway freight car; when the RGV car is traveling in a first direction, acquiring the positioning information of the RGV car in real time, based on the parameter information and The positioning information generates the first target detection area of multiple carriages in real time, and the RGV car collects a plurality of first detection images in the first target detection area; The first detection image is input to the neural network model to generate a first defect detection result; based on the first defect detection result, a second target detection area is generated when the RGV trolley travels in the second direction, and the RGV trolley collects the first defect detection area. Two second detection images in the target detection area, inputting the plurality of second detection images into the neural network model to generate a second defect detection result; based on the first defect detection result and the second defect detection As a result, a brake defect detection report is generated. Wherein, the first direction may be the direction in which the train head faces the train tail, and the second direction may be the direction in which the train tail faces the train head; since the first detection image and the second detection image are respectively generated by the RGV car in different driving directions Therefore, there is a certain difference between the first defect detection result output by the neural network model and the second defect detection result. At the same time, the reliability of the detection result output by the neural network model is affected by environmental interference, camera parameters and other factors. The manner of generating the brake machine defect detection report based on the first defect detection result and the second defect detection result can be freely selected according to the credibility of the first defect detection result and the second defect detection result output by the neural network model, For example: when the reliability of the first defect detection result and the second defect detection result is high, the brake defect detection can be generated based on the union of the first defect detection result and the second defect detection result report; in the case that the reliability of the first defect detection result and the second defect detection result is low, a brake machine defect detection report may be generated based on the intersection of the first defect detection result and the second defect detection result .

Further, acquiring the parameter information of the railway freight car includes: when the railway freight car enters the inspection yard, extracting the vehicle number information of the railway freight car and the vehicle wheelbase information and the number of vehicle axles corresponding to the vehicle number information based on the vehicle number identification unit. information; when the railway freight car enters the inspection yard, extract the number of axles of the railway freight car that has entered the inspection yard based on the axle counting sensor; based on the number of axles, the vehicle number information, the vehicle axle The distance information and the vehicle axle number information constitute the parameter information. Among them, the vehicle number recognition unit and the axle counting sensor are all mature general-purpose devices in the art, so the present application does not specifically limit their setting methods and models.

Further, obtaining the positioning information of the RGV car in real time includes: obtaining the travel distance information based on the encoder of the RGV car; the wheel sensor based on the RGV car is generated from the number of wheels of the RGV car that has passed the railway freight car, and used as auxiliary correction information; The travel distance information, the auxiliary correction information and the initial position information of the RGV car generate the positioning information of the RGV car compared to the railway freight car, wherein the initial position of the RGV car is the beginning of the front of the railway freight car. Among them, the RGV trolley is an automatic trolley based on track guidance, which is a mature general-purpose device in the field. Therefore, this application does not specifically limit its setting method and model. At the same time, the RGV trolley can be equipped with encoders, wheel sensors, cameras, data Processing unit and other conventional detection devices in the field.

Further, generating the first target detection area in real time based on the parameter information and the positioning information includes: when a brake is arranged between the second axle and the third axle of each carriage of the railway freight car Next, generate the first target detection area for detecting the area between the second axle and the third axle of each carriage of the railway freight car based on the positioning information, the vehicle wheelbase information and the vehicle axle number information . Wherein, in the art, the brake is usually arranged between the second axle and the third axle of each car of the railway freight car, so how to generate the first target detection area will be described in this case. When the brake is arranged between other axles of each car of the railway freight car, the area between the other axles can still be used as the first target detection area.

Further, pre-training the neural network model includes: constructing an initialization network model, wherein the network model includes a semantic segmentation model; obtaining a training data set and a test data set composed of brake image samples containing multi-category labels, wherein, Labeled classes include brakes and brakes of multiple classes of defects; the neural network model is trained and tested based on the training dataset and the test dataset. Among them, it should be noted that the neural network model used in this application is a conventional prior art in the field, and does not involve further improvement of the model architecture, so the type and architecture of the neural network model are not specifically limited. Among them, the type of neural network model can be YOLO-V3, FASTER RCNN, etc.

Further, inputting a plurality of the first detection images into the neural network model to generate a first defect detection result includes: inputting a plurality of the first detection images of the same carriage into the neural network model one by one, and in the When the neural network model generates the brake detection frame for the first time, stop the image recognition of the remaining first detection images in the same car and input multiple first detection images of the next car into the neural network model one by one, At the same time, if other types of brake defect detection frames are simultaneously generated when the neural network model generates the brake detection frame for the first time, the brake defect detection frame is generated based on the brake detection frame and the brake defect detection frame. Defect detection results of the carriages; when the neural network model traverses the first detection images of all the carriages, the first defect detection results are generated based on the defect detection results of all the carriages. Among them, the information of the various types of detection frames output by the neural network model includes (x, y, w, h, i), (x, y) is the upper left corner coordinate of the detection frame, and w is the detection frame. width, h is the height of the detection frame, and i is the confidence of the detection frame.

Further, generating a second target detection area when the RGV trolley travels in the second direction based on the first defect detection result includes: based on the first defect detection result, the brake detection frame of each row of carriages includes: Detect frame area information and coordinate information, and generate a second target detection area in the first target detection area of each train whose area is smaller than the area of the first target detection area and larger than the area of the detection frame, and the second target detection area includes the The area to which the brake detection frame belongs.

Further, inputting a plurality of the second detection images into the neural network model to generate a second defect detection result includes: inputting a plurality of the second detection images of the same carriage into the neural network model one by one, When the neural network model generates the brake detection frame for the first time, the image recognition of the remaining second detection images in the same car is stopped, and multiple second detection images of the next car are input into the neural network model one by one , and at the same time, if other types of brake defect detection frames are simultaneously generated when the neural network model generates the brake detection frame for the first time, then the brake detection frame and the brake defect detection frame are generated based on the brake detection frame and the brake defect detection frame. The defect detection result of the carriage; when the neural network model traverses the second detection images of all the carriages, the second defect detection result is generated based on the defect detection results of all the carriages.

The invention accurately locates the axle area where the brake is located as the first target detection area by collecting the parameter information of the railway freight car and the positioning information of the RGV trolley compared with the railway train, effectively avoiding the need for the RGV trolley to continuously collect the train in real time during the running process. The bottom photo leads to a huge number of photos, which causes a huge computational load of the data processing unit, so that the RGV car only needs to intermittently detect the first target detection area during driving, which reduces the data input to the RGV car data processing unit. It realizes the real-time detection of brake defects of the RGV trolley. At the same time, since the brake cylinder of each car may be located on the left or right side of the car body, when the RGV car is detected in the first direction, the brake can only be generated based on the axle position of each car and the positioning information of the RGV car. The axle area where it is located is used as the first target detection area, but after generating the first defect detection result, the data processing unit can determine whether the brake cylinder of each car is set on the left or right side of the car body from the brake detection frame information, In this way, the detection area is further reduced, and a more accurate second target detection area is generated, thereby further reducing the amount of data input to the RGV trolley data processing unit, realizing real-time and accurate detection of the brakes of railway freight cars, and solving the problem of traditional railway freight cars. The existing brake detection method has the problems of low detection efficiency and poor real-time performance.

The method for detecting a brake of a railway freight car provided by the embodiment of the present invention has been described in detail above. The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments can be referred to each other. As for the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant part can be referred to the description of the method. It should be pointed out that for those skilled in the art, without departing from the principle of the present invention, several improvements and modifications can also be made to the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

Professionals may further realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two, in order to clearly illustrate the possibilities of hardware and software. Interchangeability, the above description has generally described the components and steps of each example in terms of functionality. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Skilled artisans may implement the described functionality using different methods for each particular application, but such implementations should not be considered beyond the scope of the present invention.

The steps of a method or algorithm described in conjunction with the embodiments disclosed herein may be directly implemented in hardware, a software module executed by a processor, or a combination of the two. A software module can be placed in random access memory (RAM), internal memory, read only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other in the technical field. in any other known form of storage medium.

Claims (8)

1. A brake detection method of a railway wagon is characterized by comprising the following steps:

acquiring parameter information of a railway wagon;

when the RGV trolley runs along a first direction, acquiring positioning information of the RGV trolley in real time, generating a first target detection area of a plurality of rows of carriages in real time based on the parameter information and the positioning information, and collecting a plurality of first detection images in the first target detection area by the RGV trolley;

acquiring a pre-trained neural network model, and inputting a plurality of first detection images into the neural network model to generate a first defect detection result;

generating a second target detection area when the RGV trolley runs along a second direction based on the first defect detection result, acquiring a plurality of second detection images in the second target detection area by the RGV trolley, and inputting the plurality of second detection images into the neural network model to generate a second defect detection result;

generating a brake defect detection report based on the first defect detection result and the second defect detection result.

2. The brake detection method of claim 1, wherein obtaining parameter information of a rail wagon comprises:

when a railway wagon drives into a train inspection field, extracting the wagon number information of the railway wagon and vehicle wheelbase information and vehicle axle number information corresponding to the wagon number information based on a wagon number identification unit;

when a railway wagon drives into a train inspection field, extracting the number of axles of the part of the railway wagon, which has driven into the train inspection field, based on an axle counting sensor;

and forming the parameter information based on the number of axles, the number information of the vehicle, the wheel base information of the vehicle and the number information of the axles of the vehicle.

3. The brake detection method of claim 2, wherein obtaining location information of the RGV car in real time comprises:

acquiring running distance information based on an encoder of the RGV trolley;

the wheel sensor based on the RGV trolley generates the number information of the wheels of the RGV trolley passing through the rail wagon as auxiliary correction information;

and generating the positioning information of the RGV compared with the rail wagon based on the running distance information, the auxiliary correction information and the initial position information of the RGV, wherein the initial position of the RGV is the head starting position of the rail wagon.

4. The brake detection method of claim 3, wherein generating a first target detection area in real time based on the parameter information and the positioning information comprises:

in the case where a brake is provided between the second axle and the third axle of each of the cars of the railway wagon,

generating the first target detection area for detecting an area between a second axle and a third axle of each car of the railway wagon based on the positioning information, the vehicle wheel base information, and the vehicle axle number information.

5. The brake detection method of claim 1, wherein pre-training the neural network model comprises:

constructing an initialization network model, wherein the network model comprises a semantic segmentation model;

acquiring a training data set and a testing data set which are formed by brake image samples containing multi-class marks, wherein the mark classes comprise the brakes and multi-class defects of the brakes;

training and testing the neural network model based on the training dataset and the testing dataset.

6. The brake detection method of claim 4, wherein inputting a plurality of the first inspection images into the neural network model to generate a first defect detection result comprises:

inputting a plurality of first detection images of the same compartment into the neural network model one by one, stopping image recognition of the rest of first detection images in the same compartment and inputting a plurality of first detection images of the next compartment into the neural network model one by one when a brake detection frame is generated by the neural network model for the first time, and generating a defect detection result of the compartment based on the brake detection frame and the brake defect detection frame if brake defect detection frames of other categories are generated at the same time when the brake detection frame is generated by the neural network model for the first time;

and when the neural network model finishes traversing the first detection images of all the carriages, generating a first defect detection result based on the defect detection results of all the carriages.

7. The brake detection method of claim 5, wherein generating a second target detection zone when the RGV is traveling in a second direction based on the first defect detection result comprises:

and generating a second target detection area with the area smaller than that of the first target detection area and larger than that of the detection frame in the first target detection area of each train of carriages based on the area information and the coordinate information of the detection frame contained in the brake detection frame of each train of carriages in the first defect detection result, wherein the second target detection area contains the area to which the brake detection frame belongs.

8. The brake detection method of claim 7, wherein inputting a plurality of the second detection images into the neural network model to generate a second defect detection result comprises:

inputting a plurality of second detection images of the same compartment into the neural network model one by one, stopping image recognition of the rest of second detection images in the same compartment and inputting a plurality of second detection images of the next compartment into the neural network model one by one when the neural network model generates a brake detection frame for the first time, and generating a defect detection result of the compartment based on the brake detection frame and the brake defect detection frame if brake defect detection frames of other categories are generated at the same time when the brake detection frame is generated for the first time by the neural network model;

and when the neural network model finishes traversing the second detection images of all the carriages, generating a second defect detection result based on the defect detection results of all the carriages.

Images