An ai-based optical cable acceptance method and system

Abstract

The application discloses an optical cable acceptance method and system based on AI, overcomes the problem that the difference caused by the influence of human factors in the prior art optical cable acceptance leads to the deviation of communication network quality evaluation, and comprises the following steps: an information input module inputs the basic data related to the optical cable into a background, a system test gives an acceptance node set extracted from an acceptance paragraph, an AI recognition module carries out camera calibration and shooting according to the acceptance node set to obtain AI recognition data such as fiber core test data, and outputs a conclusion whether to pass the acceptance or not, and the like. The application mainly solves the problem of human interference in the acceptance process through acceptance point test and AI recognition, standardizes the acceptance process, synchronously stores photos and data generated in the acceptance process, and realizes that there is a file to check.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of communication, in particular to an optical cable acceptance method and system based on AI. BACKGROUND

[0002] The optical cable is the most basic network of the communication network, and the construction quality and the advantages and disadvantages of the fiber core index directly determine the quality of the communication network. Strictly speaking, the optical cable project acceptance should be jointly conducted by the operator, the construction party and the supervision party, the acceptance requirements are checked item by item through on-site witnessing, an acceptance record table is formed, and archiving is conducted. However, due to the lack of supervision and real-time recording in the process, the project acceptance quality often varies due to the person. In general, the optical cable project acceptance lacks a standardized and process-oriented objective system tool. SUMMARY

[0003] The application is to overcome the problem that the existing optical cable acceptance is influenced by human factors, the communication network quality evaluation is deviated, and an optical cable acceptance method and system based on AI are provided. The interference problem of the person in the acceptance process is solved through acceptance point sampling and AI recognition, the acceptance process is standardized and standardized, the photos and data generated in the acceptance process are synchronously stored, the solid foundation is laid for the quality evaluation system of the communication network, the objective and fair, the efficient and traceable are realized.

[0004] In order to achieve the above-mentioned purpose, the application adopts the following technical scheme:

[0005] An optical cable acceptance method based on AI, comprising the following steps:

[0006] S1, basic data related to the optical cable are input into the background through an information input module; wherein the basic data includes all node names related to the optical cable project, node latitude and longitude, node type, node construction difficulty coefficient, optical cable segment laying mode, segment length information and acceptance requirements;

[0007] S2, the sampling module of the optical cable acceptance system extracts the inspection points according to the node basic information combined with the sampling algorithm, and obtains the acceptance node set; wherein the node basic information includes the node related optical cable small segment length, the node related optical cable small segment laying mode and the construction difficulty coefficient;

[0008] S3, at the acceptance node, the AI recognition module performs camera calibration and on-site shooting to obtain AI recognition data;

[0009] S4, according to the AI recognition data, the tolerable difference and the acceptance requirements set by the system are compared, if the tolerable difference is within the tolerable limit, the acceptance is passed, if the tolerable difference of the project exceeds the tolerable limit, the project and the requirements that need to be rectified are automatically output.

[0010] The scheme first introduces basic data including node information, extracts inspection points along the way by setting different weight coefficients of nodes, mainly through the S2 test algorithm, comprehensively considers the laying mode, cable length and construction difficulty coefficient, gives different test probabilities to nodes according to the weight coefficients, and finally obtains an acceptance node set including a plurality of to-be-tested points. After determining the inspection points, the on-site acceptance photos are shot, AI recognition is carried out, the difference between the data recognized by AI and the pre-set acceptance requirements in the system is compared, and the conclusion of whether the acceptance passes or not is obtained. If it does not pass, the rectification requirements are output.

[0011] The application utilizes AI to make the project acceptance link more standardized, synchronously stores the photos and data generated in the acceptance process, and realizes that the acceptance result can be traced and viewed. The system test algorithm is mainly used to select high-risk and error-prone nodes, avoid human subjective factors, and is suitable for various optical cable acceptance projects and is objective and fair. AI recognition is used to white-box the on-site situation and data in the acceptance process, exclude human interference factors, and output rectification schemes.

[0012] Further, when the basic data related to the optical cable is input into the background through the information input module, the node types in the basic data include joint points and end points.

[0013] Further, as a main part of the application, the test module of the optical cable acceptance system extracts test points according to node basic information combined with a test algorithm to obtain an acceptance node set.

[0014] The test algorithm specifically includes the following contents:

[0015] S21, the optical cable project has N optical cable sections, and the section between the end points is called an optical cable section;

[0016] The optical cable section is Li, i=1, 2, ……N, and there are N optical cable sections,

[0017] The first and last nodes of the optical cable section are the end points, denoted as Ai and Bi,

[0018] Wherein, the joint points along each optical cable section are denoted as Pij, i=1, 2, ……N, and there are N optical cable sections, j=1, 2, ……Mi, and Mi represents the number of joint points of each optical cable section,

[0019] Wherein, each optical cable section is cut into small sections by the joint points along the way, which is called an optical cable small section, and the optical cable small section is denoted as Qij, i=1, 2, ……N, and there are N optical cable sections, j=1, 2, ……Mi+1, and Mi represents the number of joint points of each optical cable section.

[0020] S22, in each optical cable section, according to the laying mode of the joint point related optical cable small section, the laying coefficient of the joint point is obtained, and the joint point of each optical cable section is provided with a corresponding construction difficulty coefficient; the construction difficulty coefficient has a value range of 0-1;

[0021] The laying mode includes direct-buried laying mode, overhead laying mode and pipeline laying mode.

[0022] The value of the laying coefficient is divided into two cases according to the number of laying modes adopted.

[0023] If the joint point involves only one laying mode, the laying coefficient is set to 0.1, and if the joint point involves more than one laying mode, the laying coefficient is set to 0.3. The setting rule of the laying coefficient can be adjusted, which can be modified in the system parameter setting.

[0024] The distance coefficient is distance coefficient = sum of distances of adjacent optical cable small sections of the node / total distance of the optical cable section.

[0025] Then, the sampling coefficient of all joint points is determined according to the construction difficulty coefficient, the distance coefficient and the laying coefficient of the joint point,

[0026] The sampling coefficient of the joint point = construction difficulty coefficient + distance coefficient + laying coefficient.

[0027] S23, then the sampling number in each optical cable section is calculated, and 2 end points are must-check points,

[0028] The sampling number of each optical cable section = 2 + roundown (distance of direct-buried optical cable section / 10, 0) + roundown (distance of overhead optical cable section / 10, 0) + roundown (distance of pipeline optical cable section / 10, 0).

[0029] S24, according to the results of the above steps, the joint point set of each optical cable section under different laying modes is obtained,

[0030] Because each joint point involves one or more laying modes, one joint point can correspond to one or more joint point sets, for example, the joint point involving more than one laying mode should belong to the joint point set under different laying modes at the same time.

[0031] At the same time, the joint points in the joint point set are arranged in descending order according to the sampling coefficient.

[0032] S25, according to the rules of the number of sampling in S23, respectively, the joint point set to extract the joint point that meets the sampling number requirement, form the acceptance node set of each cable segment, and eliminate the repeated items in the set, get the acceptance node set. Because a joint point can correspond to multiple joint point sets, so the joint point needs to be extracted to eliminate the repeated items of the joint point corresponding to multiple joint point sets, finally get an acceptance point set.

[0033] Further, the camera calibration and field shooting at the acceptance node AI recognition module obtains AI recognition data, specifically:

[0034] After arriving at the acceptance node, log in to the handheld terminal, enter the system AI recognition module, and perform camera calibration and field shooting operation according to the system prompt.

[0035] Further, the camera calibration operation includes:

[0036] S31, before shooting, the basic parameters of the camera model not in the database are extracted, that is, a set of test images are used to calculate the focal length and optical center of the current model camera.

[0037] Further, the field shooting is after the camera calibration operation, specifically:

[0038] S32, the key points of field shooting are photographed, and the field shooting photos are imported into the handheld terminal, and the handheld terminal is identified for clarity and angle, the AI recognition module has preset clarity and angle range, if the identified clarity and angle fall outside the preset clarity and angle range, the handheld terminal is prompted to re-shoot, if the identified clarity and angle fall within the preset clarity and angle range, it is determined, the handheld terminal reads out the photo information after AI recognition, and stores the photo in the background.

[0039] Further, the field shooting key points include:

[0040] Use the fiber core tester to test the fiber core loss index of each core, and get the instrument test result;

[0041] For the direct-buried laid optical cable, excavate and measure by tape measure;

[0042] For overhead laying optical cable, shoot the joint point pole line, and the distance from the adjacent pole;

[0043] For the pipeline laid optical cable, shoot the hand well optical cable disc situation, and the adjacent hand well distance measured by roller.

[0044] Further, after the field shooting, a three-dimensional model of the engineering measurement object is obtained, specifically as follows:

[0045] S33, the on-site shooting photo is transmitted back to the SaaS server;

[0046] The distortion of the lens is corrected, a depth map is constructed by matching similar information, pixels are reprojected into three-dimensional space using depth mapping, a point cloud is constructed, a network is constructed to obtain a three-dimensional model, and engineering acceptance measurement is realized according to the model.

[0047] Further, after obtaining the three-dimensional model of the engineering measurement object,

[0048] S34, the OCR and deep learning technology are used to identify and read the signboard, label and instrument reading.

[0049] An AI-based optical cable acceptance system applied to an AI-based optical cable acceptance method, the system comprising a hardware part and a software part for controlling the hardware part;

[0050] The hardware part comprises a handheld terminal, a computer operating end and a server connected with each other;

[0051] The software part comprises an information interconnected input module, a sampling module, an AI identification module and a verification comparison module.

[0052] Therefore, the present application has the following beneficial effects:

[0053] Firstly, the present application prevents human interference, avoids subjective guidance risks, and improves efficiency by sampling the acceptance point, and the sampling algorithm can cover any situation of the optical cable project, and the optical cable under various laying modes can be reasonably calculated and obtain appropriate acceptance nodes, which has universality, objectivity, fairness, efficiency and convenience.

[0054] Secondly, the present application uses AI identification, which can quickly detect various key points, construct a depth map by matching information, map the depth map to a three-dimensional space to construct a three-dimensional model, realize the visualization of engineering acceptance measurement and the traceability of acceptance results, solve the human interference factors in the acceptance process, and output the rectification scheme. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 is the flowchart of the present embodiment. DETAILED DESCRIPTION

[0056] The present application will be further described below in combination with the drawings and specific embodiments.

[0057] Cable construction quality and fiber core index are of great significance to optical fiber communication, and directly determine the quality of communication network. The present application mainly solves the problem of human interference in the acceptance process through acceptance point sampling and AI recognition, and at the same time, the acceptance process is standardized, the photos and data generated in the acceptance process are stored synchronously, and the file can be checked. First, the basic data is imported, and the inspection points along the way are extracted through the different weight coefficients of the nodes. The sampling algorithm comprehensively considers the laying mode, cable segment length and construction difficulty coefficient, and gives different sampling probabilities to the nodes according to the weight coefficients. After determining the inspection points, the on-site acceptance photos are imported, AI recognition is carried out, the difference between the data recognized by AI and the acceptance requirements is compared, and the conclusion of whether the acceptance is passed or not is obtained. If not, output the rectification requirements.

[0058] The present application provides an AI-based optical cable acceptance system, mainly including a handheld terminal, a computer operating terminal and a server, and the handheld terminal is a mobile phone in the present application. The system software includes an information input module, a sampling module, an AI recognition module and a verification comparison module, which mainly realizes the functions of sampling, AI recognition and output conclusion.

[0059] The present application provides an AI-based optical cable acceptance method, as shown in Figure 1 The method comprises the following steps:

[0060] Step 1, input the basic data related to optical cable into the background through the information input module;

[0061] Among them, the basic data includes all node names, node latitude and longitude, node type, node construction difficulty coefficient, cable segment laying mode, segment length information and acceptance requirements related to optical cable project; the node type includes the joint point and the end point.

[0062] Step 2, the sampling module of the optical cable acceptance system extracts the inspection points according to the node basic information combined with the sampling algorithm, and obtains the acceptance node set; wherein the node basic information includes the node related optical cable segment length, the node related optical cable segment laying mode and the construction difficulty coefficient;

[0063] 2-1: there are N optical cable segments in the optical cable project, and the segment between the end points is called optical cable segment;

[0064] The optical cable segment is Li, i=1, 2, ……N, and there are N optical cable segments,

[0065] The first and last nodes of the optical cable segment are the end points, denoted as Ai and Bi,

[0066] Among them, each joint point along the optical cable segment is denoted as Pij, i=1, 2, ……N, and there are N optical cable segments, j=1, 2, ……Mi, Mi represents the number of joint points of each optical cable segment,

[0067] Wherein, each optical cable section is cut into small sections by the joint points along the way, called optical cable small sections, and the optical cable small sections are denoted as Qij, i = 1, 2, …, N, N optical cable sections, and j = 1, 2, …, Mi+1, Mi represents the number of joint points of each optical cable section;

[0068] 2-2: In each optical cable section, the laying coefficient of the joint point is obtained according to the laying mode of the joint point related optical cable small section, and each joint point of the optical cable section is provided with a corresponding construction difficulty coefficient; the construction difficulty coefficient has a value range of 0-1;

[0069] Wherein, the laying mode includes direct-buried laying mode, overhead laying mode and pipeline laying mode;

[0070] The value of the laying coefficient is divided into two cases according to the number of laying modes adopted;

[0071] If the joint point involves only one laying mode, the laying coefficient is set to 0.1, and if the joint point involves more than one laying mode, the laying coefficient is set to 0.3. The setting rule of the laying coefficient can be adjusted, which can be modified in the system parameter setting;

[0072] Distance coefficient, distance coefficient = sum of distances of adjacent optical cable small sections of nodes / total distance of optical cable section;

[0073] Further, the sampling coefficient of all joint points is determined according to the construction difficulty coefficient, the distance coefficient and the laying coefficient of the joint point,

[0074] The sampling coefficient of the joint point = the construction difficulty coefficient + the distance coefficient + the laying coefficient;

[0075] 2-3: Calculate the sampling number in each optical cable section, and the two end points are mandatory points,

[0076] The sampling number of each optical cable section = 2 + roundown (direct-buried optical cable section distance / 10, 0) + roundown (overhead optical cable section distance / 10, 0) + roundown (pipeline optical cable section distance / 10, 0);

[0077] 2-4: Obtain the joint point set of each optical cable section under different laying modes,

[0078] Because each joint point involves one or more laying modes, a joint point can correspond to one or more joint point sets, for example, a joint point involving more than one laying mode should belong to the joint point set under different laying modes at the same time;

[0079] At the same time, the joint points in the joint point set are arranged in descending order according to the sampling coefficient;

[0080] 2-5: According to the sampling quantity rule in S23, the joint point set that meets the sampling quantity requirement is extracted respectively to form the acceptance node set of each cable segment, and the repeated items in the set are removed to obtain the acceptance node set Zi, wherein i = 1, 2, …, N, and N is the total number of cable segments.

[0081] Step 3, at the acceptance node, the AI recognition module performs camera calibration and on-site shooting to obtain AI recognition data.

[0082] 3-1: After arriving at the acceptance node Pij, log in to the handheld terminal, enter the system AI recognition module, and perform camera calibration and on-site shooting operation according to the system prompt;

[0083] 3-2: Before shooting, the basic parameters of the camera model not in the database are extracted, that is, the focal length and optical center of the current camera model are calculated using a set of test images;

[0084] 3-3: Photograph the key points of on-site shooting, import the on-site shooting photos into the handheld terminal, and identify the clarity and angle by the handheld terminal, the AI recognition module has preset clarity and angle range, if the identified clarity and angle fall outside the preset clarity and angle range, the handheld terminal is prompted to re-shoot, if the identified clarity and angle fall within the preset clarity and angle range, it is determined, the handheld terminal reads out the photo information after AI recognition, and stores the photo in the background;

[0085] The on-site shooting key points include but are not limited to the following contents:

[0086] Test the fiber core loss index of each core with the fiber core tester, and take a photo of the instrument test result;

[0087] For direct-buried laid optical cable, excavate and measure with a tape measure, and take a photo;

[0088] For overhead laid optical cable, take photos of the joint point pole line and the distance from the adjacent pole;

[0089] For pipe-laid optical cable, take photos of the hand well optical cable disc situation and the roller measurement of the distance between adjacent hand wells;

[0090] 3-4: Obtain a three-dimensional model of the engineering measurement object; transmit the on-site shooting photos back to the SaaS server. Correct the lens distortion, construct a depth map by matching similar information. Use depth mapping to project pixels into three-dimensional space and construct a point cloud, construct a network to obtain a three-dimensional model, and realize engineering acceptance measurement according to the model;

[0091] 3-5: Use OCR and deep learning technology to identify and read the signboard, label and instrument reading.

[0092] Step 4, according to the AI identification data in step 3, the system is set to compare the tolerable difference with the acceptance requirements, if the tolerable difference is within the tolerable limit, the acceptance is passed, if the tolerable difference of the project exceeds the tolerable limit, the project and the requirements that need to be rectified are automatically output.

[0093] The above embodiments are only used for further illustration of the application, and cannot be understood as limitation of the protection scope of the application. The skilled in the art can make some non-essential improvements and adjustments to the application according to the content of the application, which falls within the protection scope of the application.

Claims

1. An AI-based optical cable acceptance method, characterized by, The method comprises the following steps: The basic data related to the optical cable is input into the background through the information input module of the optical cable acceptance system; wherein, the basic data includes all node names, node latitude and longitude, node type, node construction difficulty coefficient, optical cable segment laying mode, segment length information and acceptance requirements related to the optical cable project; The test point extraction module of the optical cable acceptance system extracts test points according to the node basic information and the test algorithm to obtain a test node set; wherein, the node basic information includes node related optical cable segment length, node related optical cable segment laying mode and construction difficulty coefficient; At any test node, the AI recognition module of the optical cable acceptance system performs camera calibration and on-site shooting to obtain AI recognition data; According to the AI recognition data, the tolerable difference and the acceptance requirements set by the system are compared, if the tolerable difference is within the tolerable limit, the acceptance is passed, if the tolerable difference of the project exceeds the tolerable limit, the project and requirements that need to be rectified are automatically output; Wherein, the optical cable between the end points of the optical cable project is called an optical cable segment, assuming that there are N segments; the optical cable acceptance system gives a test algorithm according to the project basic information, and the test module extracts test points according to the test algorithm to obtain a test node set, the test algorithm includes: S21, there are N optical cable segments between the end points of the optical cable project; the first and last nodes of the optical cable segment are Ai and Bi respectively, each optical cable segment has a joint point along the way; the optical cable segment is cut into small segments by the joint points along the way, which is called optical cable segment; S22, in each optical cable segment, according to the laying mode of the joint point related optical cable segment, the laying coefficient of the joint point is obtained, the joint point is provided with construction difficulty coefficient, distance coefficient and laying coefficient; distance coefficient = sum of node adjacent optical cable segment distance / total optical cable segment distance; the value of laying coefficient is divided into two cases according to whether the number of laying modes used is more than one; if the joint point involves only one laying mode, the laying coefficient is set to 0.1, if the joint point involves more than one laying mode, the laying coefficient is set to 0.3; the setting rule of laying coefficient can be adjusted, which can be modified in the system parameter setting; the test coefficient of the joint point = construction difficulty coefficient + distance coefficient + laying coefficient; S23, then calculate the number of tests in each optical cable segment, 2 end points are mandatory points, the number of tests of each optical cable segment = 2 + roundown(buried optical cable segment distance / 10, 0) + roundown(overhead optical cable segment distance / 10, 0) + roundown(pipeline optical cable segment distance / 10, 0); S24, obtain the joint point set of each optical cable segment under different laying modes, the joint points related to more than one laying mode belong to the joint point set under different laying modes at the same time; the joint points in the joint point set are arranged in descending order according to the test coefficient; S25, according to the test number rule in S23, the joint points in the joint point set that meet the test number requirement are extracted respectively to form the test node set of each optical cable segment, and the repeated items in the set are removed to obtain the test node set. At the acceptance node, the AI recognition module performs camera calibration and on-site shooting to obtain AI recognition data, including: before shooting, basic parameter extraction is performed on camera models not in the database, and the focal length and optical center of the current model camera are calculated using a set of test images; key points of on-site shooting are photographed, and on-site shooting photos are imported into a handheld terminal, and the handheld terminal respectively identifies the clarity and angle, the AI recognition module has preset clarity and angle ranges, if the identified clarity and angle falls outside the preset clarity and angle range, the handheld terminal is prompted to re-shoot, if the identified clarity and angle falls within the preset clarity and angle range, it is determined, the handheld terminal reads out the photo information after AI recognition, and the photo is stored in the background at the same time; wherein the key points of on-site shooting include: testing the fiber core loss index of each core with a fiber core tester, and photographing the instrument test results; for the direct-buried laid optical cable, the optical cable is excavated and measured by pulling the ruler, and the photograph is recorded; for the overhead laid optical cable, the joint point pole road and the distance from the adjacent pole are photographed; for the pipe laid optical cable, the hand well optical cable disc remains and the roller measured adjacent hand well distance photos are photographed; a three-dimensional model is obtained for engineering measurement objects; the on-site photographed photos are transmitted back to the SaaS server; the distortion of the lens is corrected, a depth map is constructed by matching similar information; the pixels are re-projected into a three-dimensional space using depth mapping, and a point cloud is constructed, a network is constructed to obtain a three-dimensional model, and engineering acceptance measurement is realized according to the model; the signboard, label and instrument reading are identified and read by using OCR and deep learning technology.

2. The AI-based optical cable acceptance method according to claim 1, characterized in that, When the information input module of the optical cable acceptance system enters the basic data related to the optical cable into the background, the node types in the basic data include joint points and end points.

3. The AI-based optical cable acceptance method of claim 1, wherein the AI-based optical cable acceptance method is characterized by At any acceptance node, the AI recognition module of the optical cable acceptance system performs camera calibration and on-site shooting to obtain AI recognition data, specifically as follows: After arriving at any acceptance node, log in to the handheld terminal, enter the AI recognition module of the optical cable acceptance system, and perform camera calibration and on-site shooting operations according to system prompts.

4. The AI-based optical cable acceptance method of claim 3, wherein, The camera calibration operation includes: Before shooting, basic parameter extraction is performed on camera models not in the database, that is, the focal length and optical center of the current model camera are calculated using a set of test images.

5. The AI-based optical cable acceptance method of claim 3, wherein, The on-site shooting is performed after the camera calibration operation, specifically as follows: The key points of on-site shooting are photographed, and the on-site shooting photos are imported into a handheld terminal, and the handheld terminal respectively identifies the clarity and angle, the AI recognition module of the optical cable acceptance system has preset clarity and angle ranges, if the identified clarity and angle falls outside the preset clarity and angle range, the handheld terminal is prompted to re-shoot, if the identified clarity and angle falls within the preset clarity and angle range, it is determined, the handheld terminal reads out the photo information after AI recognition, and the photo is stored in the background at the same time.

6. An AI-based optical cable acceptance system applied to the AI-based optical cable acceptance method of claim 1, characterized in that, The system includes a hardware part and a software part for controlling the hardware part; The hardware part includes a handheld terminal, a computer operating end and a server connected to each other; The software part includes information entry module, test module, AI identification module of optical cable acceptance system and check comparison module.

Images