Method and device for correcting machine account data

Abstract

The invention discloses a machine account data correction method and device. The method comprises the following steps: obtaining pull-out value data of a railway overhead line system; inputting the pull-out value data into a pre-trained anchor section identification model, and outputting mileage data of the anchor section joint and the type of the anchor section joint; the anchor section recognition model is obtained by training a machine learning model according to historical pull-out value data of the railway overhead line system, mileage data of anchor section joints corresponding to the historical pull-out value data and types of the anchor section joints corresponding to the historical pull-out value data; according to the output type of the anchor section joint and the mileage data of the anchor section joint, determining the mileage data of the central column; correcting the machine account data according to the mileage data of the central column; the standing book data comprises the contact network line information, the strong association between the contact network waveform and the standing book data can be realized, the standing book data is corrected, and the standing book data accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of high-speed railway maintenance engineering, and in particular to a method and device for correcting ledger data. Background Art

[0002] This section is intended to provide a background or context for embodiments of the present invention. No description herein is admitted to be prior art by virtue of its inclusion in this section.

[0003] At present, with the continuous increase in the operating mileage of high-speed railways, the demand for inspection of line infrastructure has become increasingly urgent. The contact network is an important equipment for providing power guarantee for operating vehicles, and the ledger is an important basic data for recording contact network line information. Through the association between the two, on-site operators can be quickly guided to repair the detected fault points.

[0004] The existing method of associating the contact network and the ledger only uses the mileage information in the waveform data to manually compare and search with the mileage information in the ledger. However, due to the statistical errors in the construction mileage when the ledger is established and the cumulative mileage errors in the detection data, the matching results often have large deviations. Taking a 50m span as an example, when the error between the detected pillar mileage and the pillar mileage in the ledger exceeds 25m, there will be uncertainty in the matching. Summary of the invention

[0005] The embodiment of the present invention provides a method for correcting ledger data, which can achieve a strong association between the overhead line waveform and the ledger data, correct the ledger data, and improve the accuracy of the ledger data. The method includes:

[0006] Acquire the pull-out value data of the railway contact network; the pull-out value data is the pull-out value waveform data of the railway contact network corresponding to the railway mileage data; the pull-out value data includes the pull-out value data having the pull-out value characteristics of the anchor section joint;

[0007] Input the pull-out value data into a pre-trained anchor segment recognition model, and output the mileage data of the anchor segment joint and the type of the anchor segment joint; the anchor segment recognition model is obtained by training a machine learning model based on the historical pull-out value data of the railway contact network, the mileage data of the anchor segment joint corresponding to the historical pull-out value data, and the type of the anchor segment joint corresponding to the historical pull-out value data;

[0008] Determine the mileage data of the center column according to the output anchor segment joint type and the mileage data of the anchor segment joint;

[0009] The ledger data is corrected according to the mileage data of the center column; the ledger data includes the contact network line information.

[0010] The embodiment of the present invention provides a ledger data correction device, which can realize a strong association between the overhead line waveform and the ledger data, correct the ledger data, and improve the accuracy of the ledger data. The device includes:

[0011] An acquisition module is used to acquire the pull-out value data of the railway contact network; the pull-out value data is the pull-out value waveform data of the railway contact network corresponding to the railway mileage data; the pull-out value data includes the pull-out value data having the pull-out value characteristics of the anchor section joint;

[0012] An output module is used to input the pull-out value data into a pre-trained anchor segment recognition model, and output the mileage data of the anchor segment joint and the type of the anchor segment joint; the anchor segment recognition model is obtained by training a machine learning model with the historical pull-out value data of the railway contact network, the mileage data of the anchor segment joint corresponding to the historical pull-out value data, and the type of the anchor segment joint corresponding to the historical pull-out value data;

[0013] A determination module, used to determine the mileage data of the center column according to the output type of the anchor segment joint and the mileage data of the anchor segment joint;

[0014] The correction module is used to correct the ledger data according to the mileage data of the center column; the ledger data includes the contact network line information.

[0015] An embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned ledger data correction method when executing the computer program.

[0016] An embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned ledger data correction method is implemented.

[0017] An embodiment of the present invention also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the above-mentioned ledger data correction method is implemented.

[0018] Compared with the scheme of the prior art which uses the mileage information in the waveform data and manually compares and searches the mileage information in the ledger, the embodiment of the present invention obtains the pull-out value data of the railway contact network; the pull-out value data is the pull-out value waveform data of the railway contact network corresponding to the railway mileage data; the pull-out value data includes the pull-out value data with the pull-out value characteristics of the anchor segment joint; the pull-out value data is input into a pre-trained anchor segment recognition model, and the mileage data of the anchor segment joint and the type of the anchor segment joint are output; the anchor segment recognition model is obtained by training a machine learning model with the historical pull-out value data of the railway contact network, the mileage data of the anchor segment joint corresponding to the historical pull-out value data, and the type of the anchor segment joint corresponding to the historical pull-out value data; the mileage data of the center column is determined according to the output type of the anchor segment joint and the mileage data of the anchor segment joint; the ledger data is corrected according to the mileage data of the center column; the ledger data includes the contact network line information, which can realize the strong correlation between the contact network waveform and the ledger data, correct the ledger data, and improve the accuracy of the ledger data. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0020] Figure 1 Flow chart of the method for correcting ledger data in an embodiment of the present invention;

[0021] Figure 2 This is a schematic diagram of ledger data in an embodiment of the present invention;

[0022] Figure 3 This is a schematic diagram of waveform data of the overhead contact network 1c in an embodiment of the present invention;

[0023] Figure 4 Schematic diagram of the contact network anchor section joint in an embodiment of the present invention;

[0024] Figure 5 Schematic diagram of the pull-out value waveform at the joint of the anchor section in an embodiment of the present invention;

[0025] Figure 6 is a flow chart of an anchor segment recognition model training method according to an embodiment of the present invention;

[0026] Figure 7 It is a flowchart of a specific example of the method for correcting ledger data in an embodiment of the present invention;

[0027] Figure 8This is a schematic diagram of ledger data in an embodiment of the present invention;

[0028] Fig. 9 A schematic diagram of a specific example of ledger data in an embodiment of the present invention;

[0029] Fig.10 A schematic diagram of a specific example of ledger data in an embodiment of the present invention;

[0030] Fig.11 A schematic diagram of a closed-loop matching logic in an embodiment of the present invention;

[0031] Fig.12 A schematic diagram of pole number positioning in an embodiment of the present invention;

[0032] Fig.13 Schematic diagram of a ledger data correction device in an embodiment of the present invention. DETAILED DESCRIPTION

[0033] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0034] In order to achieve a strong association between the contact network waveform and the ledger data, the ledger data is corrected to improve the accuracy of the ledger data. The embodiment of the present invention proposes a ledger data correction method and device, which uses neural network technology to train and identify the anchor segment joint feature segments in the waveform data, and improves the existing ledger according to the ledger information specification, and finally associates the waveform data with the ledger information based on the matching logic.

[0035] Figure 1 Flow chart of the method for correcting ledger data in an embodiment of the present invention. Figure 1 As shown, the method includes:

[0036] Step 101, obtaining the pull-out value data of the railway contact network; the pull-out value data is the pull-out value waveform data of the railway contact network corresponding to the railway mileage data; the pull-out value data includes the pull-out value data having the pull-out value characteristics of the anchor section joint;

[0037] Step 102, input the pull-out value data into a pre-trained anchor segment recognition model, and output the mileage data of the anchor segment joint and the type of the anchor segment joint; the anchor segment recognition model is obtained by training a machine learning model with the historical pull-out value data of the railway contact network, the mileage data of the anchor segment joint corresponding to the historical pull-out value data, and the type of the anchor segment joint corresponding to the historical pull-out value data;

[0038] Step 103, determining the mileage data of the center column according to the output anchor segment joint type and the mileage data of the anchor segment joint;

[0039] Step 104, correct the ledger data according to the mileage data of the center column; the ledger data includes overhead line information.

[0040] First, the inventory data and contact network waveform data of this article are introduced.

[0041] The ledger is an important basic data for recording the information of overhead line, including line name data, line name data, bureau name data, station area name data, pillar number data, kilometer mark data, pillar type data, structure mark data, line number data, etc. Figure 2 shown.

[0042] The overhead line is an important device that provides power guarantee for running vehicles. The 1C waveform is obtained through the pantograph integrated detection device. Electric locomotives and EMUs obtain power from the overhead line through the pantograph. The operating status of the pantograph is presented in the form of an "electrocardiogram" through the sensor installed on the integrated detection train, such as Figure 3 As shown, the embodiment of the present invention applies the pulled value data in the waveform data.

[0043] The line segment of the overhead contact network for mechanical segmentation is called an anchor segment, and the connecting section of adjacent anchor segments is called an anchor segment joint. The anchor segment joint that only plays the role of mechanical segmentation is called a non-insulated anchor segment joint. The anchor segment joint that plays not only the role of mechanical segmentation but also the role of electrical segmentation is called an insulated anchor segment joint. The anchor segment joint with a neutral embedded section that plays both the role of mechanical segmentation and the function of electrical segmentation is called an electrical segmentation anchor segment joint. Figure 4 shown.

[0044] The anchor joint in the overhead line will appear once every 1 km, which can realize high-frequency data correlation correction. The pull-out value waveform at the anchor joint is as follows: Figure 5 As shown, its waveform structure is special, and its central column can be intelligently identified through neural network technology. Figure 5 In the test, the point with the kilometer mark of 1600.180km was tested at a speed of 299km / h, a height of 5341mm, and a pull-out value of 272mm. The number of sparks was 0, the height of the contact wire was the height of the contact network, and the number of sparks was the number of times the contact network arced. The mileage information of the center column detected was extracted and identified, and the mileage information and pole number information of the center column at the anchor section joint were compared with the ledger records, and the center column's own mileage information was associated with the ledger information. The center column number in the ledger was added to the waveform center column pole position information through matching logic, and the anchor section and the pillar number and mileage between the anchor sections were supplemented and corrected according to the ledger and waveform pole position information, so as to achieve a strong association between the waveform and the ledger data.

[0045] In one embodiment, before the pull-out value data is input into a pre-trained anchor segment identification model, it may also include: obtaining historical pull-out value data of the railway contact network, mileage data of the anchor segment joint corresponding to the historical pull-out value data, and the type of the anchor segment joint corresponding to the historical pull-out value data; preprocessing the mileage data of the anchor segment joint corresponding to the historical pull-out value data and the type of the anchor segment joint corresponding to the historical pull-out value data; the preprocessing includes sampling, scaling, and windowing or any combination thereof; using the historical pull-out value data of the railway contact network, the preprocessed mileage data of the anchor segment joint corresponding to the historical pull-out value data, and the preprocessed type of the anchor segment joint corresponding to the historical pull-out value data as sample data to construct a training set and a validation set; training a pre-selected machine learning model according to the training set to obtain an anchor segment identification model; using the validation set to evaluate the trained anchor segment identification model to generate an evaluation result; and adjusting the parameters and model structure of the anchor segment identification model according to the evaluation result.

[0046] First, in order to improve the accuracy of anchor segment joint recognition, a target detection deep neural network or machine learning model training can be performed based on the anchor segment joint data in the contact network 1C pull-out data waveform. The embodiment of the present invention provides an anchor segment recognition model training method, Figure 6 FIG. 1 is a flow chart of the anchor segment recognition model training method according to an embodiment of the present invention. Figure 6 As shown, the anchor segment recognition model training method may include:

[0047] 1. Data collection and annotation: Collect 1C pull-out value data, find the anchor joint position manually, and annotate the mileage section and anchor joint type of the anchor joint in the annotation file.

[0048] 2. Data preprocessing: Map the data to the two-dimensional space xy, map the mileage data to the x dimension, and map the pull-out value data to the y dimension. Preprocess the labeled data, including sampling, scaling, windowing, etc., to meet the input requirements of the neural network.

[0049] 3. Model training: Select a suitable machine learning model or target detection neural network model, such as the YOLO series model, on the preprocessed data set and train it. The model output is the position x1, x2, y1, y2 of the anchor segment joint in two-dimensional space, as well as the type and confidence of the anchor segment joint. During the training process, it is necessary to define an appropriate loss function, usually including position loss and category loss, to optimize the network parameters.

[0050] 4. Output post-processing: Map the y-dimensional data output by the model back to the mileage area, that is, the mileage where the anchor segment joint is located.

[0051] 5. Model evaluation: Use the validation set to evaluate the trained object detection model, and measure the performance of the model by calculating indicators such as precision, recall, and F1 score.

[0052] 6. Model tuning: Based on the evaluation results, the model can be tuned, such as adjusting hyperparameters, increasing the amount of data, adjusting the model structure, etc., to achieve better performance.

[0053] In one embodiment, after completing the model training, we can enter the model application phase. The goal is to apply the trained target detection neural network to the actual scenario of contact network anchor segment joint recognition to perform anchor segment joint recognition. Figure 7 As shown, this process involves the following steps:

[0054] 1. Model deployment: The trained model can be deployed to the contact network 1C data processor.

[0055] 2. Data reading: Input the 1C pull-out value data into the anchor segment recognition model in segments.

[0056] 3. Data preprocessing: Map the data to the two-dimensional space xy, map the mileage data to the x dimension, and map the pull-out value data to the y dimension. Preprocess the labeled data, including sampling, scaling, windowing, and other operations, to meet the input requirements of the machine learning model or neural network model.

[0057] 4. Model reasoning: Input the preprocessed data into the anchor segment joint recognition model, and output the result after model reasoning. If the input data does not contain the anchor segment joint, the model has no output; when the input data contains the anchor segment joint, the model will output the coordinate values ​​x1, x2, y1, y2 of the anchor segment joint in the two-dimensional data space after preprocessing, as well as the type and confidence of the anchor segment joint.

[0058] 5. Post-processing of results: Map the y-dimensional data output by the model back to the mileage area to obtain the mileage section where the anchor segment joint is located.

[0059] 6. Determination of the center column of the anchor section joint: Determine the position of the center column according to the type and section of the output anchor section joint. For example, there are two center columns in the four-span anchor section joint, which are located at the beginning and end of the mileage section; there is one center column in the five-span anchor section joint, which is located in the center of the mileage section.

[0060] 7. Ledger verification: Verify the identified mileage of the anchor section joint center column with the ledger, and find the anchor section joint center column in the ledger corresponding to the test data within the 200-meter error range.

[0061] In one embodiment, before correcting the ledger data according to the mileage data of the center column, it can also include: determining whether the ledger data is complete, and if incomplete, supplementing the ledger data; filtering the ledger data according to the station area name in the ledger data; arranging the filtered ledger data according to the pillar number in the filtered ledger data; determining whether the mileage data corresponding to the pillar number is abnormal, and if so, correcting the abnormal data; determining whether the pillar type in the arranged ledger data is abnormal according to the pillar type sequence corresponding to the preset anchor segment joint type, and if so, correcting the abnormal data.

[0062] First, you need to confirm that the ledger data is complete, for example, Figure 8 As shown, it is necessary to confirm that the line name, line name, bureau name, station area name, pillar number, kilometer mark, pillar type, structure mark and other information are complete.

[0063] Secondly, it is necessary to confirm whether the correspondence between the pole number and the mileage information is accurate: filter the information in the ledger by the station area name, and after filtering, arrange them in ascending order by the pillar number in each station area. At this time, observe whether there is any abnormality in the mileage information corresponding to the pillar number, such as Fig. 9 The mileage at the kilometer mark corresponding to the pillar numbered 201 in the middle box is obviously abnormal, and it is the mileage of the center column of the anchor section joint, which should be corrected.

[0064] It is also necessary to confirm that the description of the pillar type is complete and unified: In general, the pillar types of the five-span joints are in the following order: anchor pillar, conversion pillar, center pillar, center pillar, conversion pillar, anchor pillar; the pillar types of the four-span joints are in the following order: anchor pillar, conversion pillar, center pillar, conversion pillar, anchor pillar. Some 12-span and 13-span anchor joints are actually composed of two 5-span joints, and the center pillar should also be indicated. Fig.10 The "inner conversion column" in the box corresponding to the joint of the non-insulated anchor section in the middle five spans should be changed to "center column" in sequence.

[0065] Finally, it is also necessary to confirm that the structural landmark description is complete and accurate: for the anchor segment joints, the structural landmark description should be accurate and cannot be missing.

[0066] In one embodiment, the pull value data may also include pole position information.

[0067] In one embodiment, correcting the ledger data according to the mileage data of the center column may include: correcting the center column information in the ledger data according to the mileage data of the center column; determining the number of pole positions between adjacent anchor segment joints in the pole position information; comparing the number of pole positions between adjacent anchor segment joints in the pole position information with the number of pole positions between adjacent anchor segment joints in the ledger data; if the number of pole positions between adjacent anchor segment joints in the pole position information is inconsistent with the number of pole positions between adjacent anchor segment joints in the ledger data, checking the pole position information according to the pull-out value data; if the check is incorrect, modifying the pole position information according to the ledger data; modifying the ledger data according to the pole position information; if the check is correct, modifying the ledger data according to the pole position information.

[0068] After completing the characteristic waveform recognition in the waveform data and correcting the ledger data information, Fig.11 The closed-loop matching logic diagram shown associates the waveform with the ledger data.

[0069] The following is a schematic diagram of the closed-loop matching logic according to an embodiment of the present invention. Fig.11 Combined with the pole number positioning diagram Fig.12 To narrate.

[0070] First, the mileage of the anchor segment is obtained and located. If the input data of the anchor segment recognition model does not contain the pull-out value data with the pull-out value characteristics of the anchor segment joint, the model has no output, indicating that the original pole position data in the waveform is missing and the center column mileage cannot be obtained. This anchor segment information is invalidated and the next anchor segment is searched. If the input data of the anchor segment recognition model contains the pull-out value data with the pull-out value characteristics of the anchor segment joint, the mileage data of the center column is located with the help of the original pole position data, the ledger information is queried based on the mileage data of the center column, and the mileage data of the center column is assigned to the center column information in the ledger data. For example, Fig.12 In the process, the anchor segment recognition model is used to identify the type of anchor segment joint and the mileage data of the anchor segment joint, and then the mileage data 1633.856 and 1633.901 of the center column are determined. The center column numbers 1297 and 1299 with similar mileage data are queried in the ledger, and the mileage data 1633.856 and 1633.901 of the center column are used to correct the kilometer marks 1633.773 and 1633.818 corresponding to 1297 and 1299 in the ledger respectively. If the ledger information is missing, the ledger is repaired according to the mileage data of the center column, and then the mileage data of the center column is assigned to the center column information in the ledger data.

[0071] exist Fig.11 After the mileage correction of one anchor segment is completed, the mileage of the adjacent anchor segment is obtained and located, which is similar to the above steps and will not be repeated here. Fig.12In the process, determine the mileage data 1634.634 and 1634.683 of the next center column, query the center column numbers 1331 and 1333 with similar mileage data in the ledger, and use the mileage data 1634.634 and 1634.683 of the center column to correct the kilometer marks 1634.549 and 1634.599 corresponding to 1331 and 1333 in the ledger respectively. If the ledger information is missing, repair the ledger according to the mileage data of the center column, and then assign the mileage data of the center column to the center column information in the ledger data.

[0072] After correcting the mileage information of all center columns in the ledger data, the pillar information between adjacent anchor segments can also be corrected. The number of pole positions between adjacent anchor segment joints in the pole position information can be determined; the number of pole positions between adjacent anchor segment joints in the pole position information is compared with the number of pole positions between adjacent anchor segment joints in the ledger data. If the number of pole positions between adjacent anchor segment joints in the pole position information is inconsistent with the number of pole positions between adjacent anchor segment joints in the ledger data, the pole position information is checked according to the pull-out value data. If the check is incorrect, the pole position information is modified according to the ledger data, and the ledger data is modified according to the pole position information. If the check is correct, the ledger data is modified according to the pole position information. In this embodiment, the step of correcting the pillar information between adjacent anchor segments can also be interspersed with the step of correcting the mileage information of the center column in the ledger data.

[0073] In summary, compared with the prior art scheme of manually comparing and searching the mileage information in the waveform data with the mileage information in the ledger, the embodiment of the present invention obtains the pull-out value data of the railway contact network; the pull-out value data is the pull-out value waveform data of the railway contact network corresponding to the railway mileage data; the pull-out value data includes the pull-out value data with the pull-out value characteristics of the anchor segment joint; the pull-out value data is input into a pre-trained anchor segment recognition model, and the mileage data of the anchor segment joint and the type of the anchor segment joint are output; the anchor segment recognition model is obtained by training a machine learning model with the historical pull-out value data of the railway contact network, the mileage data of the anchor segment joint corresponding to the historical pull-out value data, and the type of the anchor segment joint corresponding to the historical pull-out value data; the mileage data of the center column is determined according to the output type of the anchor segment joint and the mileage data of the anchor segment joint; the ledger data is corrected according to the mileage data of the center column; the ledger data includes the contact network line information, which can realize the strong correlation between the contact network waveform and the ledger data, correct the ledger data, and improve the accuracy of the ledger data. The embodiments of the present invention provide a waveform and ledger information data matching method, a characteristic waveform training and identification method, a ledger data standardization method, and a closed-loop matching logic for waveform and ledger data. Neural network training and identification are performed on waveform features, and ledger information is standardized. Finally, closed-loop matching logic is applied to complete the association between the two types of data.

[0074] The embodiment of the present invention further proposes a ledger data correction device, the principle of which is similar to the ledger data correction method and will not be repeated here.

[0075] Fig.13 Schematic diagram of a ledger data correction device in an embodiment of the present invention, Fig.13 As shown, the ledger data correction device may include:

[0076] The acquisition module 1301 is used to acquire the pull-out value data of the railway contact network; the pull-out value data is the pull-out value waveform data of the railway contact network corresponding to the railway mileage data; the pull-out value data includes the pull-out value data having the pull-out value characteristics of the anchor section joint;

[0077] The output module 1302 is used to input the pull-out value data into a pre-trained anchor segment identification model, and output the mileage data of the anchor segment joint and the type of the anchor segment joint; the anchor segment identification model is obtained by training a machine learning model with the historical pull-out value data of the railway contact network, the mileage data of the anchor segment joint corresponding to the historical pull-out value data, and the type of the anchor segment joint corresponding to the historical pull-out value data;

[0078] A determination module 1303 is used to determine the mileage data of the center column according to the output type of the anchor segment joint and the mileage data of the anchor segment joint;

[0079] The correction module 1304 is used to correct the ledger data according to the mileage data of the center column; the ledger data includes the contact network line information.

[0080] In one embodiment, the pull value data also includes pole position information.

[0081] In one embodiment, the correction module 1304 is specifically used to:

[0082] Correct the center column information in the ledger data according to the mileage data of the center column;

[0083] Determine the number of pole positions between adjacent anchor segment joints in the pole position information;

[0084] Compare the number of pole positions between adjacent anchor segment joints in the pole position information with the number of pole positions between adjacent anchor segment joints in the ledger data. If the number of pole positions between adjacent anchor segment joints in the pole position information is inconsistent with the number of pole positions between adjacent anchor segment joints in the ledger data, check the pole position information according to the pull-out value data. If the check is incorrect, modify the pole position information according to the ledger data. Modify the ledger data according to the pole position information. If the check is correct, modify the ledger data according to the pole position information.

[0085] In one embodiment, the ledger data correction device further includes: a training module for

[0086] Obtaining historical pull-out value data of the railway contact network, mileage data of the anchor section joint corresponding to the historical pull-out value data, and types of the anchor section joint corresponding to the historical pull-out value data;

[0087] Preprocessing the mileage data of the anchor segment joint corresponding to the historical pull-out value data and the type of the anchor segment joint corresponding to the historical pull-out value data; the preprocessing includes one or any combination of sampling, scaling, and windowing;

[0088] The historical pull-out value data of the railway contact network, the pre-processed mileage data of the anchor section joints corresponding to the historical pull-out value data, and the pre-processed types of the anchor section joints corresponding to the historical pull-out value data are used as sample data to construct a training set and a validation set;

[0089] The pre-selected machine learning model is trained according to the training set to obtain an anchor segment recognition model;

[0090] Use the validation set to evaluate the trained anchor segment recognition model and generate evaluation results;

[0091] According to the evaluation results, the parameters and model structure of the anchor segment identification model are adjusted.

[0092] In one embodiment, the ledger data correction device further includes: a ledger data preprocessing module for

[0093] Determine whether the ledger data is complete. If incomplete, supplement the ledger data;

[0094] Filter the ledger data according to the station area name in the ledger data;

[0095] Arrange the filtered ledger data according to the pillar number in the filtered ledger data;

[0096] Determine whether the mileage data corresponding to the pillar number is abnormal. If abnormal, correct the abnormal data;

[0097] According to the preset sequence of pillar types corresponding to the anchor segment joint types, determine whether the pillar types in the arranged ledger data are abnormal. If abnormal, correct the abnormal data.

[0098] In summary, the device proposed in the embodiment of the present invention is compared with the solution in the prior art that uses the mileage information in the waveform data to manually compare and search with the mileage information in the ledger. The device is used to obtain the pull-out value data of the railway contact network through an acquisition module; the pull-out value data is the pull-out value waveform data of the railway contact network corresponding to the railway mileage data; the pull-out value data includes the pull-out value data with the pull-out value characteristics of the anchor segment joint; the output module is used to input the pull-out value data into a pre-trained anchor segment recognition model, and output the mileage data of the anchor segment joint and the type of the anchor segment joint; the anchor segment recognition model The model is obtained by training the machine learning model with the historical pull-out value data of the railway contact network, the mileage data of the anchor section joint corresponding to the historical pull-out value data, and the type of the anchor section joint corresponding to the historical pull-out value data; the determination module is used to determine the mileage data of the center column according to the output anchor section joint type and the mileage data of the anchor section joint; the correction module is used to correct the ledger data according to the mileage data of the center column; the ledger data includes the contact network line information, which can realize the strong correlation between the contact network waveform and the ledger data, correct the ledger data, and improve the accuracy of the ledger data.

[0099] The beneficial effects of the present invention are: intelligent identification of the center column is carried out through neural network technology, and the existing ledger is improved according to the ledger information specification, the ledger center column number is added to the waveform center column pole position information through matching logic, and the anchor section and the pillar number and mileage between the anchor sections are completed and corrected according to the ledger and waveform pole position information, so as to achieve a strong correlation between the waveform and the ledger data, correct the ledger data, and improve the accuracy of the ledger data.

[0100] An embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned ledger data correction method when executing the computer program.

[0101] An embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned ledger data correction method is implemented.

[0102] An embodiment of the present invention also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the above-mentioned ledger data correction method is implemented.

[0103] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0104] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0105] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0106] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0107] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for correcting ledger data, characterized in that: include: Obtain the pull-out value data of the railway contact network; The pull-out value data is the pull-out value waveform data of the railway contact network corresponding to the railway mileage data; The pull-out value data includes pull-out value data having pull-out value characteristics of an anchor segment joint; Input the pull-out value data into a pre-trained anchor segment recognition model, and output the mileage data of the anchor segment joint and the type of the anchor segment joint; the anchor segment recognition model is obtained by training a machine learning model based on the historical pull-out value data of the railway contact network, the mileage data of the anchor segment joint corresponding to the historical pull-out value data, and the type of the anchor segment joint corresponding to the historical pull-out value data; Determine the mileage data of the center column according to the output anchor segment joint type and the mileage data of the anchor segment joint; The ledger data is corrected according to the mileage data of the center column; the ledger data includes the contact network line information.

2. The method according to claim 1, characterized in that The pull-out value data also includes pole position information.

3. The method according to claim 2, characterized in that Correct the ledger data according to the mileage data of the center column, including: Correct the center column information in the ledger data according to the mileage data of the center column; Determine the number of pole positions between adjacent anchor segment joints in the pole position information; Compare the number of pole positions between adjacent anchor segment joints in the pole position information with the number of pole positions between adjacent anchor segment joints in the ledger data. If the number of pole positions between adjacent anchor segment joints in the pole position information is inconsistent with the number of pole positions between adjacent anchor segment joints in the ledger data, check the pole position information according to the pull-out value data. If the check is incorrect, modify the pole position information according to the ledger data. Modify the ledger data according to the pole position information. If the check is correct, modify the ledger data according to the pole position information.

4. The method according to claim 1, characterized in that Before the pulled value data is fed into the pre-trained anchor segment recognition model, it also includes: Obtaining historical pull-out value data of the railway contact network, mileage data of the anchor section joint corresponding to the historical pull-out value data, and types of the anchor section joint corresponding to the historical pull-out value data; Preprocessing the mileage data of the anchor segment joint corresponding to the historical pull-out value data and the type of the anchor segment joint corresponding to the historical pull-out value data; the preprocessing includes one or any combination of sampling, scaling, and windowing; The historical pull-out value data of the railway contact network, the pre-processed mileage data of the anchor section joints corresponding to the historical pull-out value data, and the pre-processed types of the anchor section joints corresponding to the historical pull-out value data are used as sample data to construct a training set and a validation set; The pre-selected machine learning model is trained according to the training set to obtain an anchor segment recognition model; Use the validation set to evaluate the trained anchor segment recognition model and generate evaluation results; According to the evaluation results, the parameters and model structure of the anchor segment identification model are adjusted.

5. The method according to claim 1, characterized in that Before the ledger data is corrected according to the mileage data of the center column, it also includes: Determine whether the ledger data is complete. If incomplete, supplement the ledger data; Filter the ledger data according to the station area name in the ledger data; Arrange the filtered ledger data according to the pillar number in the filtered ledger data; Determine whether the mileage data corresponding to the pillar number is abnormal. If abnormal, correct the abnormal data; According to the preset sequence of pillar types corresponding to the anchor segment joint types, determine whether the pillar types in the arranged ledger data are abnormal. If abnormal, correct the abnormal data.

6. A ledger data correction device, characterized in that: include: An acquisition module is used to obtain the pull-out value data of the railway contact network; The pull-out value data is the pull-out value waveform data of the railway contact network corresponding to the railway mileage data; The pull-out value data includes pull-out value data having pull-out value characteristics of an anchor segment joint; An output module is used to input the pull-out value data into a pre-trained anchor segment recognition model, and output the mileage data of the anchor segment joint and the type of the anchor segment joint; the anchor segment recognition model is obtained by training a machine learning model with the historical pull-out value data of the railway contact network, the mileage data of the anchor segment joint corresponding to the historical pull-out value data, and the type of the anchor segment joint corresponding to the historical pull-out value data; A determination module, used to determine the mileage data of the center column according to the output type of the anchor segment joint and the mileage data of the anchor segment joint; The correction module is used to correct the ledger data according to the mileage data of the center column; the ledger data includes the contact network line information.

7. The device according to claim 6, characterized in that The pull-out value data also includes pole position information.

8. The device according to claim 7, characterized in that The correction module is specifically used for: Correct the center column information in the ledger data according to the mileage data of the center column; Determine the number of pole positions between adjacent anchor segment joints in the pole position information; Compare the number of pole positions between adjacent anchor segment joints in the pole position information with the number of pole positions between adjacent anchor segment joints in the ledger data. If the number of pole positions between adjacent anchor segment joints in the pole position information is inconsistent with the number of pole positions between adjacent anchor segment joints in the ledger data, check the pole position information according to the pull-out value data. If the check is incorrect, modify the pole position information according to the ledger data. Modify the ledger data according to the pole position information. If the check is correct, modify the ledger data according to the pole position information.

9. The device according to claim 6, characterized in that Also includes: Training module for Obtaining historical pull-out value data of the railway contact network, mileage data of the anchor section joint corresponding to the historical pull-out value data, and types of the anchor section joint corresponding to the historical pull-out value data; Preprocessing the mileage data of the anchor segment joint corresponding to the historical pull-out value data and the type of the anchor segment joint corresponding to the historical pull-out value data; the preprocessing includes one or any combination of sampling, scaling, and windowing; The historical pull-out value data of the railway contact network, the pre-processed mileage data of the anchor section joints corresponding to the historical pull-out value data, and the pre-processed types of the anchor section joints corresponding to the historical pull-out value data are used as sample data to construct a training set and a validation set; The pre-selected machine learning model is trained according to the training set to obtain an anchor segment recognition model; Use the validation set to evaluate the trained anchor segment recognition model and generate evaluation results; According to the evaluation results, the parameters and model structure of the anchor segment identification model are adjusted.

10. The device according to claim 6, characterized in that Also includes: Ledger data preprocessing module, used for Determine whether the ledger data is complete. If incomplete, supplement the ledger data; Filter the ledger data according to the station area name in the ledger data; Arrange the filtered ledger data according to the pillar number in the filtered ledger data; Determine whether the mileage data corresponding to the pillar number is abnormal. If abnormal, correct the abnormal data; According to the preset sequence of pillar types corresponding to the anchor segment joint types, determine whether the pillar types in the arranged ledger data are abnormal. If abnormal, correct the abnormal data.

11. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 5 is implemented.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

13. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images