Automatic correction method of overhead line pillar foundation database based on GPS positioning

Abstract

The present invention discloses an automatic correction method for a basic database of a contact network pillar based on GPS positioning, comprising the following steps: S1, a collection device is installed on a detection vehicle, and the system automatically completes the collection of GPS longitude and latitude; at the same time, when the train runs to the pillar, the trigger module sends a signal, and the pillar camera takes a picture containing the pillar number identification after receiving the signal, and performs an archiving operation. S2, the basic database correction software is used to automatically analyze the collected data, the pillar number is identified by the pillar identification algorithm, and the system calculates the most correct GPS coordinates in a time-space synchronization manner and associates them with the pillar number. S3, the GPS coordinates processed by the system are intuitively displayed by loading a map; abnormal data points are manually confirmed, and the abnormal data are corrected. S4, after manual confirmation, the system automatically calculates the span length between pillars, and automatically completes the basic database span modification, and the automatic correction of the basic database is completed.

Description

Technical Field

[0001] The invention belongs to the field of GPS positioning technology, and in particular relates to an automatic correction method for a contact network pillar basic database based on GPS positioning. Background Art

[0002] With the rapid development of electrified railways in my country, in order to ensure the safety of power supply equipment and realize accurate maintenance quickly, the real-time positioning accuracy of the contact network detection train must be guaranteed, the detected defect information should be accurate, and it should be able to accurately reflect the actual defect information on the site. The defects existing on the site should not be missed or over-detected, and the detected defects of the line power supply equipment should accurately correspond to the actual position. Therefore, the detection vehicle is highly dependent on the basic database of the contact network pillars. The more accurate the basic database span and the pillar GPS longitude and latitude information are, the more accurate the dynamic detection positioning will be. The existing positioning technology uses the basic database information such as the contact network pillar number, pillar span, line kilometer mark, GPS coordinates along the railway as a reference point to calculate the real-time position of the train. At present, the basic database of railway lines is completely compiled manually based on the previous planning drawings. Due to factors such as construction and personnel, there are deviations between the actual construction and design of the line, which affects the accuracy of the basic database of the line; thus, the positioning of the detection vehicle is biased, which is also a major problem that troubles the detection personnel. On the other hand, the pillar GPS coordinates are all marked by manual following the car, which has problems such as human error, coordinate offset and inaccurate position. As for the above problems of GPS manual marking and basic database, there is no efficient measure to automatically complete GPS marking and automatically correct the basic database of line pillars.

[0003] The standard span between contact network pillars is 50 meters, but due to construction or environmental factors, the span cannot meet the standard requirements; if the basic database defaults to a span of 50 meters per pole, this will cause an error between the basic database and the actual line span, resulting in cumulative errors during the detection process, which will affect the positioning accuracy of the detection vehicle. Summary of the invention

[0004] The purpose of the present invention is to provide a method for automatically correcting a basic database of contact network pillars based on GPS positioning in order to overcome the defects of the prior art.

[0005] The object of the present invention is achieved by the following technical solution: a method for automatically correcting a basic database of a contact network support based on GPS positioning, comprising the following steps:

[0006] S1. The inspection vehicle is equipped with a data acquisition device, which automatically collects GPS coordinate information and captures pillar image data, names the pillar image with the current computer time, adds the current computer time tag to each set of latitude and longitude data packets, and generates a GPS coordinate database;

[0007] S2. Run the basic database automatic correction software offline to read the collected GPS data and pole number image data, process the photographed pole number image through the pole number recognition algorithm, calculate the recognized pole number data with the basic database, identify the current correct pole number position and match it with the collected GPS coordinate information, calculate the most accurate GPS coordinate and then correct the GPS coordinate of the pole number position in the basic database;

[0008] S3. Confirm the corrected basic database. If abnormal data is found, delete the abnormal data, manually find the abnormal pole position data, perform manual secondary matching and correct the basic database;

[0009] S4. After confirming that there is no abnormality in the basic database data, the span is automatically calculated and the basic database is updated.

[0010] Furthermore: the data acquisition device in step S1 includes a trigger module, a capture module, a comprehensive positioning module and a GPS module.

[0011] Furthermore: the method for collecting the GPS coordinate information is: collecting the longitude and latitude information in real time at a frequency of 50 Hz through the GPS module.

[0012] Further: Step S1 is specifically as follows: when the detection train travels to the pillar position, the trigger module recognizes the pillar signal through the real-time recognition algorithm, and sends a capture command to the camera trigger board through the serial port. The pillar capture camera immediately takes a picture of the pillar upon receiving the trigger signal. The captured pillar picture, like the GPS coordinate data, contains a unique time tag accurate to milliseconds and is stored locally in a specified format. All modules are processed by the time synchronization software to maintain relative time synchronization.

[0013] Furthermore: the identification of the pillar number data in step S2 includes pillar number area positioning, pillar number character segmentation and pillar number character classification identification.

[0014] Further: the specific steps of the pillar number area positioning are:

[0015] A1. During the training phase, collect no less than 1,000 images of pillar numbers of the line, set ROI to obtain images of the pillar number area, mark the pillar number area with a rectangular box as a positive sample, and the area outside the rectangular box as the background, which is marked as a negative sample;

[0016] A2. Input the CNN features of the pillar area and the background area into the CNN classifier for training. After the training, the final model is the pillar area and the background.

[0017] A3. In the equipment stage, the input original pillar number image is reduced by 5 times, from 2448×2048 to 489×409;

[0018] A4. According to the size rule of the actual pillar number, set the minimum and maximum size ranges, slide the size window from left to right and from top to bottom on the reduced image, and calculate the CNN features of each sliding window;

[0019] A5. Based on the CNN feature vector of the sliding window and the trained positioning model, CNN is used for classification judgment. The CNN output is 1 for the pillar number area and 0 for the background.

[0020] Further: the specific steps of the pillar number character segmentation are:

[0021] B1. Use the reduced image of the pillar number area as input for character segmentation;

[0022] B2. Use the erosion operation in morphology to filter. The erosion operation is to select a 12*6 structure kernel, calculate the minimum value of all pixels of the kernel as the final output, and filter out the area with complex background of the pillar number;

[0023] B3, select a fixed threshold value T = 80 as the binarization threshold to perform binarization processing on the reduced image;

[0024] B4, extract the bounding rectangles of all white areas in the binary image as candidate characters;

[0025] B5. Filter out candidate character frames based on prior knowledge of the size of pillar number characters, the spacing between pillar number characters, and the number of pillar number characters in the actual scene. The filtered character area is the pillar number character area.

[0026] Furthermore: the pillar number character classification and recognition adopts the VGGNet network structure, and the VGGNet divides the network into 5 segments, each segment connects multiple 3*3 convolutional networks in series, each convolution is followed by a maximum pooling layer, and finally there are 3 fully connected layers and a softmax layer.

[0027] Furthermore: the specific method of information matching in step S2 is: extracting the time tag of the identified pillar number picture through image intelligent recognition technology, and calculating the GPS coordinate point with the smallest error with the current picture time tag through binary search in the GPS coordinate data; thus completing the GPS marking and correction of a single pillar number, and automatically updating the information of the current pillar number in the basic database.

[0028] Further: the calculation method of the span information in step S4 is:

[0029] After the first GPS coordinates and pillar number matching and correction, the system screens out abnormal data points according to the pillar number arrangement rules along the railway, and manually confirms the operation. After the manual confirmation operation is completed, the automatic span correction operation is run to calculate the span and automatically correct the basic database through the GPS coordinates of the front and rear pillars;

[0030] The formula for calculating the span between two GPS coordinates is:

[0031] D=R*arccos(sin(LatA)sin(LatB)+cos(LatA)cos(LatB)*cos(LngA-LngB))

[0032] In the above formula, D is the distance between points A and B, R is the radius of the earth, LngA and LatA are the longitude and latitude of point A, and LngB and LatB are the longitude and latitude of point B.

[0033] The aforementioned main scheme of the present invention and its further options can be freely combined to form multiple schemes, all of which are schemes that can be adopted and claimed for protection by the present invention. After understanding the scheme of the present invention, those skilled in the art can understand that there are multiple combinations based on the prior art and common knowledge, all of which are technical schemes to be protected by the present invention, and they are not exhaustively listed here.

[0034] Beneficial effects of the present invention:

[0035] 1. The present invention does not require a dedicated person to follow the vehicle and automatically record the GPS coordinates, thereby improving work efficiency;

[0036] 2. The present invention does not require manual recognition of the pillar number information of the image, thus reducing the workload;

[0037] 3. The present invention does not require manual entry of pillar span data, thus reducing errors in the basic database;

[0038] 4. The present invention does not require manual correction of the basic database. It improves work efficiency and reduces manual operation errors through intelligent image recognition, automatic matching with GPS, and automatic span calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of the present invention;

[0040] Figure 2 It is a flowchart of pillar number character segmentation in the present invention;

[0041] Figure 3 It is a flowchart of pillar number character classification and recognition in the present invention. DETAILED DESCRIPTION

[0042] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0043] It should be noted that in order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described below. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments.

[0044] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc. indicate the orientation or position relationship that the product of the invention is usually placed in when in use. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0045] In addition, the terms "horizontal", "vertical", "overhanging" and the like do not mean that the components are required to be absolutely horizontal or overhanging, but can be slightly tilted. For example, "horizontal" only means that its direction is more horizontal than "vertical", and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0046] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0047] In addition, the present invention should be pointed out that, in the present invention, unless the specific structure, connection relationship, positional relationship, power source relationship, etc. are specifically written out, the structure, connection relationship, positional relationship, power source relationship, etc. involved in the present invention are all known by those skilled in the art on the basis of the prior art without creative work.

[0048] Embodiment 1:

[0049] refer to Figure 1 As shown, the present invention discloses a method for automatically correcting a basic database of a contact network pillar based on GPS positioning, comprising the following steps:

[0050] S1. The inspection vehicle is equipped with a data acquisition device, which automatically collects GPS coordinate information and captures pillar image data, names the pillar image with the current computer time, adds the current computer time tag to each set of latitude and longitude data packets, and generates a GPS coordinate database;

[0051] It is necessary to install data acquisition devices on the train, including trigger modules, capture modules, integrated positioning modules and GPS modules, as shown in Table 1; at the same time, it is necessary to deploy trigger software, acquisition software, integrated positioning software and time synchronization software on the train.

[0052] Table 1 Data acquisition hardware configuration table

[0053]

[0054] During the inspection process, the GPS module of the inspection vehicle collects longitude and latitude information in real time at a frequency of 50Hz. A time tag with the current computer accuracy to milliseconds is added to each set of longitude and latitude data packets, and the local data is stored in the specified format.

[0055] When the detection train reaches the pillar position, the trigger module recognizes the pillar signal through the real-time recognition algorithm, and sends a capture command to the camera trigger board through the serial port. The pillar capture camera immediately takes a picture of the pillar upon receiving the trigger signal. The pillar picture taken contains a unique time tag accurate to milliseconds, just like the GPS coordinate data, and is stored locally in the specified format. All modules are processed by the time synchronization software to maintain relative time synchronization.

[0056] S2. Run the basic database automatic correction software offline to read the collected GPS data and pole number image data, process the photographed pole number image through the pole number recognition algorithm, calculate the recognized pole number data with the basic database, identify the current correct pole number position and match it with the collected GPS coordinate information, calculate the most accurate GPS coordinate and then correct the GPS coordinate of the pole number position in the basic database;

[0057] Through the pillar number image intelligent recognition algorithm, the pillar number data of the captured image is extracted;

[0058] 1) Pillar number recognition algorithm:

[0059] The pillar number recognition algorithm is divided into three main stages: pillar number area positioning, pillar number character segmentation, and pillar number character classification and recognition, among which:

[0060] 1.1) The steps for locating the pillar number area are as follows:

[0061] A1: First, during the training phase, collect no less than 1,000 images of pillar numbers on the line, set the ROI cut-off to obtain a large area image of the pillar number, and then mark the pillar number area with a rectangular box as a positive sample, and the area outside the rectangular box as the background, which is marked as a negative sample.

[0062] A2: Input the CNN features of the pillar number area and the background area into the CNN classifier for training. After the training, the final model is two categories: the pillar number area and the background.

[0063] A3: During the recognition stage, in order to quickly locate the pillar number area, the input original pillar number image needs to be reduced by 5 times, from 2448×2048 to 489×409.

[0064] A4: According to the size rule of the actual pillar number, set the minimum and maximum size range (50-120), slide the size window from left to right and from top to bottom in the reduced image, and calculate the CNN features of each sliding window.

[0065] A5: Based on the CNN feature vector of the sliding window and the trained positioning model, CNN is used for classification. If CNN outputs 1, it is the pillar area; if the output is 0, it is the background.

[0066] 1.2) If Figure 2 As shown, the steps for pillar number character segmentation are as follows:

[0067] B1: Pillar number area image input: The small image locating the pillar number area is used as the input for character segmentation.

[0068] B2: Morphological filtering: The erosion operation in morphology is used for filtering. The erosion operation is to select a 12*6 structure kernel and calculate the minimum value of all pixels of the kernel as the final output; the area with a more complex background of the pillar number can be filtered out.

[0069] B3: Image binarization processing: Select a fixed threshold T=80 as the binarization threshold for binarization processing.

[0070] B4: Image connectivity analysis: Extract the bounding rectangles of all white areas in the binary image as candidate characters.

[0071] B5: Pillar number character filtering: Candidate character frames are filtered out based on prior knowledge such as the size of the pillar number characters, the spacing between pillar number characters, and the number of pillar number characters in the actual scene. The filtered character area is the pillar number character area.

[0072] 1.3) If Figure 3As shown, the pillar number character classification and recognition are specifically as follows:

[0073] The feature extraction in character recognition uses the VGGNet network structure. VGGNet divides the network into 5 sections, each of which connects multiple 3*3 convolutional networks in series. Each convolution is followed by a maximum pooling layer, and finally three fully connected layers and a softmax layer. VGG uses multiple convolutional layers with smaller convolutional kernels (3x3) instead of one convolutional layer with a larger convolutional kernel, which enhances the network's fitting ability. At the same time, its structure is simple, the number of layers is deeper, and the feature map is wider.

[0074] 2) GPS coordinates match the pillar number

[0075] The time tag of the identified pillar number image is extracted through image intelligent recognition technology, and the GPS coordinate point with the smallest error with the current image time tag is calculated through binary search in the GPS coordinate data; the GPS labeling and correction of a single pillar number is completed, and the information of the current pillar number in the basic database is automatically updated. Similarly, the correction operation of the pillar basic data of all routes is automatically completed.

[0076] S3. Confirm the corrected database. If abnormal data is found, delete the abnormal data, manually find the abnormal pole position data, perform manual secondary matching and correct the database;

[0077] S4. After confirming that there is no abnormality in the database data, the span is automatically calculated and the database is updated.

[0078] 1) Calculation of span between pillar numbers

[0079] After the first GPS coordinates and pillar number matching and correction, the system screens out abnormal data points according to the pillar number arrangement rules along the railway, and manually confirms the operation. After the manual confirmation operation is completed, the automatic span correction operation is run, and the span is calculated and the basic database is automatically corrected through the GPS coordinates of the front and rear pillars.

[0080] The formula for calculating the span between two GPS coordinates is:

[0081] D=R*arccos(sin(LatA)sin(LatB)+cos(LatA)cos(LatB)*cos(LngA-LngB))

[0082] In the above formula, D is the distance between points A and B, R is the radius of the earth, LngA and LatA are the longitude and latitude of point A, and LngB and LatB are the longitude and latitude of point B.

[0083] Through the above steps, the automatic matching of the pillar number and GPS coordinates in the contact network pillar basic database and the automatic correction of the span between two pillars are completed.

[0084] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for automatically correcting the basic database of overhead line pillars based on GPS positioning, characterized in that: The following steps are involved: S1. The inspection vehicle is equipped with a data acquisition device, which automatically collects GPS coordinate information and captures pillar image data, names the pillar image with the current computer time, adds the current computer time tag to each set of latitude and longitude data packets, and generates a GPS coordinate database; S2, run the basic database automatic correction software, read the collected GPS data and pole number image data, process the photographed pole number image through the pole number recognition algorithm, calculate the recognized pole number data with the basic database, identify the current correct pole number position and then perform time and space synchronization matching with the collected GPS coordinate information, calculate the most accurate GPS coordinates and then correct the GPS coordinates of the pole number position in the basic database; S3. Manually confirm the corrected basic database. If abnormal data is found, delete the abnormal data, manually find the abnormal pole position data, perform manual secondary matching and correct the basic database; S4. After confirming that there is no abnormality in the basic database data, the span is automatically calculated and the basic database is automatically updated; The data acquisition device in step S1 includes a trigger module, a capture module, a comprehensive positioning module and a GPS module; The step S1 is specifically as follows: when the detection train travels to the pillar position, the trigger module recognizes the pillar signal through the real-time recognition algorithm, and sends a capture command to the camera trigger board through the serial port. The pillar capture camera immediately takes a picture of the pillar upon receiving the trigger signal. The captured pillar picture and GPS coordinate data contain a unique time tag accurate to milliseconds, and are stored locally in a specified format; all modules are processed by the time synchronization software to maintain relative time synchronization; The identification of the pillar number data in step S2 includes pillar number area positioning, pillar number character segmentation and pillar number character classification identification; The pillar number character classification and recognition adopts the VGGNet network structure. The VGGNet divides the network into 5 sections. Each section connects multiple 3*3 convolutional networks in series. Each convolution is followed by a maximum pooling layer, and finally 3 fully connected layers and a softmax layer. The specific method of the information time-space synchronization matching in step S2 is: extract the time tag of the pillar number picture identified by the image intelligent recognition technology, and calculate the GPS coordinate point with the smallest error with the current picture time tag in the GPS coordinate data by binary search; thus completing the GPS marking and correction of the single pillar number, and automatically updating the information of the current pillar number in the basic database; The calculation method of the span information in step S4 is: After the first GPS coordinates and pillar number matching and correction, the system screens out abnormal data points according to the pillar number arrangement rules along the railway, and manually confirms the operation. After the manual confirmation operation is completed, the automatic span correction operation is run to calculate the span and automatically correct the basic database through the GPS coordinates of the front and rear pillars; The formula for calculating the span between two GPS coordinates is: D=R*arccos(sin(LatA)sin(LatB)+cos(LatA)cos(LatB)*cos(LngA-Lng B)) In the above formula, D is the distance between points A and B, R is the radius of the earth, LngA and LatA are the longitude and latitude of point A, and LngB and LatB are the longitude and latitude of point B.

2. The method for automatically correcting the basic database of overhead line pillars based on GPS positioning according to claim 1 is characterized in that: The method for collecting GPS coordinate information is: using a GPS module to collect longitude and latitude information in real time at a frequency of 50 Hz.

3. The method for automatically correcting the basic database of overhead line pillars based on GPS positioning according to claim 1 is characterized in that: The specific steps of the pillar number area positioning are: A1. During the training phase, collect no less than 1,000 images of pillar numbers of the line, set ROI to obtain images of the pillar number area, mark the pillar number area with a rectangular box as a positive sample, and the area outside the rectangular box as the background, which is marked as a negative sample; A2. Input the CNN features of the pillar area and the background area into the CNN classifier for training. After the training, the final model is the pillar area and the background. A3. In the equipment stage, the input original pillar number image is reduced by 5 times, from 2448×2048 to 489×409; A4. According to the size rule of the actual pillar number, set the minimum and maximum size ranges, slide the size window from left to right and from top to bottom on the reduced image, and calculate the CNN features of each sliding window; A5. Based on the CNN feature vector of the sliding window and the trained positioning model, CNN is used for classification judgment. The CNN output is 1 for the pillar number area and 0 for the background.

4. The method for automatically correcting the basic database of overhead line pillars based on GPS positioning according to claim 1 is characterized in that: The specific steps of the pillar number character segmentation are: B1. Use the reduced image of the pillar number area as input for character segmentation; B2. Use the erosion operation in morphology to filter. The erosion operation is to select a 12*6 structure kernel, calculate the minimum value of all pixels of the kernel as the final output, and filter out the area with complex background of the pillar number; B3, select a fixed threshold value T = 80 as the binarization threshold to perform binarization processing on the reduced image; B4, extract the bounding rectangles of all white areas in the binary image as candidate characters; B5. Filter out candidate character frames based on prior knowledge of the size of pillar number characters, the spacing between pillar number characters, and the number of pillar number characters in the actual scene. The filtered character area is the pillar number character area.

Images